[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-openai-boltz-protein-design":3,"mdc-pfcunp-key":33,"related-repo-openai-boltz-protein-design":1922,"related-org-openai-boltz-protein-design":2045},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":22,"repoUrl":23,"updatedAt":24,"license":25,"forks":26,"topics":27,"repo":28,"sourceUrl":31,"mdContent":32},"boltz-protein-design","design novel protein binders","Design new protein binders with Boltz. Use when generating protein, peptide, antibody, nanobody, or custom binder candidates for a target. Not for screening existing proteins or small molecules.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"openai","OpenAI","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fopenai.png",[12,16,19],{"name":13,"slug":14,"type":15},"Research","research","tag",{"name":17,"slug":18,"type":15},"Life Sciences","life-sciences",{"name":20,"slug":21,"type":15},"Bioinformatics","bioinformatics",3992,"https:\u002F\u002Fgithub.com\u002Fopenai\u002Fplugins","2026-06-30T19:00:57.102",null,465,[],{"repoUrl":23,"stars":22,"forks":26,"topics":29,"description":30},[],"OpenAI Plugins","https:\u002F\u002Fgithub.com\u002Fopenai\u002Fplugins\u002Ftree\u002FHEAD\u002Fplugins\u002Fboltz-api-cli\u002Fskills\u002Fboltz-protein-design","---\nname: boltz-protein-design\ndescription: Design new protein binders with Boltz. Use when generating protein, peptide, antibody, nanobody, or custom binder candidates for a target. Not for screening existing proteins or small molecules.\n---\n\n## Workflow\n\nIf `boltz-api` is missing from `PATH`, use `boltz-cli-setup` for install\u002Fupdate guidance before retrying.\nIf a command reports missing or expired authentication, use `boltz-cli-setup` to start `boltz-api auth login --device-code` before retrying; do not ask permission first.\nIf the agent host sandbox blocks `boltz-api` install\u002Fauth\u002FAPI calls, use `boltz-cli-setup` to request the host sandbox bypass\u002Fescalation needed for user-wide CLI install, browser login, credential storage, temp files, or API access before retrying.\n\nUse this skill when the user wants de novo protein \u002F peptide \u002F antibody \u002F nanobody binders.\n\n1. **Decide on target exploration first (new targets).** For a new target where the user hasn't already fixed the binding site and crop, your first action — before authoring a payload, normalizing the target, or running `estimate-cost` — is to raise the choice between a target-exploration pass and designing directly, with a **recommendation** for this target:\n   - Unknown site, or a multi-domain \u002F large target → recommend exploration (it scouts different input configurations for generation, ≈50 designs each, and finds the best before a full run).\n   - A well-characterized site → it's fine to recommend going (mostly) direct, perhaps with a quick check of whether conditioning on the epitope beats letting the model find its own spot. State this plainly, as part of a conversation with the user about their target and goals, and let them choose.\n\n   Phrase it as a question that works *with* the user (they may know their target's biology), e.g.:\n\n   > \"This is a fresh target — I'd suggest a quick exploration pass that scouts a few framings and picks the best before a full run. Or, if you already know the site and crop, we can design directly. Which would you like?\"\n\n   **Do not mention a campaign size or tier here** — not even folded into this opening approach question. The full-run size is settled later, after the scouting runs pick a winner (its yield informs the tier), so don't ask it up front when exploration is on the table. If the user opts into exploration — or has already said they want to explore \u002F let the design find its own epitope — read [references\u002Ftarget-exploration.md](references\u002Ftarget-exploration.md), follow it, then resume at step 8 with the chosen framing and recommended `num_proteins`. If they want to design directly, continue below.\n2. Normalize the target (same shape as protein-screen): `structure_template` if a CIF\u002FPDB is available, else `no_template`.\n3. Pick the `binder_specification` variant. Supported variants include:\n   - `boltz_curated` — recommended default for antibody and nanobody design. Boltz selects from maintained scaffold\u002Ftemplate lists (`binder: boltz_antibody` or `boltz_nanobody`).\n   - `structure_template` — redesign motifs in an existing binder scaffold (CIF + `design_motifs` with `replacement` \u002F `insertion` segments).\n   - `no_template` — generate from the sequence DSL (fixed residues + designed segments like `5..10` or `8`).\n4. For antibody or nanobody requests, ask before authoring the payload: \"I recommend Boltz's curated antibody\u002Fnanobody scaffolds for this. Do you want the curated default, or do you have custom scaffold structures\u002FCDR motifs to use?\" If the user picks curated, use `type: boltz_curated`; if they want custom scaffold control, use `type: structure_template`.\n5. Pick `modality`: `peptide`, `antibody`, `nanobody`, or `custom_protein` for `structure_template` and `no_template` (use `custom_protein` for a \"miniprotein\" or generic \"protein binder\"). **If the user already named the modality, take it as given — don't ask again.** Do not include `modality` on `boltz_curated`; use `binder` instead.\n6. Pick `num_proteins` — see [Run sizing](#run-sizing). Valid range is **10 to 1,000,000** (server rejects outside it); **10** is the hard floor but it is a test size, not a campaign. When the user has not given a count, propose a campaign tier (default **50,000**), not the floor.\n7. Supported optional features include rules such as excluded amino acids, excluded sequence motifs with `X` wildcards, and max hydrophobic fraction. Add `rules` only on request; read [references\u002Fapi.md](references\u002Fapi.md) for exact shapes and examples.\n8. Author the payload YAML or JSON, then run `estimate-cost` and apply the **spending gate** (Always Do This) before `start`. (Cost model — tiered by total complex length, `estimate-cost` is the only source: see [`## Cost`](references\u002Fapi.md) in api.md.)\n9. `start` to submit. Capture the ID.\n10. Launch `download-results` with the agent runtime's background\u002Fnon-blocking command facility. In Claude Code, use Bash with `run_in_background: true`. In Codex, run `download-results` as a foreground shell command with `yield_time_ms: 1000`; if Codex returns a `session_id`, keep it for optional same-thread polling, but treat `download-status` plus the run directory as the durable source of truth. In Codex app\u002Fdesktop runtimes that expose same-thread heartbeat automations, create a heartbeat that checks `download-status` periodically and posts a concise completion or failure update when the download reaches a terminal state. After launching the downloader, always report the job ID, run name, and output directory. Include the next check cadence if the heartbeat was created; otherwise include the `download-status` command.\n11. Rank from `\u003Coutput-root>\u002F\u003Crun-name>\u002Fresults\u002Findex.jsonl` by `binding_confidence` descending. Use `iptm` and `min_interaction_pae` as tiebreakers. `optimization_score` is not emitted for this endpoint. Read [references\u002Fresults.md](references\u002Fresults.md) for output layout and metric details.\n\n## Run sizing\n\nDe novo design is a generate-and-filter campaign: you make many binders and keep the rare good ones, so a real run is **large**. Do not anchor on the `num_proteins` floor of 10 — that is only useful for a quick setup test. When the user names a count, honor it (≥10). When they do not, explain the tiers and propose one:\n\n| Tier | `num_proteins` | When |\n|---|---|---|\n| Small screen | 20,000 | Quick look \u002F tight budget |\n| **Medium (recommended)** | **50,000** | Default for a real campaign |\n| Large | 100,000 | Hard target \u002F maximal coverage |\n\nPresent the tiers as design **counts**, not dollars: don't put a price next to a tier unless `estimate-cost` returned it — run `estimate-cost` on the tier the user leans toward and show that figure, and never extrapolate one estimate across the others. Then apply the **spending gate** (Always Do This) before submitting. If the setup is unproven, suggest a small test run (tens of designs) or the full [target-exploration](references\u002Ftarget-exploration.md) pass first.\n\n## Command Pattern\n\n```bash\n# Replace placeholders with concrete absolute paths before running.\n# Use a short descriptive run name, for example: protein-design-\u003Cmodality>-\u003Ctarget>-v1\n\nboltz-api protein:design estimate-cost \\\n  --input @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml\n\nboltz-api protein:design start \\\n       --idempotency-key \"\u003Crun-name>\" \\\n       --input @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml \\\n       --raw-output --transform id\n\n# Copy the printed job ID into this command, then launch it in the agent\n# runtime's background\u002Fnon-blocking mode.\n# Claude Code: Bash with run_in_background=true.\n# Codex: foreground shell command with yield_time_ms=1000; keep the returned session_id if one is provided.\n# Do not append \"&\" or use nohup in Codex.\nboltz-api download-results \\\n  --id \"\u003Cjob-id-from-start>\" --name \"\u003Crun-name>\" \\\n  --root-dir \"\u002Fabsolute\u002Fpath\u002Fboltz-experiments\" \\\n  --poll-interval-seconds 60\n```\n\nPayload keys are `num_proteins`, `target`, `binder_specification` — API body field names.\n\n## Always Do This\n\n- **For a new target, your first move is the step-1 conversation — never jump straight to a payload.** Establish what the binder is *for* and whether the user has already fixed the binding site\u002Fcrop, then recommend a target-exploration pass vs. designing directly and let them choose (step 1). Only normalize the target and author a payload after that. If they've already fixed the site\u002Fcrop or explicitly want to design directly, proceed.\n- Enforce `10 \u003C= num_proteins \u003C= 1,000,000` before calling `estimate-cost` (server rejects outside that range), but 10 is the floor, not a campaign — see [Run sizing](#run-sizing) and propose a tier (default 50,000) when the user gives no count.\n- **Spending gate — explicit go-ahead before every `start`.** `start` spends real money. A plan you already described, an earlier phase's approval, or a cost that looks \"trivial\" are **not** authorization — even a cheap run needs a fresh yes. Run `estimate-cost`, show the `estimated_cost_usd` it returns (summed for a batch), and wait for the user to say go. This holds **even when tool calls are pre-approved** (accept-edits \u002F auto-accept \u002F bypass modes) — there you are the only cost gate. Never quote or assume a dollar figure you didn't get from `estimate-cost` (cost model: [`## Cost`](references\u002Fapi.md) in api.md).\n- For antibody or nanobody design, recommend `binder_specification.type: boltz_curated` and ask the user to confirm they do not want custom scaffold\u002FCDR control before building the payload. Use `binder: boltz_antibody` for antibody\u002FFab requests and `binder: boltz_nanobody` for nanobody\u002FVHH requests.\n- Residue indices are 0-based everywhere (`design_motifs.start_index`\u002F`end_index`, `after_residue_index`, `epitope_residues`, `flexible_residues`, bonds, constraints).\n- For CIF\u002FPDB bytes, use `@data:\u002F\u002F\u002Fabs\u002Fpath\u002Ffile.cif` inside `structure.data`. Don't use bare `@path`.\n- Sequence DSL for `designed_protein.value`: uppercase letters = fixed residues; integer `N` = exactly `N` designed residues; `MIN..MAX` = variable-length designed segment. Examples: `\"20\"`, `\"5..10\"`, `\"ACDE8GHI\"`, `\"MKTAYI5..10VKSHFSRQ\"`.\n- Keep payload field names exactly as the API body names shown in `references\u002Fapi.md`.\n- Use absolute paths for the output root, payload files, and embedded target files. Do not `cd` into the run directory for follow-up commands; pass the same `--root-dir` and use absolute paths so later relative paths do not drift.\n- Prefer one merged top-level payload via `--input @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml` or `@json:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.json` for `estimate-cost` and `start`. Keep `--idempotency-key` and `--workspace-id` top-level; if they also appear inside `--input`, the top-level flags win.\n- Direct object flags still work as overrides, such as `--target @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Ftarget.yaml` or `--binder-specification @json:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fbinder.json`. Piped YAML \u002F JSON on stdin also works, but it must use API body field names. Use the same slug for both `--idempotency-key` and `--name`.\n- In permission-gated agents such as Claude Code, keep each Boltz call as a top-level command that starts with `boltz-api`. Prefer concrete arguments over `sh -c`, inline environment assignments, aliases, wrapper scripts, loops, or pipelines around the `boltz-api` invocation unless the user already allowed that exact command form. Use `--raw-output --transform id`, read the printed ID, then paste that literal ID into the next `download-results` command.\n- Prefer the agent runtime's background\u002Fnon-blocking command mode for `download-results`. In Codex specifically, keep `download-results` in the foreground and set the shell tool yield to 1000 ms; Codex will return a `session_id` if the command is still running. Do not append `&` or use `nohup` in Codex because the tool runner may clean up shell-backgrounded descendants before `.boltz-run.json` is fully written.\n- After the background\u002Fsession starts, do not manually wait on it or run ad hoc polling loops. Wall-clock time scales roughly with `num_proteins`: under 100 often finishes in a few minutes, 100-1,000 may take several minutes to tens of minutes, and larger runs can take longer or hours depending on inputs and system load. Don't quote a fixed duration. `--poll-interval-seconds 60` is a sensible downloader default. `download-results` emits JSONL progress on stderr by default; add `--progress-format text --verbose` only when you explicitly want human-readable logs.\n- In Codex app\u002Fdesktop runtimes with same-thread heartbeat automation support, schedule a heartbeat after launching `download-results`. The heartbeat should run `boltz-api --format json download-status --name \"\u003Crun-name>\" --root-dir \"\u002Fabsolute\u002Fpath\u002Fboltz-experiments\"` and stop once terminal. Choose cadence by `num_proteins`: under 100 -> every 1-2 minutes; 100-1,000 -> every 5 minutes; over 1,000 -> every 15 minutes. Post only material status changes or terminal completion\u002Ffailure. Poll the saved `session_id` with an empty `write_stdin` only for interactive, user-requested progress checks. Never run a manual poll loop in the current turn.\n- If the current host has no heartbeat automation support, do not claim an automatic next check. Report the job ID, run name, output directory, and the command needed to check `download-status`.\n- If detached download needs to be restarted, re-run `boltz-api download-results` with the same `--name \"\u003Crun-name>\"` and the same `--root-dir`.\n- Only add `rules` on explicit user request.\n\n## Escape Hatch\n\n- Payload reference: \u003Chttps:\u002F\u002Fapi.boltz.bio\u002Fdocs\u002Fapi\u002Fresources\u002Fprotein\u002Fsubresources\u002Fdesign\u002Fmethods\u002Fstart\u002F>\n- CLI flag names: `boltz-api protein:design start --help`\n\nRead [references\u002Fapi.md](references\u002Fapi.md) for all `binder_specification` variants, motif shapes, sequence DSL, rules, modalities, and `target` variants. Read [references\u002Fresults.md](references\u002Fresults.md) after download when ranking designed binders or explaining outputs.\n\n## Outputs\n\nRank from `results\u002Findex.jsonl` after `download-results`; use [references\u002Fresults.md](references\u002Fresults.md) for local file layout, metric meanings, and the designed-binder entity type gotcha.\n",{"data":34,"body":35},{"name":4,"description":6},{"type":36,"children":37},"root",[38,47,107,112,682,687,706,801,840,846,1177,1202,1208,1823,1829,1855,1886,1892,1916],{"type":39,"tag":40,"props":41,"children":43},"element","h2",{"id":42},"workflow",[44],{"type":45,"value":46},"text","Workflow",{"type":39,"tag":48,"props":49,"children":50},"p",{},[51,53,60,62,68,70,76,78,83,85,91,93,98,100,105],{"type":45,"value":52},"If ",{"type":39,"tag":54,"props":55,"children":57},"code",{"className":56},[],[58],{"type":45,"value":59},"boltz-api",{"type":45,"value":61}," is missing from ",{"type":39,"tag":54,"props":63,"children":65},{"className":64},[],[66],{"type":45,"value":67},"PATH",{"type":45,"value":69},", use ",{"type":39,"tag":54,"props":71,"children":73},{"className":72},[],[74],{"type":45,"value":75},"boltz-cli-setup",{"type":45,"value":77}," for install\u002Fupdate guidance before retrying.\nIf a command reports missing or expired authentication, use ",{"type":39,"tag":54,"props":79,"children":81},{"className":80},[],[82],{"type":45,"value":75},{"type":45,"value":84}," to start ",{"type":39,"tag":54,"props":86,"children":88},{"className":87},[],[89],{"type":45,"value":90},"boltz-api auth login --device-code",{"type":45,"value":92}," before retrying; do not ask permission first.\nIf the agent host sandbox blocks ",{"type":39,"tag":54,"props":94,"children":96},{"className":95},[],[97],{"type":45,"value":59},{"type":45,"value":99}," install\u002Fauth\u002FAPI calls, use ",{"type":39,"tag":54,"props":101,"children":103},{"className":102},[],[104],{"type":45,"value":75},{"type":45,"value":106}," to request the host sandbox bypass\u002Fescalation needed for user-wide CLI install, browser login, credential storage, temp files, or API access before retrying.",{"type":39,"tag":48,"props":108,"children":109},{},[110],{"type":45,"value":111},"Use this skill when the user wants de novo protein \u002F peptide \u002F antibody \u002F nanobody binders.",{"type":39,"tag":113,"props":114,"children":115},"ol",{},[116,206,227,328,348,442,482,510,555,565,631],{"type":39,"tag":117,"props":118,"children":119},"li",{},[120,126,128,134,136,141,143,157,161,163,169,171,180,183,188,190,196,198,204],{"type":39,"tag":121,"props":122,"children":123},"strong",{},[124],{"type":45,"value":125},"Decide on target exploration first (new targets).",{"type":45,"value":127}," For a new target where the user hasn't already fixed the binding site and crop, your first action — before authoring a payload, normalizing the target, or running ",{"type":39,"tag":54,"props":129,"children":131},{"className":130},[],[132],{"type":45,"value":133},"estimate-cost",{"type":45,"value":135}," — is to raise the choice between a target-exploration pass and designing directly, with a ",{"type":39,"tag":121,"props":137,"children":138},{},[139],{"type":45,"value":140},"recommendation",{"type":45,"value":142}," for this target:",{"type":39,"tag":144,"props":145,"children":146},"ul",{},[147,152],{"type":39,"tag":117,"props":148,"children":149},{},[150],{"type":45,"value":151},"Unknown site, or a multi-domain \u002F large target → recommend exploration (it scouts different input configurations for generation, ≈50 designs each, and finds the best before a full run).",{"type":39,"tag":117,"props":153,"children":154},{},[155],{"type":45,"value":156},"A well-characterized site → it's fine to recommend going (mostly) direct, perhaps with a quick check of whether conditioning on the epitope beats letting the model find its own spot. State this plainly, as part of a conversation with the user about their target and goals, and let them choose.",{"type":39,"tag":158,"props":159,"children":160},"br",{},[],{"type":45,"value":162},"Phrase it as a question that works ",{"type":39,"tag":164,"props":165,"children":166},"em",{},[167],{"type":45,"value":168},"with",{"type":45,"value":170}," the user (they may know their target's biology), e.g.:",{"type":39,"tag":172,"props":173,"children":174},"blockquote",{},[175],{"type":39,"tag":48,"props":176,"children":177},{},[178],{"type":45,"value":179},"\"This is a fresh target — I'd suggest a quick exploration pass that scouts a few framings and picks the best before a full run. Or, if you already know the site and crop, we can design directly. Which would you like?\"",{"type":39,"tag":158,"props":181,"children":182},{},[],{"type":39,"tag":121,"props":184,"children":185},{},[186],{"type":45,"value":187},"Do not mention a campaign size or tier here",{"type":45,"value":189}," — not even folded into this opening approach question. The full-run size is settled later, after the scouting runs pick a winner (its yield informs the tier), so don't ask it up front when exploration is on the table. If the user opts into exploration — or has already said they want to explore \u002F let the design find its own epitope — read ",{"type":39,"tag":191,"props":192,"children":194},"a",{"href":193},"references\u002Ftarget-exploration.md",[195],{"type":45,"value":193},{"type":45,"value":197},", follow it, then resume at step 8 with the chosen framing and recommended ",{"type":39,"tag":54,"props":199,"children":201},{"className":200},[],[202],{"type":45,"value":203},"num_proteins",{"type":45,"value":205},". If they want to design directly, continue below.",{"type":39,"tag":117,"props":207,"children":208},{},[209,211,217,219,225],{"type":45,"value":210},"Normalize the target (same shape as protein-screen): ",{"type":39,"tag":54,"props":212,"children":214},{"className":213},[],[215],{"type":45,"value":216},"structure_template",{"type":45,"value":218}," if a CIF\u002FPDB is available, else ",{"type":39,"tag":54,"props":220,"children":222},{"className":221},[],[223],{"type":45,"value":224},"no_template",{"type":45,"value":226},".",{"type":39,"tag":117,"props":228,"children":229},{},[230,232,238,240],{"type":45,"value":231},"Pick the ",{"type":39,"tag":54,"props":233,"children":235},{"className":234},[],[236],{"type":45,"value":237},"binder_specification",{"type":45,"value":239}," variant. Supported variants include:",{"type":39,"tag":144,"props":241,"children":242},{},[243,270,304],{"type":39,"tag":117,"props":244,"children":245},{},[246,252,254,260,262,268],{"type":39,"tag":54,"props":247,"children":249},{"className":248},[],[250],{"type":45,"value":251},"boltz_curated",{"type":45,"value":253}," — recommended default for antibody and nanobody design. Boltz selects from maintained scaffold\u002Ftemplate lists (",{"type":39,"tag":54,"props":255,"children":257},{"className":256},[],[258],{"type":45,"value":259},"binder: boltz_antibody",{"type":45,"value":261}," or ",{"type":39,"tag":54,"props":263,"children":265},{"className":264},[],[266],{"type":45,"value":267},"boltz_nanobody",{"type":45,"value":269},").",{"type":39,"tag":117,"props":271,"children":272},{},[273,278,280,286,288,294,296,302],{"type":39,"tag":54,"props":274,"children":276},{"className":275},[],[277],{"type":45,"value":216},{"type":45,"value":279}," — redesign motifs in an existing binder scaffold (CIF + ",{"type":39,"tag":54,"props":281,"children":283},{"className":282},[],[284],{"type":45,"value":285},"design_motifs",{"type":45,"value":287}," with ",{"type":39,"tag":54,"props":289,"children":291},{"className":290},[],[292],{"type":45,"value":293},"replacement",{"type":45,"value":295}," \u002F ",{"type":39,"tag":54,"props":297,"children":299},{"className":298},[],[300],{"type":45,"value":301},"insertion",{"type":45,"value":303}," segments).",{"type":39,"tag":117,"props":305,"children":306},{},[307,312,314,320,321,327],{"type":39,"tag":54,"props":308,"children":310},{"className":309},[],[311],{"type":45,"value":224},{"type":45,"value":313}," — generate from the sequence DSL (fixed residues + designed segments like ",{"type":39,"tag":54,"props":315,"children":317},{"className":316},[],[318],{"type":45,"value":319},"5..10",{"type":45,"value":261},{"type":39,"tag":54,"props":322,"children":324},{"className":323},[],[325],{"type":45,"value":326},"8",{"type":45,"value":269},{"type":39,"tag":117,"props":329,"children":330},{},[331,333,339,341,347],{"type":45,"value":332},"For antibody or nanobody requests, ask before authoring the payload: \"I recommend Boltz's curated antibody\u002Fnanobody scaffolds for this. Do you want the curated default, or do you have custom scaffold structures\u002FCDR motifs to use?\" If the user picks curated, use ",{"type":39,"tag":54,"props":334,"children":336},{"className":335},[],[337],{"type":45,"value":338},"type: boltz_curated",{"type":45,"value":340},"; if they want custom scaffold control, use ",{"type":39,"tag":54,"props":342,"children":344},{"className":343},[],[345],{"type":45,"value":346},"type: structure_template",{"type":45,"value":226},{"type":39,"tag":117,"props":349,"children":350},{},[351,353,359,361,367,369,375,376,382,384,390,392,397,399,404,406,411,413,418,420,425,427,432,434,440],{"type":45,"value":352},"Pick ",{"type":39,"tag":54,"props":354,"children":356},{"className":355},[],[357],{"type":45,"value":358},"modality",{"type":45,"value":360},": ",{"type":39,"tag":54,"props":362,"children":364},{"className":363},[],[365],{"type":45,"value":366},"peptide",{"type":45,"value":368},", ",{"type":39,"tag":54,"props":370,"children":372},{"className":371},[],[373],{"type":45,"value":374},"antibody",{"type":45,"value":368},{"type":39,"tag":54,"props":377,"children":379},{"className":378},[],[380],{"type":45,"value":381},"nanobody",{"type":45,"value":383},", or ",{"type":39,"tag":54,"props":385,"children":387},{"className":386},[],[388],{"type":45,"value":389},"custom_protein",{"type":45,"value":391}," for ",{"type":39,"tag":54,"props":393,"children":395},{"className":394},[],[396],{"type":45,"value":216},{"type":45,"value":398}," and ",{"type":39,"tag":54,"props":400,"children":402},{"className":401},[],[403],{"type":45,"value":224},{"type":45,"value":405}," (use ",{"type":39,"tag":54,"props":407,"children":409},{"className":408},[],[410],{"type":45,"value":389},{"type":45,"value":412}," for a \"miniprotein\" or generic \"protein binder\"). ",{"type":39,"tag":121,"props":414,"children":415},{},[416],{"type":45,"value":417},"If the user already named the modality, take it as given — don't ask again.",{"type":45,"value":419}," Do not include ",{"type":39,"tag":54,"props":421,"children":423},{"className":422},[],[424],{"type":45,"value":358},{"type":45,"value":426}," on ",{"type":39,"tag":54,"props":428,"children":430},{"className":429},[],[431],{"type":45,"value":251},{"type":45,"value":433},"; use ",{"type":39,"tag":54,"props":435,"children":437},{"className":436},[],[438],{"type":45,"value":439},"binder",{"type":45,"value":441}," instead.",{"type":39,"tag":117,"props":443,"children":444},{},[445,446,451,453,459,461,466,468,473,475,480],{"type":45,"value":352},{"type":39,"tag":54,"props":447,"children":449},{"className":448},[],[450],{"type":45,"value":203},{"type":45,"value":452}," — see ",{"type":39,"tag":191,"props":454,"children":456},{"href":455},"#run-sizing",[457],{"type":45,"value":458},"Run sizing",{"type":45,"value":460},". Valid range is ",{"type":39,"tag":121,"props":462,"children":463},{},[464],{"type":45,"value":465},"10 to 1,000,000",{"type":45,"value":467}," (server rejects outside it); ",{"type":39,"tag":121,"props":469,"children":470},{},[471],{"type":45,"value":472},"10",{"type":45,"value":474}," is the hard floor but it is a test size, not a campaign. When the user has not given a count, propose a campaign tier (default ",{"type":39,"tag":121,"props":476,"children":477},{},[478],{"type":45,"value":479},"50,000",{"type":45,"value":481},"), not the floor.",{"type":39,"tag":117,"props":483,"children":484},{},[485,487,493,495,501,503,508],{"type":45,"value":486},"Supported optional features include rules such as excluded amino acids, excluded sequence motifs with ",{"type":39,"tag":54,"props":488,"children":490},{"className":489},[],[491],{"type":45,"value":492},"X",{"type":45,"value":494}," wildcards, and max hydrophobic fraction. Add ",{"type":39,"tag":54,"props":496,"children":498},{"className":497},[],[499],{"type":45,"value":500},"rules",{"type":45,"value":502}," only on request; read ",{"type":39,"tag":191,"props":504,"children":506},{"href":505},"references\u002Fapi.md",[507],{"type":45,"value":505},{"type":45,"value":509}," for exact shapes and examples.",{"type":39,"tag":117,"props":511,"children":512},{},[513,515,520,522,527,529,535,537,542,544,553],{"type":45,"value":514},"Author the payload YAML or JSON, then run ",{"type":39,"tag":54,"props":516,"children":518},{"className":517},[],[519],{"type":45,"value":133},{"type":45,"value":521}," and apply the ",{"type":39,"tag":121,"props":523,"children":524},{},[525],{"type":45,"value":526},"spending gate",{"type":45,"value":528}," (Always Do This) before ",{"type":39,"tag":54,"props":530,"children":532},{"className":531},[],[533],{"type":45,"value":534},"start",{"type":45,"value":536},". (Cost model — tiered by total complex length, ",{"type":39,"tag":54,"props":538,"children":540},{"className":539},[],[541],{"type":45,"value":133},{"type":45,"value":543}," is the only source: see ",{"type":39,"tag":191,"props":545,"children":546},{"href":505},[547],{"type":39,"tag":54,"props":548,"children":550},{"className":549},[],[551],{"type":45,"value":552},"## Cost",{"type":45,"value":554}," in api.md.)",{"type":39,"tag":117,"props":556,"children":557},{},[558,563],{"type":39,"tag":54,"props":559,"children":561},{"className":560},[],[562],{"type":45,"value":534},{"type":45,"value":564}," to submit. Capture the ID.",{"type":39,"tag":117,"props":566,"children":567},{},[568,570,576,578,584,586,591,593,599,601,607,609,615,617,622,624,629],{"type":45,"value":569},"Launch ",{"type":39,"tag":54,"props":571,"children":573},{"className":572},[],[574],{"type":45,"value":575},"download-results",{"type":45,"value":577}," with the agent runtime's background\u002Fnon-blocking command facility. In Claude Code, use Bash with ",{"type":39,"tag":54,"props":579,"children":581},{"className":580},[],[582],{"type":45,"value":583},"run_in_background: true",{"type":45,"value":585},". In Codex, run ",{"type":39,"tag":54,"props":587,"children":589},{"className":588},[],[590],{"type":45,"value":575},{"type":45,"value":592}," as a foreground shell command with ",{"type":39,"tag":54,"props":594,"children":596},{"className":595},[],[597],{"type":45,"value":598},"yield_time_ms: 1000",{"type":45,"value":600},"; if Codex returns a ",{"type":39,"tag":54,"props":602,"children":604},{"className":603},[],[605],{"type":45,"value":606},"session_id",{"type":45,"value":608},", keep it for optional same-thread polling, but treat ",{"type":39,"tag":54,"props":610,"children":612},{"className":611},[],[613],{"type":45,"value":614},"download-status",{"type":45,"value":616}," plus the run directory as the durable source of truth. In Codex app\u002Fdesktop runtimes that expose same-thread heartbeat automations, create a heartbeat that checks ",{"type":39,"tag":54,"props":618,"children":620},{"className":619},[],[621],{"type":45,"value":614},{"type":45,"value":623}," periodically and posts a concise completion or failure update when the download reaches a terminal state. After launching the downloader, always report the job ID, run name, and output directory. Include the next check cadence if the heartbeat was created; otherwise include the ",{"type":39,"tag":54,"props":625,"children":627},{"className":626},[],[628],{"type":45,"value":614},{"type":45,"value":630}," command.",{"type":39,"tag":117,"props":632,"children":633},{},[634,636,642,644,650,652,658,659,665,667,673,675,680],{"type":45,"value":635},"Rank from ",{"type":39,"tag":54,"props":637,"children":639},{"className":638},[],[640],{"type":45,"value":641},"\u003Coutput-root>\u002F\u003Crun-name>\u002Fresults\u002Findex.jsonl",{"type":45,"value":643}," by ",{"type":39,"tag":54,"props":645,"children":647},{"className":646},[],[648],{"type":45,"value":649},"binding_confidence",{"type":45,"value":651}," descending. Use ",{"type":39,"tag":54,"props":653,"children":655},{"className":654},[],[656],{"type":45,"value":657},"iptm",{"type":45,"value":398},{"type":39,"tag":54,"props":660,"children":662},{"className":661},[],[663],{"type":45,"value":664},"min_interaction_pae",{"type":45,"value":666}," as tiebreakers. ",{"type":39,"tag":54,"props":668,"children":670},{"className":669},[],[671],{"type":45,"value":672},"optimization_score",{"type":45,"value":674}," is not emitted for this endpoint. Read ",{"type":39,"tag":191,"props":676,"children":678},{"href":677},"references\u002Fresults.md",[679],{"type":45,"value":677},{"type":45,"value":681}," for output layout and metric details.",{"type":39,"tag":40,"props":683,"children":685},{"id":684},"run-sizing",[686],{"type":45,"value":458},{"type":39,"tag":48,"props":688,"children":689},{},[690,692,697,699,704],{"type":45,"value":691},"De novo design is a generate-and-filter campaign: you make many binders and keep the rare good ones, so a real run is ",{"type":39,"tag":121,"props":693,"children":694},{},[695],{"type":45,"value":696},"large",{"type":45,"value":698},". Do not anchor on the ",{"type":39,"tag":54,"props":700,"children":702},{"className":701},[],[703],{"type":45,"value":203},{"type":45,"value":705}," floor of 10 — that is only useful for a quick setup test. When the user names a count, honor it (≥10). When they do not, explain the tiers and propose one:",{"type":39,"tag":707,"props":708,"children":709},"table",{},[710,737],{"type":39,"tag":711,"props":712,"children":713},"thead",{},[714],{"type":39,"tag":715,"props":716,"children":717},"tr",{},[718,724,732],{"type":39,"tag":719,"props":720,"children":721},"th",{},[722],{"type":45,"value":723},"Tier",{"type":39,"tag":719,"props":725,"children":726},{},[727],{"type":39,"tag":54,"props":728,"children":730},{"className":729},[],[731],{"type":45,"value":203},{"type":39,"tag":719,"props":733,"children":734},{},[735],{"type":45,"value":736},"When",{"type":39,"tag":738,"props":739,"children":740},"tbody",{},[741,760,783],{"type":39,"tag":715,"props":742,"children":743},{},[744,750,755],{"type":39,"tag":745,"props":746,"children":747},"td",{},[748],{"type":45,"value":749},"Small screen",{"type":39,"tag":745,"props":751,"children":752},{},[753],{"type":45,"value":754},"20,000",{"type":39,"tag":745,"props":756,"children":757},{},[758],{"type":45,"value":759},"Quick look \u002F tight budget",{"type":39,"tag":715,"props":761,"children":762},{},[763,771,778],{"type":39,"tag":745,"props":764,"children":765},{},[766],{"type":39,"tag":121,"props":767,"children":768},{},[769],{"type":45,"value":770},"Medium (recommended)",{"type":39,"tag":745,"props":772,"children":773},{},[774],{"type":39,"tag":121,"props":775,"children":776},{},[777],{"type":45,"value":479},{"type":39,"tag":745,"props":779,"children":780},{},[781],{"type":45,"value":782},"Default for a real campaign",{"type":39,"tag":715,"props":784,"children":785},{},[786,791,796],{"type":39,"tag":745,"props":787,"children":788},{},[789],{"type":45,"value":790},"Large",{"type":39,"tag":745,"props":792,"children":793},{},[794],{"type":45,"value":795},"100,000",{"type":39,"tag":745,"props":797,"children":798},{},[799],{"type":45,"value":800},"Hard target \u002F maximal coverage",{"type":39,"tag":48,"props":802,"children":803},{},[804,806,811,813,818,820,825,827,831,833,838],{"type":45,"value":805},"Present the tiers as design ",{"type":39,"tag":121,"props":807,"children":808},{},[809],{"type":45,"value":810},"counts",{"type":45,"value":812},", not dollars: don't put a price next to a tier unless ",{"type":39,"tag":54,"props":814,"children":816},{"className":815},[],[817],{"type":45,"value":133},{"type":45,"value":819}," returned it — run ",{"type":39,"tag":54,"props":821,"children":823},{"className":822},[],[824],{"type":45,"value":133},{"type":45,"value":826}," on the tier the user leans toward and show that figure, and never extrapolate one estimate across the others. Then apply the ",{"type":39,"tag":121,"props":828,"children":829},{},[830],{"type":45,"value":526},{"type":45,"value":832}," (Always Do This) before submitting. If the setup is unproven, suggest a small test run (tens of designs) or the full ",{"type":39,"tag":191,"props":834,"children":835},{"href":193},[836],{"type":45,"value":837},"target-exploration",{"type":45,"value":839}," pass first.",{"type":39,"tag":40,"props":841,"children":843},{"id":842},"command-pattern",[844],{"type":45,"value":845},"Command Pattern",{"type":39,"tag":847,"props":848,"children":853},"pre",{"className":849,"code":850,"language":851,"meta":852,"style":852},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","# Replace placeholders with concrete absolute paths before running.\n# Use a short descriptive run name, for example: protein-design-\u003Cmodality>-\u003Ctarget>-v1\n\nboltz-api protein:design estimate-cost \\\n  --input @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml\n\nboltz-api protein:design start \\\n       --idempotency-key \"\u003Crun-name>\" \\\n       --input @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml \\\n       --raw-output --transform id\n\n# Copy the printed job ID into this command, then launch it in the agent\n# runtime's background\u002Fnon-blocking mode.\n# Claude Code: Bash with run_in_background=true.\n# Codex: foreground shell command with yield_time_ms=1000; keep the returned session_id if one is provided.\n# Do not append \"&\" or use nohup in Codex.\nboltz-api download-results \\\n  --id \"\u003Cjob-id-from-start>\" --name \"\u003Crun-name>\" \\\n  --root-dir \"\u002Fabsolute\u002Fpath\u002Fboltz-experiments\" \\\n  --poll-interval-seconds 60\n","bash","",[854],{"type":39,"tag":54,"props":855,"children":856},{"__ignoreMap":852},[857,869,878,888,914,928,936,957,986,1004,1023,1031,1040,1049,1058,1067,1076,1093,1136,1162],{"type":39,"tag":858,"props":859,"children":862},"span",{"class":860,"line":861},"line",1,[863],{"type":39,"tag":858,"props":864,"children":866},{"style":865},"--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#546E7A;--shiki-default-font-style:italic;--shiki-dark:#676E95;--shiki-dark-font-style:italic",[867],{"type":45,"value":868},"# Replace placeholders with concrete absolute paths before running.\n",{"type":39,"tag":858,"props":870,"children":872},{"class":860,"line":871},2,[873],{"type":39,"tag":858,"props":874,"children":875},{"style":865},[876],{"type":45,"value":877},"# Use a short descriptive run name, for example: protein-design-\u003Cmodality>-\u003Ctarget>-v1\n",{"type":39,"tag":858,"props":879,"children":881},{"class":860,"line":880},3,[882],{"type":39,"tag":858,"props":883,"children":885},{"emptyLinePlaceholder":884},true,[886],{"type":45,"value":887},"\n",{"type":39,"tag":858,"props":889,"children":891},{"class":860,"line":890},4,[892,897,903,908],{"type":39,"tag":858,"props":893,"children":895},{"style":894},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[896],{"type":45,"value":59},{"type":39,"tag":858,"props":898,"children":900},{"style":899},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[901],{"type":45,"value":902}," protein:design",{"type":39,"tag":858,"props":904,"children":905},{"style":899},[906],{"type":45,"value":907}," estimate-cost",{"type":39,"tag":858,"props":909,"children":911},{"style":910},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[912],{"type":45,"value":913}," \\\n",{"type":39,"tag":858,"props":915,"children":917},{"class":860,"line":916},5,[918,923],{"type":39,"tag":858,"props":919,"children":920},{"style":899},[921],{"type":45,"value":922},"  --input",{"type":39,"tag":858,"props":924,"children":925},{"style":899},[926],{"type":45,"value":927}," @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml\n",{"type":39,"tag":858,"props":929,"children":931},{"class":860,"line":930},6,[932],{"type":39,"tag":858,"props":933,"children":934},{"emptyLinePlaceholder":884},[935],{"type":45,"value":887},{"type":39,"tag":858,"props":937,"children":939},{"class":860,"line":938},7,[940,944,948,953],{"type":39,"tag":858,"props":941,"children":942},{"style":894},[943],{"type":45,"value":59},{"type":39,"tag":858,"props":945,"children":946},{"style":899},[947],{"type":45,"value":902},{"type":39,"tag":858,"props":949,"children":950},{"style":899},[951],{"type":45,"value":952}," start",{"type":39,"tag":858,"props":954,"children":955},{"style":910},[956],{"type":45,"value":913},{"type":39,"tag":858,"props":958,"children":960},{"class":860,"line":959},8,[961,966,972,977,982],{"type":39,"tag":858,"props":962,"children":963},{"style":899},[964],{"type":45,"value":965},"       --idempotency-key",{"type":39,"tag":858,"props":967,"children":969},{"style":968},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[970],{"type":45,"value":971}," \"",{"type":39,"tag":858,"props":973,"children":974},{"style":899},[975],{"type":45,"value":976},"\u003Crun-name>",{"type":39,"tag":858,"props":978,"children":979},{"style":968},[980],{"type":45,"value":981},"\"",{"type":39,"tag":858,"props":983,"children":984},{"style":910},[985],{"type":45,"value":913},{"type":39,"tag":858,"props":987,"children":989},{"class":860,"line":988},9,[990,995,1000],{"type":39,"tag":858,"props":991,"children":992},{"style":899},[993],{"type":45,"value":994},"       --input",{"type":39,"tag":858,"props":996,"children":997},{"style":899},[998],{"type":45,"value":999}," @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml",{"type":39,"tag":858,"props":1001,"children":1002},{"style":910},[1003],{"type":45,"value":913},{"type":39,"tag":858,"props":1005,"children":1007},{"class":860,"line":1006},10,[1008,1013,1018],{"type":39,"tag":858,"props":1009,"children":1010},{"style":899},[1011],{"type":45,"value":1012},"       --raw-output",{"type":39,"tag":858,"props":1014,"children":1015},{"style":899},[1016],{"type":45,"value":1017}," --transform",{"type":39,"tag":858,"props":1019,"children":1020},{"style":899},[1021],{"type":45,"value":1022}," id\n",{"type":39,"tag":858,"props":1024,"children":1026},{"class":860,"line":1025},11,[1027],{"type":39,"tag":858,"props":1028,"children":1029},{"emptyLinePlaceholder":884},[1030],{"type":45,"value":887},{"type":39,"tag":858,"props":1032,"children":1034},{"class":860,"line":1033},12,[1035],{"type":39,"tag":858,"props":1036,"children":1037},{"style":865},[1038],{"type":45,"value":1039},"# Copy the printed job ID into this command, then launch it in the agent\n",{"type":39,"tag":858,"props":1041,"children":1043},{"class":860,"line":1042},13,[1044],{"type":39,"tag":858,"props":1045,"children":1046},{"style":865},[1047],{"type":45,"value":1048},"# runtime's background\u002Fnon-blocking mode.\n",{"type":39,"tag":858,"props":1050,"children":1052},{"class":860,"line":1051},14,[1053],{"type":39,"tag":858,"props":1054,"children":1055},{"style":865},[1056],{"type":45,"value":1057},"# Claude Code: Bash with run_in_background=true.\n",{"type":39,"tag":858,"props":1059,"children":1061},{"class":860,"line":1060},15,[1062],{"type":39,"tag":858,"props":1063,"children":1064},{"style":865},[1065],{"type":45,"value":1066},"# Codex: foreground shell command with yield_time_ms=1000; keep the returned session_id if one is provided.\n",{"type":39,"tag":858,"props":1068,"children":1070},{"class":860,"line":1069},16,[1071],{"type":39,"tag":858,"props":1072,"children":1073},{"style":865},[1074],{"type":45,"value":1075},"# Do not append \"&\" or use nohup in Codex.\n",{"type":39,"tag":858,"props":1077,"children":1079},{"class":860,"line":1078},17,[1080,1084,1089],{"type":39,"tag":858,"props":1081,"children":1082},{"style":894},[1083],{"type":45,"value":59},{"type":39,"tag":858,"props":1085,"children":1086},{"style":899},[1087],{"type":45,"value":1088}," download-results",{"type":39,"tag":858,"props":1090,"children":1091},{"style":910},[1092],{"type":45,"value":913},{"type":39,"tag":858,"props":1094,"children":1096},{"class":860,"line":1095},18,[1097,1102,1106,1111,1115,1120,1124,1128,1132],{"type":39,"tag":858,"props":1098,"children":1099},{"style":899},[1100],{"type":45,"value":1101},"  --id",{"type":39,"tag":858,"props":1103,"children":1104},{"style":968},[1105],{"type":45,"value":971},{"type":39,"tag":858,"props":1107,"children":1108},{"style":899},[1109],{"type":45,"value":1110},"\u003Cjob-id-from-start>",{"type":39,"tag":858,"props":1112,"children":1113},{"style":968},[1114],{"type":45,"value":981},{"type":39,"tag":858,"props":1116,"children":1117},{"style":899},[1118],{"type":45,"value":1119}," --name",{"type":39,"tag":858,"props":1121,"children":1122},{"style":968},[1123],{"type":45,"value":971},{"type":39,"tag":858,"props":1125,"children":1126},{"style":899},[1127],{"type":45,"value":976},{"type":39,"tag":858,"props":1129,"children":1130},{"style":968},[1131],{"type":45,"value":981},{"type":39,"tag":858,"props":1133,"children":1134},{"style":910},[1135],{"type":45,"value":913},{"type":39,"tag":858,"props":1137,"children":1139},{"class":860,"line":1138},19,[1140,1145,1149,1154,1158],{"type":39,"tag":858,"props":1141,"children":1142},{"style":899},[1143],{"type":45,"value":1144},"  --root-dir",{"type":39,"tag":858,"props":1146,"children":1147},{"style":968},[1148],{"type":45,"value":971},{"type":39,"tag":858,"props":1150,"children":1151},{"style":899},[1152],{"type":45,"value":1153},"\u002Fabsolute\u002Fpath\u002Fboltz-experiments",{"type":39,"tag":858,"props":1155,"children":1156},{"style":968},[1157],{"type":45,"value":981},{"type":39,"tag":858,"props":1159,"children":1160},{"style":910},[1161],{"type":45,"value":913},{"type":39,"tag":858,"props":1163,"children":1165},{"class":860,"line":1164},20,[1166,1171],{"type":39,"tag":858,"props":1167,"children":1168},{"style":899},[1169],{"type":45,"value":1170},"  --poll-interval-seconds",{"type":39,"tag":858,"props":1172,"children":1174},{"style":1173},"--shiki-light:#F76D47;--shiki-default:#F78C6C;--shiki-dark:#F78C6C",[1175],{"type":45,"value":1176}," 60\n",{"type":39,"tag":48,"props":1178,"children":1179},{},[1180,1182,1187,1188,1194,1195,1200],{"type":45,"value":1181},"Payload keys are ",{"type":39,"tag":54,"props":1183,"children":1185},{"className":1184},[],[1186],{"type":45,"value":203},{"type":45,"value":368},{"type":39,"tag":54,"props":1189,"children":1191},{"className":1190},[],[1192],{"type":45,"value":1193},"target",{"type":45,"value":368},{"type":39,"tag":54,"props":1196,"children":1198},{"className":1197},[],[1199],{"type":45,"value":237},{"type":45,"value":1201}," — API body field names.",{"type":39,"tag":40,"props":1203,"children":1205},{"id":1204},"always-do-this",[1206],{"type":45,"value":1207},"Always Do This",{"type":39,"tag":144,"props":1209,"children":1210},{},[1211,1228,1254,1323,1351,1393,1421,1485,1496,1517,1572,1605,1646,1696,1731,1773,1784,1811],{"type":39,"tag":117,"props":1212,"children":1213},{},[1214,1219,1221,1226],{"type":39,"tag":121,"props":1215,"children":1216},{},[1217],{"type":45,"value":1218},"For a new target, your first move is the step-1 conversation — never jump straight to a payload.",{"type":45,"value":1220}," Establish what the binder is ",{"type":39,"tag":164,"props":1222,"children":1223},{},[1224],{"type":45,"value":1225},"for",{"type":45,"value":1227}," and whether the user has already fixed the binding site\u002Fcrop, then recommend a target-exploration pass vs. designing directly and let them choose (step 1). Only normalize the target and author a payload after that. If they've already fixed the site\u002Fcrop or explicitly want to design directly, proceed.",{"type":39,"tag":117,"props":1229,"children":1230},{},[1231,1233,1239,1241,1246,1248,1252],{"type":45,"value":1232},"Enforce ",{"type":39,"tag":54,"props":1234,"children":1236},{"className":1235},[],[1237],{"type":45,"value":1238},"10 \u003C= num_proteins \u003C= 1,000,000",{"type":45,"value":1240}," before calling ",{"type":39,"tag":54,"props":1242,"children":1244},{"className":1243},[],[1245],{"type":45,"value":133},{"type":45,"value":1247}," (server rejects outside that range), but 10 is the floor, not a campaign — see ",{"type":39,"tag":191,"props":1249,"children":1250},{"href":455},[1251],{"type":45,"value":458},{"type":45,"value":1253}," and propose a tier (default 50,000) when the user gives no count.",{"type":39,"tag":117,"props":1255,"children":1256},{},[1257,1268,1270,1275,1277,1282,1284,1289,1291,1297,1299,1304,1306,1311,1313,1321],{"type":39,"tag":121,"props":1258,"children":1259},{},[1260,1262,1267],{"type":45,"value":1261},"Spending gate — explicit go-ahead before every ",{"type":39,"tag":54,"props":1263,"children":1265},{"className":1264},[],[1266],{"type":45,"value":534},{"type":45,"value":226},{"type":45,"value":1269}," ",{"type":39,"tag":54,"props":1271,"children":1273},{"className":1272},[],[1274],{"type":45,"value":534},{"type":45,"value":1276}," spends real money. A plan you already described, an earlier phase's approval, or a cost that looks \"trivial\" are ",{"type":39,"tag":121,"props":1278,"children":1279},{},[1280],{"type":45,"value":1281},"not",{"type":45,"value":1283}," authorization — even a cheap run needs a fresh yes. Run ",{"type":39,"tag":54,"props":1285,"children":1287},{"className":1286},[],[1288],{"type":45,"value":133},{"type":45,"value":1290},", show the ",{"type":39,"tag":54,"props":1292,"children":1294},{"className":1293},[],[1295],{"type":45,"value":1296},"estimated_cost_usd",{"type":45,"value":1298}," it returns (summed for a batch), and wait for the user to say go. This holds ",{"type":39,"tag":121,"props":1300,"children":1301},{},[1302],{"type":45,"value":1303},"even when tool calls are pre-approved",{"type":45,"value":1305}," (accept-edits \u002F auto-accept \u002F bypass modes) — there you are the only cost gate. Never quote or assume a dollar figure you didn't get from ",{"type":39,"tag":54,"props":1307,"children":1309},{"className":1308},[],[1310],{"type":45,"value":133},{"type":45,"value":1312}," (cost model: ",{"type":39,"tag":191,"props":1314,"children":1315},{"href":505},[1316],{"type":39,"tag":54,"props":1317,"children":1319},{"className":1318},[],[1320],{"type":45,"value":552},{"type":45,"value":1322}," in api.md).",{"type":39,"tag":117,"props":1324,"children":1325},{},[1326,1328,1334,1336,1341,1343,1349],{"type":45,"value":1327},"For antibody or nanobody design, recommend ",{"type":39,"tag":54,"props":1329,"children":1331},{"className":1330},[],[1332],{"type":45,"value":1333},"binder_specification.type: boltz_curated",{"type":45,"value":1335}," and ask the user to confirm they do not want custom scaffold\u002FCDR control before building the payload. Use ",{"type":39,"tag":54,"props":1337,"children":1339},{"className":1338},[],[1340],{"type":45,"value":259},{"type":45,"value":1342}," for antibody\u002FFab requests and ",{"type":39,"tag":54,"props":1344,"children":1346},{"className":1345},[],[1347],{"type":45,"value":1348},"binder: boltz_nanobody",{"type":45,"value":1350}," for nanobody\u002FVHH requests.",{"type":39,"tag":117,"props":1352,"children":1353},{},[1354,1356,1362,1364,1370,1371,1377,1378,1384,1385,1391],{"type":45,"value":1355},"Residue indices are 0-based everywhere (",{"type":39,"tag":54,"props":1357,"children":1359},{"className":1358},[],[1360],{"type":45,"value":1361},"design_motifs.start_index",{"type":45,"value":1363},"\u002F",{"type":39,"tag":54,"props":1365,"children":1367},{"className":1366},[],[1368],{"type":45,"value":1369},"end_index",{"type":45,"value":368},{"type":39,"tag":54,"props":1372,"children":1374},{"className":1373},[],[1375],{"type":45,"value":1376},"after_residue_index",{"type":45,"value":368},{"type":39,"tag":54,"props":1379,"children":1381},{"className":1380},[],[1382],{"type":45,"value":1383},"epitope_residues",{"type":45,"value":368},{"type":39,"tag":54,"props":1386,"children":1388},{"className":1387},[],[1389],{"type":45,"value":1390},"flexible_residues",{"type":45,"value":1392},", bonds, constraints).",{"type":39,"tag":117,"props":1394,"children":1395},{},[1396,1398,1404,1406,1412,1414,1420],{"type":45,"value":1397},"For CIF\u002FPDB bytes, use ",{"type":39,"tag":54,"props":1399,"children":1401},{"className":1400},[],[1402],{"type":45,"value":1403},"@data:\u002F\u002F\u002Fabs\u002Fpath\u002Ffile.cif",{"type":45,"value":1405}," inside ",{"type":39,"tag":54,"props":1407,"children":1409},{"className":1408},[],[1410],{"type":45,"value":1411},"structure.data",{"type":45,"value":1413},". Don't use bare ",{"type":39,"tag":54,"props":1415,"children":1417},{"className":1416},[],[1418],{"type":45,"value":1419},"@path",{"type":45,"value":226},{"type":39,"tag":117,"props":1422,"children":1423},{},[1424,1426,1432,1434,1440,1442,1447,1449,1455,1457,1463,1464,1470,1471,1477,1478,1484],{"type":45,"value":1425},"Sequence DSL for ",{"type":39,"tag":54,"props":1427,"children":1429},{"className":1428},[],[1430],{"type":45,"value":1431},"designed_protein.value",{"type":45,"value":1433},": uppercase letters = fixed residues; integer ",{"type":39,"tag":54,"props":1435,"children":1437},{"className":1436},[],[1438],{"type":45,"value":1439},"N",{"type":45,"value":1441}," = exactly ",{"type":39,"tag":54,"props":1443,"children":1445},{"className":1444},[],[1446],{"type":45,"value":1439},{"type":45,"value":1448}," designed residues; ",{"type":39,"tag":54,"props":1450,"children":1452},{"className":1451},[],[1453],{"type":45,"value":1454},"MIN..MAX",{"type":45,"value":1456}," = variable-length designed segment. Examples: ",{"type":39,"tag":54,"props":1458,"children":1460},{"className":1459},[],[1461],{"type":45,"value":1462},"\"20\"",{"type":45,"value":368},{"type":39,"tag":54,"props":1465,"children":1467},{"className":1466},[],[1468],{"type":45,"value":1469},"\"5..10\"",{"type":45,"value":368},{"type":39,"tag":54,"props":1472,"children":1474},{"className":1473},[],[1475],{"type":45,"value":1476},"\"ACDE8GHI\"",{"type":45,"value":368},{"type":39,"tag":54,"props":1479,"children":1481},{"className":1480},[],[1482],{"type":45,"value":1483},"\"MKTAYI5..10VKSHFSRQ\"",{"type":45,"value":226},{"type":39,"tag":117,"props":1486,"children":1487},{},[1488,1490,1495],{"type":45,"value":1489},"Keep payload field names exactly as the API body names shown in ",{"type":39,"tag":54,"props":1491,"children":1493},{"className":1492},[],[1494],{"type":45,"value":505},{"type":45,"value":226},{"type":39,"tag":117,"props":1497,"children":1498},{},[1499,1501,1507,1509,1515],{"type":45,"value":1500},"Use absolute paths for the output root, payload files, and embedded target files. Do not ",{"type":39,"tag":54,"props":1502,"children":1504},{"className":1503},[],[1505],{"type":45,"value":1506},"cd",{"type":45,"value":1508}," into the run directory for follow-up commands; pass the same ",{"type":39,"tag":54,"props":1510,"children":1512},{"className":1511},[],[1513],{"type":45,"value":1514},"--root-dir",{"type":45,"value":1516}," and use absolute paths so later relative paths do not drift.",{"type":39,"tag":117,"props":1518,"children":1519},{},[1520,1522,1528,1529,1535,1536,1541,1542,1547,1549,1555,1556,1562,1564,1570],{"type":45,"value":1521},"Prefer one merged top-level payload via ",{"type":39,"tag":54,"props":1523,"children":1525},{"className":1524},[],[1526],{"type":45,"value":1527},"--input @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.yaml",{"type":45,"value":261},{"type":39,"tag":54,"props":1530,"children":1532},{"className":1531},[],[1533],{"type":45,"value":1534},"@json:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fpayload.json",{"type":45,"value":391},{"type":39,"tag":54,"props":1537,"children":1539},{"className":1538},[],[1540],{"type":45,"value":133},{"type":45,"value":398},{"type":39,"tag":54,"props":1543,"children":1545},{"className":1544},[],[1546],{"type":45,"value":534},{"type":45,"value":1548},". Keep ",{"type":39,"tag":54,"props":1550,"children":1552},{"className":1551},[],[1553],{"type":45,"value":1554},"--idempotency-key",{"type":45,"value":398},{"type":39,"tag":54,"props":1557,"children":1559},{"className":1558},[],[1560],{"type":45,"value":1561},"--workspace-id",{"type":45,"value":1563}," top-level; if they also appear inside ",{"type":39,"tag":54,"props":1565,"children":1567},{"className":1566},[],[1568],{"type":45,"value":1569},"--input",{"type":45,"value":1571},", the top-level flags win.",{"type":39,"tag":117,"props":1573,"children":1574},{},[1575,1577,1583,1584,1590,1592,1597,1598,1604],{"type":45,"value":1576},"Direct object flags still work as overrides, such as ",{"type":39,"tag":54,"props":1578,"children":1580},{"className":1579},[],[1581],{"type":45,"value":1582},"--target @yaml:\u002F\u002F\u002Fabsolute\u002Fpath\u002Ftarget.yaml",{"type":45,"value":261},{"type":39,"tag":54,"props":1585,"children":1587},{"className":1586},[],[1588],{"type":45,"value":1589},"--binder-specification @json:\u002F\u002F\u002Fabsolute\u002Fpath\u002Fbinder.json",{"type":45,"value":1591},". Piped YAML \u002F JSON on stdin also works, but it must use API body field names. Use the same slug for both ",{"type":39,"tag":54,"props":1593,"children":1595},{"className":1594},[],[1596],{"type":45,"value":1554},{"type":45,"value":398},{"type":39,"tag":54,"props":1599,"children":1601},{"className":1600},[],[1602],{"type":45,"value":1603},"--name",{"type":45,"value":226},{"type":39,"tag":117,"props":1606,"children":1607},{},[1608,1610,1615,1617,1623,1625,1630,1632,1638,1640,1645],{"type":45,"value":1609},"In permission-gated agents such as Claude Code, keep each Boltz call as a top-level command that starts with ",{"type":39,"tag":54,"props":1611,"children":1613},{"className":1612},[],[1614],{"type":45,"value":59},{"type":45,"value":1616},". Prefer concrete arguments over ",{"type":39,"tag":54,"props":1618,"children":1620},{"className":1619},[],[1621],{"type":45,"value":1622},"sh -c",{"type":45,"value":1624},", inline environment assignments, aliases, wrapper scripts, loops, or pipelines around the ",{"type":39,"tag":54,"props":1626,"children":1628},{"className":1627},[],[1629],{"type":45,"value":59},{"type":45,"value":1631}," invocation unless the user already allowed that exact command form. Use ",{"type":39,"tag":54,"props":1633,"children":1635},{"className":1634},[],[1636],{"type":45,"value":1637},"--raw-output --transform id",{"type":45,"value":1639},", read the printed ID, then paste that literal ID into the next ",{"type":39,"tag":54,"props":1641,"children":1643},{"className":1642},[],[1644],{"type":45,"value":575},{"type":45,"value":630},{"type":39,"tag":117,"props":1647,"children":1648},{},[1649,1651,1656,1658,1663,1665,1670,1672,1678,1680,1686,1688,1694],{"type":45,"value":1650},"Prefer the agent runtime's background\u002Fnon-blocking command mode for ",{"type":39,"tag":54,"props":1652,"children":1654},{"className":1653},[],[1655],{"type":45,"value":575},{"type":45,"value":1657},". In Codex specifically, keep ",{"type":39,"tag":54,"props":1659,"children":1661},{"className":1660},[],[1662],{"type":45,"value":575},{"type":45,"value":1664}," in the foreground and set the shell tool yield to 1000 ms; Codex will return a ",{"type":39,"tag":54,"props":1666,"children":1668},{"className":1667},[],[1669],{"type":45,"value":606},{"type":45,"value":1671}," if the command is still running. Do not append ",{"type":39,"tag":54,"props":1673,"children":1675},{"className":1674},[],[1676],{"type":45,"value":1677},"&",{"type":45,"value":1679}," or use ",{"type":39,"tag":54,"props":1681,"children":1683},{"className":1682},[],[1684],{"type":45,"value":1685},"nohup",{"type":45,"value":1687}," in Codex because the tool runner may clean up shell-backgrounded descendants before ",{"type":39,"tag":54,"props":1689,"children":1691},{"className":1690},[],[1692],{"type":45,"value":1693},".boltz-run.json",{"type":45,"value":1695}," is fully written.",{"type":39,"tag":117,"props":1697,"children":1698},{},[1699,1701,1706,1708,1714,1716,1721,1723,1729],{"type":45,"value":1700},"After the background\u002Fsession starts, do not manually wait on it or run ad hoc polling loops. Wall-clock time scales roughly with ",{"type":39,"tag":54,"props":1702,"children":1704},{"className":1703},[],[1705],{"type":45,"value":203},{"type":45,"value":1707},": under 100 often finishes in a few minutes, 100-1,000 may take several minutes to tens of minutes, and larger runs can take longer or hours depending on inputs and system load. Don't quote a fixed duration. ",{"type":39,"tag":54,"props":1709,"children":1711},{"className":1710},[],[1712],{"type":45,"value":1713},"--poll-interval-seconds 60",{"type":45,"value":1715}," is a sensible downloader default. ",{"type":39,"tag":54,"props":1717,"children":1719},{"className":1718},[],[1720],{"type":45,"value":575},{"type":45,"value":1722}," emits JSONL progress on stderr by default; add ",{"type":39,"tag":54,"props":1724,"children":1726},{"className":1725},[],[1727],{"type":45,"value":1728},"--progress-format text --verbose",{"type":45,"value":1730}," only when you explicitly want human-readable logs.",{"type":39,"tag":117,"props":1732,"children":1733},{},[1734,1736,1741,1743,1749,1751,1756,1758,1763,1765,1771],{"type":45,"value":1735},"In Codex app\u002Fdesktop runtimes with same-thread heartbeat automation support, schedule a heartbeat after launching ",{"type":39,"tag":54,"props":1737,"children":1739},{"className":1738},[],[1740],{"type":45,"value":575},{"type":45,"value":1742},". The heartbeat should run ",{"type":39,"tag":54,"props":1744,"children":1746},{"className":1745},[],[1747],{"type":45,"value":1748},"boltz-api --format json download-status --name \"\u003Crun-name>\" --root-dir \"\u002Fabsolute\u002Fpath\u002Fboltz-experiments\"",{"type":45,"value":1750}," and stop once terminal. Choose cadence by ",{"type":39,"tag":54,"props":1752,"children":1754},{"className":1753},[],[1755],{"type":45,"value":203},{"type":45,"value":1757},": under 100 -> every 1-2 minutes; 100-1,000 -> every 5 minutes; over 1,000 -> every 15 minutes. Post only material status changes or terminal completion\u002Ffailure. Poll the saved ",{"type":39,"tag":54,"props":1759,"children":1761},{"className":1760},[],[1762],{"type":45,"value":606},{"type":45,"value":1764}," with an empty ",{"type":39,"tag":54,"props":1766,"children":1768},{"className":1767},[],[1769],{"type":45,"value":1770},"write_stdin",{"type":45,"value":1772}," only for interactive, user-requested progress checks. Never run a manual poll loop in the current turn.",{"type":39,"tag":117,"props":1774,"children":1775},{},[1776,1778,1783],{"type":45,"value":1777},"If the current host has no heartbeat automation support, do not claim an automatic next check. Report the job ID, run name, output directory, and the command needed to check ",{"type":39,"tag":54,"props":1779,"children":1781},{"className":1780},[],[1782],{"type":45,"value":614},{"type":45,"value":226},{"type":39,"tag":117,"props":1785,"children":1786},{},[1787,1789,1795,1797,1803,1805,1810],{"type":45,"value":1788},"If detached download needs to be restarted, re-run ",{"type":39,"tag":54,"props":1790,"children":1792},{"className":1791},[],[1793],{"type":45,"value":1794},"boltz-api download-results",{"type":45,"value":1796}," with the same ",{"type":39,"tag":54,"props":1798,"children":1800},{"className":1799},[],[1801],{"type":45,"value":1802},"--name \"\u003Crun-name>\"",{"type":45,"value":1804}," and the same ",{"type":39,"tag":54,"props":1806,"children":1808},{"className":1807},[],[1809],{"type":45,"value":1514},{"type":45,"value":226},{"type":39,"tag":117,"props":1812,"children":1813},{},[1814,1816,1821],{"type":45,"value":1815},"Only add ",{"type":39,"tag":54,"props":1817,"children":1819},{"className":1818},[],[1820],{"type":45,"value":500},{"type":45,"value":1822}," on explicit user request.",{"type":39,"tag":40,"props":1824,"children":1826},{"id":1825},"escape-hatch",[1827],{"type":45,"value":1828},"Escape Hatch",{"type":39,"tag":144,"props":1830,"children":1831},{},[1832,1844],{"type":39,"tag":117,"props":1833,"children":1834},{},[1835,1837],{"type":45,"value":1836},"Payload reference: ",{"type":39,"tag":191,"props":1838,"children":1842},{"href":1839,"rel":1840},"https:\u002F\u002Fapi.boltz.bio\u002Fdocs\u002Fapi\u002Fresources\u002Fprotein\u002Fsubresources\u002Fdesign\u002Fmethods\u002Fstart\u002F",[1841],"nofollow",[1843],{"type":45,"value":1839},{"type":39,"tag":117,"props":1845,"children":1846},{},[1847,1849],{"type":45,"value":1848},"CLI flag names: ",{"type":39,"tag":54,"props":1850,"children":1852},{"className":1851},[],[1853],{"type":45,"value":1854},"boltz-api protein:design start --help",{"type":39,"tag":48,"props":1856,"children":1857},{},[1858,1860,1864,1866,1871,1873,1878,1880,1884],{"type":45,"value":1859},"Read ",{"type":39,"tag":191,"props":1861,"children":1862},{"href":505},[1863],{"type":45,"value":505},{"type":45,"value":1865}," for all ",{"type":39,"tag":54,"props":1867,"children":1869},{"className":1868},[],[1870],{"type":45,"value":237},{"type":45,"value":1872}," variants, motif shapes, sequence DSL, rules, modalities, and ",{"type":39,"tag":54,"props":1874,"children":1876},{"className":1875},[],[1877],{"type":45,"value":1193},{"type":45,"value":1879}," variants. Read ",{"type":39,"tag":191,"props":1881,"children":1882},{"href":677},[1883],{"type":45,"value":677},{"type":45,"value":1885}," after download when ranking designed binders or explaining outputs.",{"type":39,"tag":40,"props":1887,"children":1889},{"id":1888},"outputs",[1890],{"type":45,"value":1891},"Outputs",{"type":39,"tag":48,"props":1893,"children":1894},{},[1895,1896,1902,1904,1909,1910,1914],{"type":45,"value":635},{"type":39,"tag":54,"props":1897,"children":1899},{"className":1898},[],[1900],{"type":45,"value":1901},"results\u002Findex.jsonl",{"type":45,"value":1903}," after ",{"type":39,"tag":54,"props":1905,"children":1907},{"className":1906},[],[1908],{"type":45,"value":575},{"type":45,"value":433},{"type":39,"tag":191,"props":1911,"children":1912},{"href":677},[1913],{"type":45,"value":677},{"type":45,"value":1915}," for local file layout, metric meanings, and the designed-binder entity type gotcha.",{"type":39,"tag":1917,"props":1918,"children":1919},"style",{},[1920],{"type":45,"value":1921},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":1923,"total":2044},[1924,1942,1958,1970,1990,2012,2032],{"slug":1925,"name":1925,"fn":1926,"description":1927,"org":1928,"tags":1929,"stars":22,"repoUrl":23,"updatedAt":24},"accessibility-and-inclusive-visualization","make data visualizations accessible","Make data visualizations accessible and inclusive. Use when the user needs chart or diagram accessibility guidance, text alternatives for complex visuals, color and contrast review, keyboard support, reduced-motion behavior for animation or parallax, or an accessibility QA workflow for exported figures, UML-like diagrams, and dashboards.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1930,1933,1936,1939],{"name":1931,"slug":1932,"type":15},"Accessibility","accessibility",{"name":1934,"slug":1935,"type":15},"Charts","charts",{"name":1937,"slug":1938,"type":15},"Data Visualization","data-visualization",{"name":1940,"slug":1941,"type":15},"Design","design",{"slug":1943,"name":1943,"fn":1944,"description":1945,"org":1946,"tags":1947,"stars":22,"repoUrl":23,"updatedAt":1957},"agent-browser","automate browser interactions for agents","Browser automation CLI for AI agents. Use when the user needs to interact with websites, verify dev server output, test web apps, navigate pages, fill forms, click buttons, take screenshots, extract data, or automate any browser task. Also triggers when a dev server starts so you can verify it visually.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1948,1951,1954],{"name":1949,"slug":1950,"type":15},"Agents","agents",{"name":1952,"slug":1953,"type":15},"Browser Automation","browser-automation",{"name":1955,"slug":1956,"type":15},"Testing","testing","2026-04-06T18:41:03.44016",{"slug":1959,"name":1959,"fn":1960,"description":1961,"org":1962,"tags":1963,"stars":22,"repoUrl":23,"updatedAt":1969},"agent-browser-verify","verify dev server output with automated browser","Automated browser verification for dev servers. Triggers when a dev server starts to run a visual gut-check with agent-browser — verifies the page loads, checks for console errors, validates key UI elements, and reports pass\u002Ffail before continuing.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1964,1965,1968],{"name":1952,"slug":1953,"type":15},{"name":1966,"slug":1967,"type":15},"Local Development","local-development",{"name":1955,"slug":1956,"type":15},"2026-04-06T18:41:17.526867",{"slug":1971,"name":1971,"fn":1972,"description":1973,"org":1974,"tags":1975,"stars":22,"repoUrl":23,"updatedAt":1989},"agents-sdk","build AI agents on Cloudflare Workers","Build AI agents on Cloudflare Workers using the Agents SDK. Load when creating stateful agents, durable workflows, real-time WebSocket apps, scheduled tasks, MCP servers, or chat applications. Covers Agent class, state management, callable RPC, Workflows integration, and React hooks. Biases towards retrieval from Cloudflare docs over pre-trained knowledge.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1976,1977,1980,1983,1986],{"name":1949,"slug":1950,"type":15},{"name":1978,"slug":1979,"type":15},"Cloudflare Workers","cloudflare-workers",{"name":1981,"slug":1982,"type":15},"SDK","sdk",{"name":1984,"slug":1985,"type":15},"Serverless","serverless",{"name":1987,"slug":1988,"type":15},"WebSockets","websockets","2026-04-06T18:39:51.717063",{"slug":1991,"name":1991,"fn":1992,"description":1993,"org":1994,"tags":1995,"stars":22,"repoUrl":23,"updatedAt":2011},"ai-elements","build chat UIs with AI Elements","AI Elements component library guidance — pre-built React components for AI interfaces built on shadcn\u002Fui. Use when building chat UIs, message displays, tool call rendering, streaming responses, reasoning panels, or any AI-native interface with the AI SDK.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1996,1999,2002,2005,2008],{"name":1997,"slug":1998,"type":15},"Frontend","frontend",{"name":2000,"slug":2001,"type":15},"React","react",{"name":2003,"slug":2004,"type":15},"shadcn\u002Fui","shadcn-ui",{"name":2006,"slug":2007,"type":15},"UI Components","ui-components",{"name":2009,"slug":2010,"type":15},"Vercel","vercel","2026-04-06T18:40:59.619419",{"slug":2013,"name":2013,"fn":2014,"description":2015,"org":2016,"tags":2017,"stars":22,"repoUrl":23,"updatedAt":2031},"ai-gateway","configure Vercel AI Gateway","Vercel AI Gateway expert guidance. Use when configuring model routing, provider failover, cost tracking, or managing multiple AI providers through a unified API.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2018,2021,2024,2027,2030],{"name":2019,"slug":2020,"type":15},"AI Infrastructure","ai-infrastructure",{"name":2022,"slug":2023,"type":15},"Cost Optimization","cost-optimization",{"name":2025,"slug":2026,"type":15},"LLM","llm",{"name":2028,"slug":2029,"type":15},"Performance","performance",{"name":2009,"slug":2010,"type":15},"2026-04-06T18:40:44.377464",{"slug":2033,"name":2033,"fn":2034,"description":2035,"org":2036,"tags":2037,"stars":22,"repoUrl":23,"updatedAt":2043},"ai-generation-persistence","implement persistence patterns for AI generations","AI generation persistence patterns — unique IDs, addressable URLs, database storage, and cost tracking for every LLM generation",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2038,2039,2042],{"name":2022,"slug":2023,"type":15},{"name":2040,"slug":2041,"type":15},"Database","database",{"name":2025,"slug":2026,"type":15},"2026-04-06T18:41:08.513425",600,{"items":2046,"total":2243},[2047,2068,2091,2108,2124,2141,2160,2172,2186,2200,2212,2227],{"slug":2048,"name":2048,"fn":2049,"description":2050,"org":2051,"tags":2052,"stars":2065,"repoUrl":2066,"updatedAt":2067},"prior-auth-packet-builder","build healthcare prior authorization packets","Build a concise prior authorization packet from local case files and payer policy docs.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2053,2056,2059,2062],{"name":2054,"slug":2055,"type":15},"Documents","documents",{"name":2057,"slug":2058,"type":15},"Healthcare","healthcare",{"name":2060,"slug":2061,"type":15},"Insurance","insurance",{"name":2063,"slug":2064,"type":15},"Regulatory Compliance","regulatory-compliance",28169,"https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python","2026-04-16T05:11:39.180399",{"slug":2069,"name":2069,"fn":2070,"description":2071,"org":2072,"tags":2073,"stars":2088,"repoUrl":2089,"updatedAt":2090},"aspnet-core","build ASP.NET Core web applications","Build, review, refactor, or architect ASP.NET Core web applications using current official guidance for .NET web development. Use when working on Blazor Web Apps, Razor Pages, MVC, Minimal APIs, controller-based Web APIs, SignalR, gRPC, middleware, dependency injection, configuration, authentication, authorization, testing, performance, deployment, or ASP.NET Core upgrades.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2074,2077,2079,2082,2085],{"name":2075,"slug":2076,"type":15},".NET","dotnet",{"name":2078,"slug":2069,"type":15},"ASP.NET Core",{"name":2080,"slug":2081,"type":15},"Blazor","blazor",{"name":2083,"slug":2084,"type":15},"C#","csharp",{"name":2086,"slug":2087,"type":15},"Web Development","web-development",23787,"https:\u002F\u002Fgithub.com\u002Fopenai\u002Fskills","2026-04-12T05:07:02.819491",{"slug":2092,"name":2092,"fn":2093,"description":2094,"org":2095,"tags":2096,"stars":2088,"repoUrl":2089,"updatedAt":2107},"chatgpt-apps","build ChatGPT Apps SDK applications","Build, scaffold, refactor, and troubleshoot ChatGPT Apps SDK applications that combine an MCP server and widget UI. Use when Codex needs to design tools, register UI resources, wire the MCP Apps bridge or ChatGPT compatibility APIs, apply Apps SDK metadata or CSP or domain settings, or produce a docs-aligned project scaffold. Prefer a docs-first workflow by invoking the openai-docs skill or OpenAI developer docs MCP tools before generating code.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2097,2100,2103,2106],{"name":2098,"slug":2099,"type":15},"Apps SDK","apps-sdk",{"name":2101,"slug":2102,"type":15},"ChatGPT","chatgpt",{"name":2104,"slug":2105,"type":15},"MCP","mcp",{"name":9,"slug":8,"type":15},"2026-04-12T05:07:05.468097",{"slug":2109,"name":2109,"fn":2110,"description":2111,"org":2112,"tags":2113,"stars":2088,"repoUrl":2089,"updatedAt":2123},"cli-creator","build CLIs from API docs","Build a composable CLI for Codex from API docs, an OpenAPI spec, existing curl examples, an SDK, a web app, an admin tool, or a local script. Use when the user wants Codex to create a command-line tool that can run from any repo, expose composable read\u002Fwrite commands, return stable JSON, manage auth, and pair with a companion skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2114,2117,2120],{"name":2115,"slug":2116,"type":15},"API Development","api-development",{"name":2118,"slug":2119,"type":15},"CLI","cli",{"name":2121,"slug":2122,"type":15},"Codex","codex","2026-04-12T05:07:04.132762",{"slug":2125,"name":2125,"fn":2126,"description":2127,"org":2128,"tags":2129,"stars":2088,"repoUrl":2089,"updatedAt":2140},"cloudflare-deploy","deploy projects to Cloudflare","Deploy applications and infrastructure to Cloudflare using Workers, Pages, and related platform services. Use when the user asks to deploy, host, publish, or set up a project on Cloudflare.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2130,2133,2136,2137],{"name":2131,"slug":2132,"type":15},"Cloudflare","cloudflare",{"name":2134,"slug":2135,"type":15},"Cloudflare Pages","cloudflare-pages",{"name":1978,"slug":1979,"type":15},{"name":2138,"slug":2139,"type":15},"Deployment","deployment","2026-04-12T05:07:14.275118",{"slug":2142,"name":2142,"fn":2143,"description":2144,"org":2145,"tags":2146,"stars":2088,"repoUrl":2089,"updatedAt":2159},"define-goal","define and set measurable project goals","Help the user define a concrete, measurable goal before starting work, especially when they ask to use the goal tool, create a goal, set an objective, clarify success criteria, or turn a fuzzy intention into a quantitative outcome. Use this skill for goal creation and goal refinement only; it does not manage durable snapshots, decision logs, or long-running execution artifacts.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2147,2150,2153,2156],{"name":2148,"slug":2149,"type":15},"Productivity","productivity",{"name":2151,"slug":2152,"type":15},"Project Management","project-management",{"name":2154,"slug":2155,"type":15},"Strategy","strategy",{"name":2157,"slug":2158,"type":15},"Task Management","task-management","2026-05-23T06:17:16.870838",{"slug":2161,"name":2161,"fn":2162,"description":2163,"org":2164,"tags":2165,"stars":2088,"repoUrl":2089,"updatedAt":2171},"figma","translate Figma designs into code","Use the Figma MCP server to fetch design context, screenshots, variables, and assets from Figma, and to translate Figma nodes into production code. Trigger when a task involves Figma URLs, node IDs, design-to-code implementation, or Figma MCP setup and troubleshooting.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2166,2167,2169,2170],{"name":1940,"slug":1941,"type":15},{"name":2168,"slug":2161,"type":15},"Figma",{"name":1997,"slug":1998,"type":15},{"name":2104,"slug":2105,"type":15},"2026-04-12T05:06:47.939943",{"slug":2173,"name":2173,"fn":2174,"description":2175,"org":2176,"tags":2177,"stars":2088,"repoUrl":2089,"updatedAt":2185},"figma-code-connect-components","connect Figma designs to code components","Connects Figma design components to code components using Code Connect mapping tools. Use when user says \"code connect\", \"connect this component to code\", \"map this component\", \"link component to code\", \"create code connect mapping\", or wants to establish mappings between Figma designs and code implementations. For canvas writes via `use_figma`, use `figma-use`.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2178,2179,2182,2183,2184],{"name":1940,"slug":1941,"type":15},{"name":2180,"slug":2181,"type":15},"Design System","design-system",{"name":2168,"slug":2161,"type":15},{"name":1997,"slug":1998,"type":15},{"name":2006,"slug":2007,"type":15},"2026-05-10T05:59:52.971881",{"slug":2187,"name":2187,"fn":2188,"description":2189,"org":2190,"tags":2191,"stars":2088,"repoUrl":2089,"updatedAt":2199},"figma-create-design-system-rules","generate design system rules from Figma","Generates custom design system rules for the user's codebase. Use when user says \"create design system rules\", \"generate rules for my project\", \"set up design rules\", \"customize design system guidelines\", or wants to establish project-specific conventions for Figma-to-code workflows. Requires Figma MCP server connection.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2192,2193,2194,2197,2198],{"name":1940,"slug":1941,"type":15},{"name":2180,"slug":2181,"type":15},{"name":2195,"slug":2196,"type":15},"Documentation","documentation",{"name":2168,"slug":2161,"type":15},{"name":1997,"slug":1998,"type":15},"2026-05-16T06:07:47.821474",{"slug":2201,"name":2201,"fn":2202,"description":2203,"org":2204,"tags":2205,"stars":2088,"repoUrl":2089,"updatedAt":2211},"figma-implement-design","translate Figma designs into application code","Translates Figma designs into production-ready application code with 1:1 visual fidelity. Use when implementing UI code from Figma files, when user mentions \"implement design\", \"generate code\", \"implement component\", provides Figma URLs, or asks to build components matching Figma specs. For Figma canvas writes via `use_figma`, use `figma-use`.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2206,2207,2208,2209,2210],{"name":1940,"slug":1941,"type":15},{"name":2168,"slug":2161,"type":15},{"name":1997,"slug":1998,"type":15},{"name":2006,"slug":2007,"type":15},{"name":2086,"slug":2087,"type":15},"2026-05-16T06:07:40.583615",{"slug":2213,"name":2213,"fn":2214,"description":2215,"org":2216,"tags":2217,"stars":2088,"repoUrl":2089,"updatedAt":2226},"hatch-pet","create animated pets for Codex","Create, repair, validate, visually QA, and package Codex-compatible animated pets and pet spritesheets from character art, generated images, company or prospect brand cues, or visual references. Use when a user wants a lightweight-worker Codex pet workflow, a non-pixel custom pet style, a prospect or company mascot pet, or a full 8x9 animated pet atlas with transparent unused cells, QA contact sheets, and pet.json packaging. This skill composes the installed $imagegen system skill for visual generation and uses bundled scripts for deterministic spritesheet assembly.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2218,2221,2222,2225],{"name":2219,"slug":2220,"type":15},"Animation","animation",{"name":2121,"slug":2122,"type":15},{"name":2223,"slug":2224,"type":15},"Creative","creative",{"name":1940,"slug":1941,"type":15},"2026-05-02T05:31:48.48485",{"slug":2228,"name":2228,"fn":2229,"description":2230,"org":2231,"tags":2232,"stars":2088,"repoUrl":2089,"updatedAt":2242},"imagegen","generate and edit raster images","Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output should be a bitmap asset rather than repo-native code or vector. Do not use when the task is better handled by editing existing SVG\u002Fvector\u002Fcode-native assets, extending an established icon or logo system, or building the visual directly in HTML\u002FCSS\u002Fcanvas.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[2233,2234,2235,2238,2241],{"name":2223,"slug":2224,"type":15},{"name":1940,"slug":1941,"type":15},{"name":2236,"slug":2237,"type":15},"Image Generation","image-generation",{"name":2239,"slug":2240,"type":15},"Images","images",{"name":9,"slug":8,"type":15},"2026-05-15T06:23:24.312127",675]