[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-anthropic-commerce-merchant-operations":3,"mdc-52o3ki-key":37,"related-org-anthropic-commerce-merchant-operations":1121,"related-repo-anthropic-commerce-merchant-operations":1307},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":26,"repoUrl":27,"updatedAt":28,"license":29,"forks":30,"topics":31,"repo":32,"sourceUrl":35,"mdContent":36},"commerce-merchant-operations","manage merchant operations and store rules","The reference merchant agent, covering its flows, staged changes and host approval, metrics grounding, the analysis delegate, store memory, components, and marketplace rules. Load when building or reviewing an agent for the operators of a store, property, subscriber base, or venue.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"anthropic","Anthropic","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fanthropic.png","anthropics",[13,17,20,23],{"name":14,"slug":15,"type":16},"Operations","operations","tag",{"name":18,"slug":19,"type":16},"Management","management",{"name":21,"slug":22,"type":16},"E-commerce","e-commerce",{"name":24,"slug":25,"type":16},"Agents","agents",1735,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fcommerce-agents","2026-09-04T08:00:35.464555",null,277,[],{"repoUrl":27,"stars":26,"forks":30,"topics":33,"description":34},[],"Reference blueprint for building shopping and merchant agents with Claude. Examples in retail, commerce, telecom, and entertainment included.","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fcommerce-agents\u002Ftree\u002FHEAD\u002Fplugins\u002Fcommerce-builder\u002Fskills\u002Fcommerce-merchant-operations","---\nname: commerce-merchant-operations\ndescription: The reference merchant agent, covering its flows, staged changes and host approval, metrics grounding, the analysis delegate, store memory, components, and marketplace rules. Load when building or reviewing an agent for the operators of a store, property, subscriber base, or venue.\n---\n\n# Merchant agent\n\nPaths are in the reference repo: `merchant_agent\u002F` is `merchant-agent\u002Fcore\u002Fmerchant_agent\u002F`, `merchant_agent_runtime\u002F`\nis `merchant-agent\u002Fruntime-messages-api\u002Fmerchant_agent_runtime\u002F`, `merchant_agent_sdk\u002F` is\n`merchant-agent\u002Fruntime-agent-sdk\u002Fmerchant_agent_sdk\u002F`, `commerce_common\u002F` is `commerce-common\u002Fcommerce_common\u002F`.\n\n## What it does\n\nThe merchant agent works with an operator inside their back office over one `MerchantBackend` (`merchant_agent\u002Fbackend.py`).\nIts five flows are the skills in `merchant-agent\u002Fskills\u002F`: performance-insights, catalog-listings, inventory-operations,\npricing-promotions, marketing-campaigns. The mechanisms in `commerce_common\u002F` are shared; the prompt, tools, gates, and\nexecutor are its own, and every tool result comes back inside `MERCHANT_FENCE` (`merchant_agent\u002Ffencing.py`).\n\n## The staged-change contract\n\n1. Every write is a `stage_*` tool (`stage_listing_update`, `stage_price_update`, `stage_inventory_action`, `stage_promotion`,\n   `stage_campaign`). Each returns the staged record and a note, emits a `change_update` event, and, with\n   `stage_shows_preview` on (the default, `merchant_agent\u002Fconfig.py`), emits the `change_preview` card through the runner\n   `present_change_preview` uses (`MerchantToolExecutor`, `merchant_agent\u002Fexecutor.py`). The MCP server turns it off\n   because its executor events do not reach the operator, so the hosted agent calls `present_change_preview`; that tool\n   also shows an earlier change again. Live state is untouched until `apply_change`, which performs the platform write.\n2. Staging accepts only listing and campaign ids that tools returned this session, and a content edit also needs a\n   `get_listing` read; `apply_change` and `discard_change` accept only change ids that staging or `get_pending_changes`\n   returned (`merchant_agent\u002Fgates.py`). Showing the preview card marks nothing approved.\n3. `check_guardrails` in `merchant_agent\u002Fchanges.py` runs when a change is staged (`ChangeLedger.stage`, or your backend's\n   equivalent) and again in `check_apply_change` (`merchant_agent\u002Fgates.py`) under the config in force at apply time.\n   The limits are `max_items_per_change`, `max_price_delta_pct`, `max_promotion_discount_pct`, `max_restock_quantity`,\n   `max_campaign_budget`, and `max_listing_field_chars` in `merchant_agent\u002Fconfig.py`; the defaults are demonstration values.\n4. `protected_fields` can never be staged, `price_bearing_fields` are the fields the price cap reads, and\n   `listing_update_blocked_fields` may not ride a free-form listing update. A domain that prices under another name\n   (a nightly rate, a monthly fee, a tier price) appends to these tuples and never replaces them.\n5. `require_host_approval` defaults to `True`: `apply_change` succeeds only for an id in `MerchantSessionState.approved_change_ids`,\n   which only host code writes. The three surfaces in the repo are the portal's `\u002Fchanges\u002F{change_id}\u002Fapply` route\n   (`examples\u002Fdemo_common\u002Fmerchant.py`), `MerchantToolset.host_approve` (`merchant_agent_sdk\u002Fmerchant_tools.py`, prompted\n   per change by `merchant-agent\u002Fruntime-agent-sdk\u002Fmain.py`), and `always_ask` on `apply_change` in the hosted manifest;\n   there the MCP server's config sets `require_host_approval=False` (its `default_config` does; a config you pass\n   must too, or every apply is held). An approval typed into the chat sets nothing.\n6. `require_host_approval` and `approval_surface` render into the static prompt and into refusals, so they are set per\n   deployment and a change to either is a redeploy (commerce-prompt-caching). Name the real surface in `approval_surface`.\n7. A change kind a system does not support raises `ChangeNotApplicable` (`merchant_agent\u002Fchanges.py`) and the executor\n   relays it; the tool stays registered. Applied and discarded changes stay queryable as the audit trail, each stamped\n   with the operator from `MerchantSessionContext.operator`, which a production host derives from its authentication.\n\n## Grounding and follow-through\n\n- A performance question forces `get_business_snapshot`, and a change request carrying an apply phrase with nothing\n  staged this session forces `get_pending_changes` (`GROUNDING_RULES` in `merchant_agent\u002Fgrounding.py`; lexicons and\n  flags in `merchant_agent\u002Fconfig.py`, appended to and never replaced).\n- A turn that matched `change_requested` and ended on bare text, with no `stage_*` attempt and no `present_suggestions`\n  close, gets `STAGING_FOLLOWTHROUGH_REMINDER` once, as a user message (`merchant_agent\u002Fgates.py`; applied in\n  `merchant_agent_runtime\u002Forchestrator.py` and `merchant_agent_sdk\u002Fagent.py`); the reminder text is excluded from memory extraction.\n- `present_metrics` joins each pick from the snapshot, a queried series, a campaign, or a recorded analysis and drops the\n  rest with a note (`resolve_metrics` in `merchant_agent\u002Fenrichment.py`), so a card never carries a model-authored figure.\n\n## The analysis delegate\n\n- `enable_analysis` (default off) registers `run_analysis`, a `DelegateExtension` built by `build_analysis_delegate` in\n  `merchant_agent_runtime\u002Fanalysis.py`. The delegate's tools are a submit tool, a progress tool, the `ANALYSIS_READ_TOOLS`\n  (`merchant_agent\u002Fanalysis.py`), and a query tool when the backend implements `MerchantBackend.execute_analysis_query`\n  (`analysis_sql_only` then leaves the per-series reads off). It holds no staging tool; its result is recorded in\n  `seen_analyses` and rendered as its own metrics card.\n- `check_analysis_sql` refuses anything other than one SELECT without comments before the backend runs it; the backend owns\n  the read-only role and merchant scoping; `analysis_query_timeout_s` and `cap_analysis_table` (`max_analysis_rows`,\n  `max_analysis_table_chars`) bound each query, and `analysis_timeout_s`, `max_analysis_iterations`, and\n  `max_delegate_calls_per_turn` bound the run.\n- `analysis_use_code_execution` mounts the hosted sandbox and works on the Anthropic API only (including a Foundry deployment hosted on Anthropic); the query method works\n  everywhere (`docs\u002Fdeployment.md`). On the Agent SDK the same contract is a subagent whose tools are exactly the read\n  tools (`build_analysis_agent` in `merchant_agent_sdk\u002Fagent.py`).\n- A scheduled digest is one headless turn of the same agent (`merchant-agent\u002Fmanaged-agents\u002Fscheduled-digest\u002Frun_morning_digest.py`).\n\n## Memory and components\n\n- Memory is keyed by `merchant_id` (`memory_subject` in `merchant_agent\u002Fexecutor.py`); the extraction prompt in\n  `merchant_agent\u002Fmemory.py` admits what the operator stated about running the operation and excludes anything from\n  listings, reviews, buyer messages, or metrics and anything about an identifiable customer. The write filter,\n  retention, delete, purge, and `enable_memory` are the shared rules (commerce-trust-safety).\n- The built-in components (`merchant_agent\u002Fenrichment.py`) are `present_metrics`, `present_digest`, `present_change_preview`,\n  and `present_suggestions`; under host approval the prompt says no chip approves or applies. A vertical adds its own\n  (`present_occupancy_calendar`, `present_plan_mix`, `present_event_pacing` in the examples) as extensions (commerce-ui-tools).\n\n## Marketplaces\n\n- Many sellers' content goes through the one fence; there is no per-seller label and no seller whose text escapes it.\n  In a shopping deployment the seller is a search dimension (`SearchFilters.attributes`) that the components show, and\n  `search_policies` results state whose terms they are when platform and seller terms differ.\n- Provenance and caps are per session, so a session can stage against the listings its own tools returned and no others.\n- Buyer and seller messages reach the model as fenced material; a message that says a change is approved sets no mark.\n- The memory subject is the operator's own business, keyed by its `merchant_id`; a shopper's facts are keyed by the shopper.\n",{"data":38,"body":39},{"name":4,"description":6},{"type":40,"children":41},"root",[42,51,119,126,177,183,624,630,759,765,957,963,1069,1075],{"type":43,"tag":44,"props":45,"children":47},"element","h1",{"id":46},"merchant-agent",[48],{"type":49,"value":50},"text","Merchant agent",{"type":43,"tag":52,"props":53,"children":54},"p",{},[55,57,64,66,72,74,80,82,88,89,95,97,103,104,110,111,117],{"type":49,"value":56},"Paths are in the reference repo: ",{"type":43,"tag":58,"props":59,"children":61},"code",{"className":60},[],[62],{"type":49,"value":63},"merchant_agent\u002F",{"type":49,"value":65}," is ",{"type":43,"tag":58,"props":67,"children":69},{"className":68},[],[70],{"type":49,"value":71},"merchant-agent\u002Fcore\u002Fmerchant_agent\u002F",{"type":49,"value":73},", ",{"type":43,"tag":58,"props":75,"children":77},{"className":76},[],[78],{"type":49,"value":79},"merchant_agent_runtime\u002F",{"type":49,"value":81},"\nis ",{"type":43,"tag":58,"props":83,"children":85},{"className":84},[],[86],{"type":49,"value":87},"merchant-agent\u002Fruntime-messages-api\u002Fmerchant_agent_runtime\u002F",{"type":49,"value":73},{"type":43,"tag":58,"props":90,"children":92},{"className":91},[],[93],{"type":49,"value":94},"merchant_agent_sdk\u002F",{"type":49,"value":96}," is\n",{"type":43,"tag":58,"props":98,"children":100},{"className":99},[],[101],{"type":49,"value":102},"merchant-agent\u002Fruntime-agent-sdk\u002Fmerchant_agent_sdk\u002F",{"type":49,"value":73},{"type":43,"tag":58,"props":105,"children":107},{"className":106},[],[108],{"type":49,"value":109},"commerce_common\u002F",{"type":49,"value":65},{"type":43,"tag":58,"props":112,"children":114},{"className":113},[],[115],{"type":49,"value":116},"commerce-common\u002Fcommerce_common\u002F",{"type":49,"value":118},".",{"type":43,"tag":120,"props":121,"children":123},"h2",{"id":122},"what-it-does",[124],{"type":49,"value":125},"What it does",{"type":43,"tag":52,"props":127,"children":128},{},[129,131,137,139,145,147,153,155,160,162,168,169,175],{"type":49,"value":130},"The merchant agent works with an operator inside their back office over one ",{"type":43,"tag":58,"props":132,"children":134},{"className":133},[],[135],{"type":49,"value":136},"MerchantBackend",{"type":49,"value":138}," (",{"type":43,"tag":58,"props":140,"children":142},{"className":141},[],[143],{"type":49,"value":144},"merchant_agent\u002Fbackend.py",{"type":49,"value":146},").\nIts five flows are the skills in ",{"type":43,"tag":58,"props":148,"children":150},{"className":149},[],[151],{"type":49,"value":152},"merchant-agent\u002Fskills\u002F",{"type":49,"value":154},": performance-insights, catalog-listings, inventory-operations,\npricing-promotions, marketing-campaigns. The mechanisms in ",{"type":43,"tag":58,"props":156,"children":158},{"className":157},[],[159],{"type":49,"value":109},{"type":49,"value":161}," are shared; the prompt, tools, gates, and\nexecutor are its own, and every tool result comes back inside ",{"type":43,"tag":58,"props":163,"children":165},{"className":164},[],[166],{"type":49,"value":167},"MERCHANT_FENCE",{"type":49,"value":138},{"type":43,"tag":58,"props":170,"children":172},{"className":171},[],[173],{"type":49,"value":174},"merchant_agent\u002Ffencing.py",{"type":49,"value":176},").",{"type":43,"tag":120,"props":178,"children":180},{"id":179},"the-staged-change-contract",[181],{"type":49,"value":182},"The staged-change contract",{"type":43,"tag":184,"props":185,"children":186},"ol",{},[187,308,352,443,470,574,597],{"type":43,"tag":188,"props":189,"children":190},"li",{},[191,193,199,201,207,208,214,215,221,222,228,230,236,238,244,246,252,254,260,262,268,270,276,278,284,285,291,293,298,300,306],{"type":49,"value":192},"Every write is a ",{"type":43,"tag":58,"props":194,"children":196},{"className":195},[],[197],{"type":49,"value":198},"stage_*",{"type":49,"value":200}," tool (",{"type":43,"tag":58,"props":202,"children":204},{"className":203},[],[205],{"type":49,"value":206},"stage_listing_update",{"type":49,"value":73},{"type":43,"tag":58,"props":209,"children":211},{"className":210},[],[212],{"type":49,"value":213},"stage_price_update",{"type":49,"value":73},{"type":43,"tag":58,"props":216,"children":218},{"className":217},[],[219],{"type":49,"value":220},"stage_inventory_action",{"type":49,"value":73},{"type":43,"tag":58,"props":223,"children":225},{"className":224},[],[226],{"type":49,"value":227},"stage_promotion",{"type":49,"value":229},",\n",{"type":43,"tag":58,"props":231,"children":233},{"className":232},[],[234],{"type":49,"value":235},"stage_campaign",{"type":49,"value":237},"). Each returns the staged record and a note, emits a ",{"type":43,"tag":58,"props":239,"children":241},{"className":240},[],[242],{"type":49,"value":243},"change_update",{"type":49,"value":245}," event, and, with\n",{"type":43,"tag":58,"props":247,"children":249},{"className":248},[],[250],{"type":49,"value":251},"stage_shows_preview",{"type":49,"value":253}," on (the default, ",{"type":43,"tag":58,"props":255,"children":257},{"className":256},[],[258],{"type":49,"value":259},"merchant_agent\u002Fconfig.py",{"type":49,"value":261},"), emits the ",{"type":43,"tag":58,"props":263,"children":265},{"className":264},[],[266],{"type":49,"value":267},"change_preview",{"type":49,"value":269}," card through the runner\n",{"type":43,"tag":58,"props":271,"children":273},{"className":272},[],[274],{"type":49,"value":275},"present_change_preview",{"type":49,"value":277}," uses (",{"type":43,"tag":58,"props":279,"children":281},{"className":280},[],[282],{"type":49,"value":283},"MerchantToolExecutor",{"type":49,"value":73},{"type":43,"tag":58,"props":286,"children":288},{"className":287},[],[289],{"type":49,"value":290},"merchant_agent\u002Fexecutor.py",{"type":49,"value":292},"). The MCP server turns it off\nbecause its executor events do not reach the operator, so the hosted agent calls ",{"type":43,"tag":58,"props":294,"children":296},{"className":295},[],[297],{"type":49,"value":275},{"type":49,"value":299},"; that tool\nalso shows an earlier change again. Live state is untouched until ",{"type":43,"tag":58,"props":301,"children":303},{"className":302},[],[304],{"type":49,"value":305},"apply_change",{"type":49,"value":307},", which performs the platform write.",{"type":43,"tag":188,"props":309,"children":310},{},[311,313,319,321,326,328,334,336,342,344,350],{"type":49,"value":312},"Staging accepts only listing and campaign ids that tools returned this session, and a content edit also needs a\n",{"type":43,"tag":58,"props":314,"children":316},{"className":315},[],[317],{"type":49,"value":318},"get_listing",{"type":49,"value":320}," read; ",{"type":43,"tag":58,"props":322,"children":324},{"className":323},[],[325],{"type":49,"value":305},{"type":49,"value":327}," and ",{"type":43,"tag":58,"props":329,"children":331},{"className":330},[],[332],{"type":49,"value":333},"discard_change",{"type":49,"value":335}," accept only change ids that staging or ",{"type":43,"tag":58,"props":337,"children":339},{"className":338},[],[340],{"type":49,"value":341},"get_pending_changes",{"type":49,"value":343},"\nreturned (",{"type":43,"tag":58,"props":345,"children":347},{"className":346},[],[348],{"type":49,"value":349},"merchant_agent\u002Fgates.py",{"type":49,"value":351},"). Showing the preview card marks nothing approved.",{"type":43,"tag":188,"props":353,"children":354},{},[355,361,363,369,371,377,379,385,386,391,393,399,400,406,407,413,414,420,421,427,429,435,436,441],{"type":43,"tag":58,"props":356,"children":358},{"className":357},[],[359],{"type":49,"value":360},"check_guardrails",{"type":49,"value":362}," in ",{"type":43,"tag":58,"props":364,"children":366},{"className":365},[],[367],{"type":49,"value":368},"merchant_agent\u002Fchanges.py",{"type":49,"value":370}," runs when a change is staged (",{"type":43,"tag":58,"props":372,"children":374},{"className":373},[],[375],{"type":49,"value":376},"ChangeLedger.stage",{"type":49,"value":378},", or your backend's\nequivalent) and again in ",{"type":43,"tag":58,"props":380,"children":382},{"className":381},[],[383],{"type":49,"value":384},"check_apply_change",{"type":49,"value":138},{"type":43,"tag":58,"props":387,"children":389},{"className":388},[],[390],{"type":49,"value":349},{"type":49,"value":392},") under the config in force at apply time.\nThe limits are ",{"type":43,"tag":58,"props":394,"children":396},{"className":395},[],[397],{"type":49,"value":398},"max_items_per_change",{"type":49,"value":73},{"type":43,"tag":58,"props":401,"children":403},{"className":402},[],[404],{"type":49,"value":405},"max_price_delta_pct",{"type":49,"value":73},{"type":43,"tag":58,"props":408,"children":410},{"className":409},[],[411],{"type":49,"value":412},"max_promotion_discount_pct",{"type":49,"value":73},{"type":43,"tag":58,"props":415,"children":417},{"className":416},[],[418],{"type":49,"value":419},"max_restock_quantity",{"type":49,"value":229},{"type":43,"tag":58,"props":422,"children":424},{"className":423},[],[425],{"type":49,"value":426},"max_campaign_budget",{"type":49,"value":428},", and ",{"type":43,"tag":58,"props":430,"children":432},{"className":431},[],[433],{"type":49,"value":434},"max_listing_field_chars",{"type":49,"value":362},{"type":43,"tag":58,"props":437,"children":439},{"className":438},[],[440],{"type":49,"value":259},{"type":49,"value":442},"; the defaults are demonstration values.",{"type":43,"tag":188,"props":444,"children":445},{},[446,452,454,460,462,468],{"type":43,"tag":58,"props":447,"children":449},{"className":448},[],[450],{"type":49,"value":451},"protected_fields",{"type":49,"value":453}," can never be staged, ",{"type":43,"tag":58,"props":455,"children":457},{"className":456},[],[458],{"type":49,"value":459},"price_bearing_fields",{"type":49,"value":461}," are the fields the price cap reads, and\n",{"type":43,"tag":58,"props":463,"children":465},{"className":464},[],[466],{"type":49,"value":467},"listing_update_blocked_fields",{"type":49,"value":469}," may not ride a free-form listing update. A domain that prices under another name\n(a nightly rate, a monthly fee, a tier price) appends to these tuples and never replaces them.",{"type":43,"tag":188,"props":471,"children":472},{},[473,479,481,487,489,494,496,502,504,510,512,518,520,526,527,533,535,541,543,549,551,556,558,564,566,572],{"type":43,"tag":58,"props":474,"children":476},{"className":475},[],[477],{"type":49,"value":478},"require_host_approval",{"type":49,"value":480}," defaults to ",{"type":43,"tag":58,"props":482,"children":484},{"className":483},[],[485],{"type":49,"value":486},"True",{"type":49,"value":488},": ",{"type":43,"tag":58,"props":490,"children":492},{"className":491},[],[493],{"type":49,"value":305},{"type":49,"value":495}," succeeds only for an id in ",{"type":43,"tag":58,"props":497,"children":499},{"className":498},[],[500],{"type":49,"value":501},"MerchantSessionState.approved_change_ids",{"type":49,"value":503},",\nwhich only host code writes. The three surfaces in the repo are the portal's ",{"type":43,"tag":58,"props":505,"children":507},{"className":506},[],[508],{"type":49,"value":509},"\u002Fchanges\u002F{change_id}\u002Fapply",{"type":49,"value":511}," route\n(",{"type":43,"tag":58,"props":513,"children":515},{"className":514},[],[516],{"type":49,"value":517},"examples\u002Fdemo_common\u002Fmerchant.py",{"type":49,"value":519},"), ",{"type":43,"tag":58,"props":521,"children":523},{"className":522},[],[524],{"type":49,"value":525},"MerchantToolset.host_approve",{"type":49,"value":138},{"type":43,"tag":58,"props":528,"children":530},{"className":529},[],[531],{"type":49,"value":532},"merchant_agent_sdk\u002Fmerchant_tools.py",{"type":49,"value":534},", prompted\nper change by ",{"type":43,"tag":58,"props":536,"children":538},{"className":537},[],[539],{"type":49,"value":540},"merchant-agent\u002Fruntime-agent-sdk\u002Fmain.py",{"type":49,"value":542},"), and ",{"type":43,"tag":58,"props":544,"children":546},{"className":545},[],[547],{"type":49,"value":548},"always_ask",{"type":49,"value":550}," on ",{"type":43,"tag":58,"props":552,"children":554},{"className":553},[],[555],{"type":49,"value":305},{"type":49,"value":557}," in the hosted manifest;\nthere the MCP server's config sets ",{"type":43,"tag":58,"props":559,"children":561},{"className":560},[],[562],{"type":49,"value":563},"require_host_approval=False",{"type":49,"value":565}," (its ",{"type":43,"tag":58,"props":567,"children":569},{"className":568},[],[570],{"type":49,"value":571},"default_config",{"type":49,"value":573}," does; a config you pass\nmust too, or every apply is held). An approval typed into the chat sets nothing.",{"type":43,"tag":188,"props":575,"children":576},{},[577,582,583,589,591,596],{"type":43,"tag":58,"props":578,"children":580},{"className":579},[],[581],{"type":49,"value":478},{"type":49,"value":327},{"type":43,"tag":58,"props":584,"children":586},{"className":585},[],[587],{"type":49,"value":588},"approval_surface",{"type":49,"value":590}," render into the static prompt and into refusals, so they are set per\ndeployment and a change to either is a redeploy (commerce-prompt-caching). Name the real surface in ",{"type":43,"tag":58,"props":592,"children":594},{"className":593},[],[595],{"type":49,"value":588},{"type":49,"value":118},{"type":43,"tag":188,"props":598,"children":599},{},[600,602,608,609,614,616,622],{"type":49,"value":601},"A change kind a system does not support raises ",{"type":43,"tag":58,"props":603,"children":605},{"className":604},[],[606],{"type":49,"value":607},"ChangeNotApplicable",{"type":49,"value":138},{"type":43,"tag":58,"props":610,"children":612},{"className":611},[],[613],{"type":49,"value":368},{"type":49,"value":615},") and the executor\nrelays it; the tool stays registered. Applied and discarded changes stay queryable as the audit trail, each stamped\nwith the operator from ",{"type":43,"tag":58,"props":617,"children":619},{"className":618},[],[620],{"type":49,"value":621},"MerchantSessionContext.operator",{"type":49,"value":623},", which a production host derives from its authentication.",{"type":43,"tag":120,"props":625,"children":627},{"id":626},"grounding-and-follow-through",[628],{"type":49,"value":629},"Grounding and follow-through",{"type":43,"tag":631,"props":632,"children":633},"ul",{},[634,675,733],{"type":43,"tag":188,"props":635,"children":636},{},[637,639,645,647,652,653,659,660,666,668,673],{"type":49,"value":638},"A performance question forces ",{"type":43,"tag":58,"props":640,"children":642},{"className":641},[],[643],{"type":49,"value":644},"get_business_snapshot",{"type":49,"value":646},", and a change request carrying an apply phrase with nothing\nstaged this session forces ",{"type":43,"tag":58,"props":648,"children":650},{"className":649},[],[651],{"type":49,"value":341},{"type":49,"value":138},{"type":43,"tag":58,"props":654,"children":656},{"className":655},[],[657],{"type":49,"value":658},"GROUNDING_RULES",{"type":49,"value":362},{"type":43,"tag":58,"props":661,"children":663},{"className":662},[],[664],{"type":49,"value":665},"merchant_agent\u002Fgrounding.py",{"type":49,"value":667},"; lexicons and\nflags in ",{"type":43,"tag":58,"props":669,"children":671},{"className":670},[],[672],{"type":49,"value":259},{"type":49,"value":674},", appended to and never replaced).",{"type":43,"tag":188,"props":676,"children":677},{},[678,680,686,688,693,695,701,703,709,711,716,718,724,725,731],{"type":49,"value":679},"A turn that matched ",{"type":43,"tag":58,"props":681,"children":683},{"className":682},[],[684],{"type":49,"value":685},"change_requested",{"type":49,"value":687}," and ended on bare text, with no ",{"type":43,"tag":58,"props":689,"children":691},{"className":690},[],[692],{"type":49,"value":198},{"type":49,"value":694}," attempt and no ",{"type":43,"tag":58,"props":696,"children":698},{"className":697},[],[699],{"type":49,"value":700},"present_suggestions",{"type":49,"value":702},"\nclose, gets ",{"type":43,"tag":58,"props":704,"children":706},{"className":705},[],[707],{"type":49,"value":708},"STAGING_FOLLOWTHROUGH_REMINDER",{"type":49,"value":710}," once, as a user message (",{"type":43,"tag":58,"props":712,"children":714},{"className":713},[],[715],{"type":49,"value":349},{"type":49,"value":717},"; applied in\n",{"type":43,"tag":58,"props":719,"children":721},{"className":720},[],[722],{"type":49,"value":723},"merchant_agent_runtime\u002Forchestrator.py",{"type":49,"value":327},{"type":43,"tag":58,"props":726,"children":728},{"className":727},[],[729],{"type":49,"value":730},"merchant_agent_sdk\u002Fagent.py",{"type":49,"value":732},"); the reminder text is excluded from memory extraction.",{"type":43,"tag":188,"props":734,"children":735},{},[736,742,744,750,751,757],{"type":43,"tag":58,"props":737,"children":739},{"className":738},[],[740],{"type":49,"value":741},"present_metrics",{"type":49,"value":743}," joins each pick from the snapshot, a queried series, a campaign, or a recorded analysis and drops the\nrest with a note (",{"type":43,"tag":58,"props":745,"children":747},{"className":746},[],[748],{"type":49,"value":749},"resolve_metrics",{"type":49,"value":362},{"type":43,"tag":58,"props":752,"children":754},{"className":753},[],[755],{"type":49,"value":756},"merchant_agent\u002Fenrichment.py",{"type":49,"value":758},"), so a card never carries a model-authored figure.",{"type":43,"tag":120,"props":760,"children":762},{"id":761},"the-analysis-delegate",[763],{"type":49,"value":764},"The analysis delegate",{"type":43,"tag":631,"props":766,"children":767},{},[768,850,913,945],{"type":43,"tag":188,"props":769,"children":770},{},[771,777,779,785,787,793,795,801,803,809,811,817,819,825,827,833,834,840,842,848],{"type":43,"tag":58,"props":772,"children":774},{"className":773},[],[775],{"type":49,"value":776},"enable_analysis",{"type":49,"value":778}," (default off) registers ",{"type":43,"tag":58,"props":780,"children":782},{"className":781},[],[783],{"type":49,"value":784},"run_analysis",{"type":49,"value":786},", a ",{"type":43,"tag":58,"props":788,"children":790},{"className":789},[],[791],{"type":49,"value":792},"DelegateExtension",{"type":49,"value":794}," built by ",{"type":43,"tag":58,"props":796,"children":798},{"className":797},[],[799],{"type":49,"value":800},"build_analysis_delegate",{"type":49,"value":802}," in\n",{"type":43,"tag":58,"props":804,"children":806},{"className":805},[],[807],{"type":49,"value":808},"merchant_agent_runtime\u002Fanalysis.py",{"type":49,"value":810},". The delegate's tools are a submit tool, a progress tool, the ",{"type":43,"tag":58,"props":812,"children":814},{"className":813},[],[815],{"type":49,"value":816},"ANALYSIS_READ_TOOLS",{"type":49,"value":818},"\n(",{"type":43,"tag":58,"props":820,"children":822},{"className":821},[],[823],{"type":49,"value":824},"merchant_agent\u002Fanalysis.py",{"type":49,"value":826},"), and a query tool when the backend implements ",{"type":43,"tag":58,"props":828,"children":830},{"className":829},[],[831],{"type":49,"value":832},"MerchantBackend.execute_analysis_query",{"type":49,"value":818},{"type":43,"tag":58,"props":835,"children":837},{"className":836},[],[838],{"type":49,"value":839},"analysis_sql_only",{"type":49,"value":841}," then leaves the per-series reads off). It holds no staging tool; its result is recorded in\n",{"type":43,"tag":58,"props":843,"children":845},{"className":844},[],[846],{"type":49,"value":847},"seen_analyses",{"type":49,"value":849}," and rendered as its own metrics card.",{"type":43,"tag":188,"props":851,"children":852},{},[853,859,861,867,868,874,875,881,882,888,890,896,897,903,905,911],{"type":43,"tag":58,"props":854,"children":856},{"className":855},[],[857],{"type":49,"value":858},"check_analysis_sql",{"type":49,"value":860}," refuses anything other than one SELECT without comments before the backend runs it; the backend owns\nthe read-only role and merchant scoping; ",{"type":43,"tag":58,"props":862,"children":864},{"className":863},[],[865],{"type":49,"value":866},"analysis_query_timeout_s",{"type":49,"value":327},{"type":43,"tag":58,"props":869,"children":871},{"className":870},[],[872],{"type":49,"value":873},"cap_analysis_table",{"type":49,"value":138},{"type":43,"tag":58,"props":876,"children":878},{"className":877},[],[879],{"type":49,"value":880},"max_analysis_rows",{"type":49,"value":229},{"type":43,"tag":58,"props":883,"children":885},{"className":884},[],[886],{"type":49,"value":887},"max_analysis_table_chars",{"type":49,"value":889},") bound each query, and ",{"type":43,"tag":58,"props":891,"children":893},{"className":892},[],[894],{"type":49,"value":895},"analysis_timeout_s",{"type":49,"value":73},{"type":43,"tag":58,"props":898,"children":900},{"className":899},[],[901],{"type":49,"value":902},"max_analysis_iterations",{"type":49,"value":904},", and\n",{"type":43,"tag":58,"props":906,"children":908},{"className":907},[],[909],{"type":49,"value":910},"max_delegate_calls_per_turn",{"type":49,"value":912}," bound the run.",{"type":43,"tag":188,"props":914,"children":915},{},[916,922,924,930,932,938,939,944],{"type":43,"tag":58,"props":917,"children":919},{"className":918},[],[920],{"type":49,"value":921},"analysis_use_code_execution",{"type":49,"value":923}," mounts the hosted sandbox and works on the Anthropic API only (including a Foundry deployment hosted on Anthropic); the query method works\neverywhere (",{"type":43,"tag":58,"props":925,"children":927},{"className":926},[],[928],{"type":49,"value":929},"docs\u002Fdeployment.md",{"type":49,"value":931},"). On the Agent SDK the same contract is a subagent whose tools are exactly the read\ntools (",{"type":43,"tag":58,"props":933,"children":935},{"className":934},[],[936],{"type":49,"value":937},"build_analysis_agent",{"type":49,"value":362},{"type":43,"tag":58,"props":940,"children":942},{"className":941},[],[943],{"type":49,"value":730},{"type":49,"value":176},{"type":43,"tag":188,"props":946,"children":947},{},[948,950,956],{"type":49,"value":949},"A scheduled digest is one headless turn of the same agent (",{"type":43,"tag":58,"props":951,"children":953},{"className":952},[],[954],{"type":49,"value":955},"merchant-agent\u002Fmanaged-agents\u002Fscheduled-digest\u002Frun_morning_digest.py",{"type":49,"value":176},{"type":43,"tag":120,"props":958,"children":960},{"id":959},"memory-and-components",[961],{"type":49,"value":962},"Memory and components",{"type":43,"tag":631,"props":964,"children":965},{},[966,1008],{"type":43,"tag":188,"props":967,"children":968},{},[969,971,977,978,984,985,990,992,998,1000,1006],{"type":49,"value":970},"Memory is keyed by ",{"type":43,"tag":58,"props":972,"children":974},{"className":973},[],[975],{"type":49,"value":976},"merchant_id",{"type":49,"value":138},{"type":43,"tag":58,"props":979,"children":981},{"className":980},[],[982],{"type":49,"value":983},"memory_subject",{"type":49,"value":362},{"type":43,"tag":58,"props":986,"children":988},{"className":987},[],[989],{"type":49,"value":290},{"type":49,"value":991},"); the extraction prompt in\n",{"type":43,"tag":58,"props":993,"children":995},{"className":994},[],[996],{"type":49,"value":997},"merchant_agent\u002Fmemory.py",{"type":49,"value":999}," admits what the operator stated about running the operation and excludes anything from\nlistings, reviews, buyer messages, or metrics and anything about an identifiable customer. The write filter,\nretention, delete, purge, and ",{"type":43,"tag":58,"props":1001,"children":1003},{"className":1002},[],[1004],{"type":49,"value":1005},"enable_memory",{"type":49,"value":1007}," are the shared rules (commerce-trust-safety).",{"type":43,"tag":188,"props":1009,"children":1010},{},[1011,1013,1018,1020,1025,1026,1032,1033,1038,1040,1045,1047,1053,1054,1060,1061,1067],{"type":49,"value":1012},"The built-in components (",{"type":43,"tag":58,"props":1014,"children":1016},{"className":1015},[],[1017],{"type":49,"value":756},{"type":49,"value":1019},") are ",{"type":43,"tag":58,"props":1021,"children":1023},{"className":1022},[],[1024],{"type":49,"value":741},{"type":49,"value":73},{"type":43,"tag":58,"props":1027,"children":1029},{"className":1028},[],[1030],{"type":49,"value":1031},"present_digest",{"type":49,"value":73},{"type":43,"tag":58,"props":1034,"children":1036},{"className":1035},[],[1037],{"type":49,"value":275},{"type":49,"value":1039},",\nand ",{"type":43,"tag":58,"props":1041,"children":1043},{"className":1042},[],[1044],{"type":49,"value":700},{"type":49,"value":1046},"; under host approval the prompt says no chip approves or applies. A vertical adds its own\n(",{"type":43,"tag":58,"props":1048,"children":1050},{"className":1049},[],[1051],{"type":49,"value":1052},"present_occupancy_calendar",{"type":49,"value":73},{"type":43,"tag":58,"props":1055,"children":1057},{"className":1056},[],[1058],{"type":49,"value":1059},"present_plan_mix",{"type":49,"value":73},{"type":43,"tag":58,"props":1062,"children":1064},{"className":1063},[],[1065],{"type":49,"value":1066},"present_event_pacing",{"type":49,"value":1068}," in the examples) as extensions (commerce-ui-tools).",{"type":43,"tag":120,"props":1070,"children":1072},{"id":1071},"marketplaces",[1073],{"type":49,"value":1074},"Marketplaces",{"type":43,"tag":631,"props":1076,"children":1077},{},[1078,1099,1104,1109],{"type":43,"tag":188,"props":1079,"children":1080},{},[1081,1083,1089,1091,1097],{"type":49,"value":1082},"Many sellers' content goes through the one fence; there is no per-seller label and no seller whose text escapes it.\nIn a shopping deployment the seller is a search dimension (",{"type":43,"tag":58,"props":1084,"children":1086},{"className":1085},[],[1087],{"type":49,"value":1088},"SearchFilters.attributes",{"type":49,"value":1090},") that the components show, and\n",{"type":43,"tag":58,"props":1092,"children":1094},{"className":1093},[],[1095],{"type":49,"value":1096},"search_policies",{"type":49,"value":1098}," results state whose terms they are when platform and seller terms differ.",{"type":43,"tag":188,"props":1100,"children":1101},{},[1102],{"type":49,"value":1103},"Provenance and caps are per session, so a session can stage against the listings its own tools returned and no others.",{"type":43,"tag":188,"props":1105,"children":1106},{},[1107],{"type":49,"value":1108},"Buyer and seller messages reach the model as fenced material; a message that says a change is approved sets no mark.",{"type":43,"tag":188,"props":1110,"children":1111},{},[1112,1114,1119],{"type":49,"value":1113},"The memory subject is the operator's own business, keyed by its ",{"type":43,"tag":58,"props":1115,"children":1117},{"className":1116},[],[1118],{"type":49,"value":976},{"type":49,"value":1120},"; a shopper's facts are keyed by the shopper.",{"items":1122,"total":1306},[1123,1139,1158,1172,1184,1201,1217,1228,1249,1269,1283,1298],{"slug":1124,"name":1124,"fn":1125,"description":1126,"org":1127,"tags":1128,"stars":1136,"repoUrl":1137,"updatedAt":1138},"academy-guide","recommend Claude Academy resources","Stop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: \"how do I\", \"how can I\", \"getting started with\", \"what can Claude do\", \"teach me\", \"learn to use\"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1129,1130,1133],{"name":9,"slug":8,"type":16},{"name":1131,"slug":1132,"type":16},"Documentation","documentation",{"name":1134,"slug":1135,"type":16},"Education","education",161831,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fskills","2026-08-19T03:59:01.021254",{"slug":1140,"name":1140,"fn":1141,"description":1142,"org":1143,"tags":1144,"stars":1136,"repoUrl":1137,"updatedAt":1157},"algorithmic-art","create algorithmic art with p5.js","Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1145,1148,1151,1154],{"name":1146,"slug":1147,"type":16},"Creative","creative",{"name":1149,"slug":1150,"type":16},"Design","design",{"name":1152,"slug":1153,"type":16},"Generative Art","generative-art",{"name":1155,"slug":1156,"type":16},"JavaScript","javascript","2026-04-06T17:56:15.455818",{"slug":1159,"name":1159,"fn":1160,"description":1161,"org":1162,"tags":1163,"stars":1136,"repoUrl":1137,"updatedAt":1171},"brand-guidelines","apply Anthropic brand colors and typography","Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1164,1167,1168],{"name":1165,"slug":1166,"type":16},"Branding","branding",{"name":1149,"slug":1150,"type":16},{"name":1169,"slug":1170,"type":16},"Typography","typography","2026-04-06T17:56:05.042852",{"slug":1173,"name":1173,"fn":1174,"description":1175,"org":1176,"tags":1177,"stars":1136,"repoUrl":1137,"updatedAt":1183},"canvas-design","create posters and visual art as PNG or PDF","Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1178,1179,1180],{"name":1146,"slug":1147,"type":16},{"name":1149,"slug":1150,"type":16},{"name":1181,"slug":1182,"type":16},"PDF","pdf","2026-04-06T17:56:03.794732",{"slug":1185,"name":1185,"fn":1186,"description":1187,"org":1188,"tags":1189,"stars":1136,"repoUrl":1137,"updatedAt":1200},"claude-api","build apps with the Claude API","Reference for the Claude API \u002F Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration.\nTRIGGER — read BEFORE opening the target file; don't skip because it \"looks like a one-liner\" — whenever: the prompt names Claude\u002FAnthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing\u002Fmodel choice\u002Flimits\u002Fcaching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent\u002FMCP\u002Ftool-definition\u002Fmulti-agent\u002FRAG\u002FLLM-judge\u002Fcomputer-use; generate\u002Fsummarize\u002Fextract\u002Fclassify\u002Frewrite\u002Fconverse over NL; debugging refusals\u002Fcutoffs\u002Fstreaming\u002Ftool-calls\u002Ftokens).\nSKIP only when another provider is being worked on (overrides all triggers): OpenAI\u002FGPT\u002FGemini\u002FLlama\u002FMistral\u002FCohere\u002FOllama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1190,1191,1192,1195,1197],{"name":24,"slug":25,"type":16},{"name":9,"slug":8,"type":16},{"name":1193,"slug":1194,"type":16},"Anthropic SDK","anthropic-sdk",{"name":1196,"slug":1185,"type":16},"Claude API",{"name":1198,"slug":1199,"type":16},"LLM","llm","2026-09-04T07:28:24.898544",{"slug":1202,"name":1202,"fn":1203,"description":1204,"org":1205,"tags":1206,"stars":1136,"repoUrl":1137,"updatedAt":1216},"discernment-nudge","provide discernment nudges for user decisions","After you give a substantive answer or draft that the user may act on — advice or recommendations, drafted artifacts such as goals, plans, pitches, proposals, or emails, estimates or projections, analysis or interpretation of data, factual claims they may rely on, or a multi-step argument — invoke this skill BEFORE finalizing your reply and then, if it applies, append 2-3 short follow-up questions, each tied to something specific in what you just produced, that help the user check key facts, probe the reasoning or assumptions, and notice missing context. Do this at most once per conversation. Skip it when the user asked a trivial how-to or simple lookup, wants a purely educational explanation, asked you only to format, convert, or assemble a file from content they provided, is writing code they will run, is doing creative writing or casual chat, or already asked you to double-check, cite, or review — the skill file explains these boundaries and the exact output format.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1207,1210,1213],{"name":1208,"slug":1209,"type":16},"Coaching","coaching",{"name":1211,"slug":1212,"type":16},"Productivity","productivity",{"name":1214,"slug":1215,"type":16},"Strategy","strategy","2026-08-19T03:59:01.539281",{"slug":1218,"name":1218,"fn":1219,"description":1220,"org":1221,"tags":1222,"stars":1136,"repoUrl":1137,"updatedAt":1227},"doc-coauthoring","co-author documentation and technical specs","Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1223,1224],{"name":1131,"slug":1132,"type":16},{"name":1225,"slug":1226,"type":16},"Technical Writing","technical-writing","2026-04-06T17:56:14.18897",{"slug":1229,"name":1229,"fn":1230,"description":1231,"org":1232,"tags":1233,"stars":1136,"repoUrl":1137,"updatedAt":1248},"docx","create and edit Word documents","Use this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files) or Word templates (.dotx files). Triggers include: any mention of 'Word doc', 'word document', '.docx', '.dotx', or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx or .dotx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1234,1237,1239,1242,1245],{"name":1235,"slug":1236,"type":16},"Documents","documents",{"name":1238,"slug":1229,"type":16},"DOCX",{"name":1240,"slug":1241,"type":16},"Office","office",{"name":1243,"slug":1244,"type":16},"Templates","templates",{"name":1246,"slug":1247,"type":16},"Word","word","2026-07-18T05:16:23.136271",{"slug":1250,"name":1250,"fn":1251,"description":1252,"org":1253,"tags":1254,"stars":1136,"repoUrl":1137,"updatedAt":1268},"frontend-design","design production-grade frontend interfaces","Guidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1255,1256,1259,1262,1265],{"name":1149,"slug":1150,"type":16},{"name":1257,"slug":1258,"type":16},"Frontend","frontend",{"name":1260,"slug":1261,"type":16},"React","react",{"name":1263,"slug":1264,"type":16},"Tailwind CSS","tailwind-css",{"name":1266,"slug":1267,"type":16},"UI Components","ui-components","2026-09-04T07:28:23.795756",{"slug":1270,"name":1270,"fn":1271,"description":1272,"org":1273,"tags":1274,"stars":1136,"repoUrl":1137,"updatedAt":1282},"internal-comms","write internal company communications","A set of resources to help me write all kinds of internal communications, using the formats that my company likes to use. Claude should use this skill whenever asked to write some sort of internal communications (status reports, leadership updates, 3P updates, company newsletters, FAQs, incident reports, project updates, etc.).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1275,1278,1279],{"name":1276,"slug":1277,"type":16},"Communications","communications",{"name":1243,"slug":1244,"type":16},{"name":1280,"slug":1281,"type":16},"Writing","writing","2026-04-06T17:56:20.695522",{"slug":1284,"name":1284,"fn":1285,"description":1286,"org":1287,"tags":1288,"stars":1136,"repoUrl":1137,"updatedAt":1297},"mcp-builder","build MCP servers","Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node\u002FTypeScript (MCP SDK).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1289,1290,1293,1294],{"name":24,"slug":25,"type":16},{"name":1291,"slug":1292,"type":16},"API Development","api-development",{"name":1198,"slug":1199,"type":16},{"name":1295,"slug":1296,"type":16},"MCP","mcp","2026-04-06T17:56:10.357665",{"slug":1182,"name":1182,"fn":1299,"description":1300,"org":1301,"tags":1302,"stars":1136,"repoUrl":1137,"updatedAt":1305},"read edit and manipulate PDF files","Use this skill whenever the user wants to do anything with PDF files. This includes reading or extracting text\u002Ftables from PDFs, combining or merging multiple PDFs into one, splitting PDFs apart, rotating pages, adding watermarks, creating new PDFs, filling PDF forms, encrypting\u002Fdecrypting PDFs, extracting images, and OCR on scanned PDFs to make them searchable. If the user mentions a .pdf file or asks to produce one, use this skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1303,1304],{"name":1235,"slug":1236,"type":16},{"name":1181,"slug":1182,"type":16},"2026-04-06T17:56:02.483316",521,{"items":1308,"total":1406},[1309,1326,1341,1356,1363,1378,1395],{"slug":1310,"name":1310,"fn":1311,"description":1312,"org":1313,"tags":1314,"stars":26,"repoUrl":27,"updatedAt":1325},"catalog-listings","create and improve e-commerce listing content","Creating and improving listing content, covering titles, descriptions, attribute completeness, categorization fixes, image callouts written as text, edits written from material the operator supplies, and content-quality audits and bulk fixes across many listings. Not needed for questions about how a listing is performing.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1315,1318,1321,1322],{"name":1316,"slug":1317,"type":16},"Content Creation","content-creation",{"name":1319,"slug":1320,"type":16},"Content Strategy","content-strategy",{"name":21,"slug":22,"type":16},{"name":1323,"slug":1324,"type":16},"Marketing","marketing","2026-09-04T08:00:35.843558",{"slug":1327,"name":1327,"fn":1328,"description":1329,"org":1330,"tags":1331,"stars":26,"repoUrl":27,"updatedAt":1340},"commerce-architecture","design and review commerce agent architecture","How the reference commerce agents of either role are structured, covering the loop, where each rule lives, skills, the backend interface, delegates, and model fields. Load when designing or reviewing the structure of a shopping or merchant agent.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1332,1333,1336,1337],{"name":24,"slug":25,"type":16},{"name":1334,"slug":1335,"type":16},"Architecture","architecture",{"name":21,"slug":22,"type":16},{"name":1338,"slug":1339,"type":16},"Engineering","engineering","2026-09-04T08:00:39.152322",{"slug":1342,"name":1342,"fn":1343,"description":1344,"org":1345,"tags":1346,"stars":26,"repoUrl":27,"updatedAt":1355},"commerce-evals","run behavioral evals for commerce agents","Authoring and running behavioral evals for a shopping or merchant agent, covering the case shape, authoring rules, code graders and judges, the run pattern, and poisoned fixtures. Load when writing eval cases or rubrics or deciding how a suite runs.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1347,1348,1349,1352],{"name":24,"slug":25,"type":16},{"name":21,"slug":22,"type":16},{"name":1350,"slug":1351,"type":16},"Evals","evals",{"name":1353,"slug":1354,"type":16},"Testing","testing","2026-09-04T08:00:36.204114",{"slug":4,"name":4,"fn":5,"description":6,"org":1357,"tags":1358,"stars":26,"repoUrl":27,"updatedAt":28},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1359,1360,1361,1362],{"name":24,"slug":25,"type":16},{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},{"slug":1364,"name":1364,"fn":1365,"description":1366,"org":1367,"tags":1368,"stars":26,"repoUrl":27,"updatedAt":1377},"commerce-prompt-caching","configure commerce agent prompt caching","The reference agents' cache-stable request assembly, covering the static system and per-request context split, the fixed tool list, the rolling conversation breakpoint, which config fields are prompt bytes, and verification. Load when writing or reviewing a commerce agent's prompt assembly, when cache reads are zero, or when a turn's latency or cost is the question.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1369,1370,1371,1374],{"name":24,"slug":25,"type":16},{"name":21,"slug":22,"type":16},{"name":1372,"slug":1373,"type":16},"Performance","performance",{"name":1375,"slug":1376,"type":16},"Prompt Engineering","prompt-engineering","2026-09-04T08:00:39.50364",{"slug":1379,"name":1379,"fn":1380,"description":1381,"org":1382,"tags":1383,"stars":26,"repoUrl":27,"updatedAt":1394},"commerce-trust-safety","enforce trust and safety rules","The rules the reference agents enforce in code for third-party content, writes, grounding, identity, and memory, each with its module, plus two adversarial-eval rules. Load when handling untrusted tool results, guarding a write tool, or scoping what an agent remembers.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1384,1385,1388,1391],{"name":24,"slug":25,"type":16},{"name":1386,"slug":1387,"type":16},"Compliance","compliance",{"name":1389,"slug":1390,"type":16},"Memory","memory",{"name":1392,"slug":1393,"type":16},"Security","security","2026-09-04T08:00:42.609563",{"slug":1396,"name":1396,"fn":1397,"description":1398,"org":1399,"tags":1400,"stars":26,"repoUrl":27,"updatedAt":1405},"commerce-ui-tools","implement commerce agent UI components","The reference presentation-tool contract, covering server-side enrichment, suggestion chips, the event stream, progressive rendering, both roles' built-in components, and adding a vertical component. Load when building or reviewing how a commerce agent's output reaches a frontend.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1401,1402,1403,1404],{"name":1149,"slug":1150,"type":16},{"name":21,"slug":22,"type":16},{"name":1257,"slug":1258,"type":16},{"name":1266,"slug":1267,"type":16},"2026-09-04T08:00:38.801247",16]