[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-posthog-diagnosing-experiment-results":3,"mdc-3knmo1-key":37,"related-org-posthog-diagnosing-experiment-results":1123,"related-repo-posthog-diagnosing-experiment-results":1294},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":26,"repoUrl":27,"updatedAt":28,"license":29,"forks":30,"topics":31,"repo":32,"sourceUrl":35,"mdContent":36},"diagnosing-experiment-results","diagnose PostHog experiment results and anomalies","Diagnoses bias, anomalies, and strange-looking results on a specific PostHog experiment. Covers empty \u002F 0-exposure experiments, sample ratio mismatch, identity fragmentation, multi-variant exposure, uneven-split exclusion bias, significance traps (peeking, A\u002FA, Bayesian vs Frequentist), PostHog-vs-SQL discrepancies, and surprises after mid-run edits. Symptom-driven dispatch to the right diagnostic.\nTRIGGER when: user asks 'is my experiment biased?' or 'why 0 exposures?', references the bias banner, says a variant looks strange \u002F wrong \u002F off, sees significance flipping, notices PostHog numbers disagreeing with their SQL, sees an A\u002FA test showing significance, or reports surprises after mid-run edits.\nDO NOT TRIGGER when: creating a new experiment (use creating-experiments), only configuring rollout (use configuring-experiment-rollout) or metrics (use configuring-experiment-analytics), or only asking lifecycle questions (use managing-experiment-lifecycle).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"posthog","PostHog","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fposthog.png",[12,14,17,20,23],{"name":9,"slug":8,"type":13},"tag",{"name":15,"slug":16,"type":13},"Data Analysis","data-analysis",{"name":18,"slug":19,"type":13},"Analytics","analytics",{"name":21,"slug":22,"type":13},"A\u002FB Testing","a-b-testing",{"name":24,"slug":25,"type":13},"Debugging","debugging",56,"https:\u002F\u002Fgithub.com\u002FPostHog\u002Fskills","2026-05-22T06:59:58.103867",null,4,[],{"repoUrl":27,"stars":26,"forks":30,"topics":33,"description":34},[],"PostHog skills (under construction)","https:\u002F\u002Fgithub.com\u002FPostHog\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Fomnibus\u002Fdiagnosing-experiment-results","---\nname: diagnosing-experiment-results\ndescription: \"Diagnoses bias, anomalies, and strange-looking results on a specific PostHog experiment. Covers empty \u002F 0-exposure experiments, sample ratio mismatch, identity fragmentation, multi-variant exposure, uneven-split exclusion bias, significance traps (peeking, A\u002FA, Bayesian vs Frequentist), PostHog-vs-SQL discrepancies, and surprises after mid-run edits. Symptom-driven dispatch to the right diagnostic.\\nTRIGGER when: user asks 'is my experiment biased?' or 'why 0 exposures?', references the bias banner, says a variant looks strange \u002F wrong \u002F off, sees significance flipping, notices PostHog numbers disagreeing with their SQL, sees an A\u002FA test showing significance, or reports surprises after mid-run edits.\\nDO NOT TRIGGER when: creating a new experiment (use creating-experiments), only configuring rollout (use configuring-experiment-rollout) or metrics (use configuring-experiment-analytics), or only asking lifecycle questions (use managing-experiment-lifecycle).\"\n---\n\n# Diagnosing experiment results\n\nThis skill answers: **My PostHog experiment results look wrong, biased, or empty — what's going on?**\n\nMatch the user's complaint in the dispatch table, then read the matching reference file for the\ndiagnostic.\n\nEach diagnostic in the reference files is tagged `[HIGH]`, `[MEDIUM]`, or `[LOW]` based on how\nstrongly it's verified — `[HIGH]` is verified directly in PostHog code, `[MEDIUM]` is partially or\nteam-source verified, `[LOW]` describes SDK\u002Fexternal behavior that wasn't verified here. Treat `[LOW]`\nitems as hypotheses to test, not facts to assert.\n\n## Step 1 — Resolve the experiment\n\nIf the user refers to an experiment by name or description, load the `finding-experiments` skill first to\nresolve it to a concrete ID.\n\nCall `experiment-get` and pull these fields. They are inputs for almost every diagnostic:\n\n- `parameters.feature_flag_variants[].rollout_percentage` — the variant split\n- `parameters.rollout_percentage` — the overall rollout (% of users entering the experiment)\n- `exposure_criteria.multiple_variant_handling` — defaults to `\"exclude\"` if absent\n- `exposure_criteria.exposure_event` — `null` means default `$feature_flag_called`\n- `exposure_criteria.filterTestAccounts` — defaults to `true`\n- `feature_flag.active`, status (`draft` \u002F `running` \u002F `paused` \u002F `stopped`), `start_date`, `end_date`\n- `feature_flag.filters.groups[].variant` — any non-null value is a forced-variant override on the\n  matched cohort (release-condition assignment, not randomized). Surfaces A7 by default.\n- `stats_config` — Bayesian (default) or Frequentist\n\n## Step 1.5 — Pull a diagnostic snapshot (verify before asking)\n\nBefore asking the user clarifying questions, pull the diagnostic snapshot in\n[references\u002Fdiagnostic-snapshot.md](references\u002Fdiagnostic-snapshot.md). Most diagnostics in this skill\ncan be confirmed or ruled out from that data without an interview.\n\n## Step 2 — Match symptom to diagnostic\n\n| User says...                                                                               | Diagnostic group                             |\n| ------------------------------------------------------------------------------------------ | -------------------------------------------- |\n| \"Smaller variant looks biased\" \u002F banner says bias                                          | A — bias & skew                              |\n| \"Variant ratio doesn't match my split\" \u002F SRM warning                                       | A — bias & skew                              |\n| \"Why isn't it 50\u002F50?\" \u002F \"users in both groups\"                                             | A — bias & skew                              |\n| \"Users in both control and test\" \u002F high `$multiple` %                                      | A — bias & skew                              |\n| Multi-variant exposure on a server-rendered app                                            | A — bias & skew                              |\n| Banner about feature-flag\u002Fexperiment state mismatch                                        | A — bias & skew                              |\n| \"Migrating distinct_id\" \u002F \"switching from anonymous to user_id\" mid-run                    | A — bias & skew                              |\n| Metric count is much smaller than exposures (e.g. 10× or 100× gap)                         | A — bias & skew (route here before D)        |\n| \"Experiment shows 0 \u002F not enough data\" \u002F empty                                             | B — empty experiment                         |\n| \"Variant always undefined \u002F false\"                                                         | B — empty experiment                         |\n| \"$feature_flag_called fires but no exposures show up\"                                      | B — empty experiment                         |\n| \"Experiment says running but exposures haven't moved in weeks\u002Fmonths\"                      | B — empty experiment                         |\n| \"Significance keeps flipping as we run longer\"                                             | C — interpretation traps                     |\n| \"Significance was declared, then it wasn't significant anymore\"                            | C — interpretation traps                     |\n| \"30\u002F16 split at 46 exposures, is this broken?\"                                             | C — interpretation traps                     |\n| \"A\u002FA test is showing significant results\"                                                  | C — interpretation traps                     |\n| \"Many metrics — some significant, some not\"                                                | C — interpretation traps                     |\n| \"Bayesian says 96% chance to win — should we ship?\"                                        | C — interpretation traps                     |\n| \"Confidence intervals overlap — does that mean not significant?\"                           | C — interpretation traps                     |\n| \"An external tool (significance calculator or AI agent) disagrees with PostHog\"            | C — interpretation traps                     |\n| \"Should I ship? Primary is up but a secondary is down\"                                     | C — interpretation traps                     |\n| \"PostHog numbers ≠ my SQL count\"                                                           | D — numbers vs SQL                           |\n| \"Funnel says X% but my raw event count says Y\"                                             | D — numbers vs SQL                           |\n| \"Sum of revenue looks wrong\" \u002F \"breakdown shows 'none'\"                                    | D — numbers vs SQL                           |\n| \"Recordings panel doesn't match the stats\"                                                 | D — numbers vs SQL                           |\n| \"I applied a filter but the user count didn't change\"                                      | D — numbers vs SQL                           |\n| \"I want to slice results by current person properties (as of now, not as of exposure)\"     | D — numbers vs SQL                           |\n| \"Changed split \u002F rollout \u002F metric \u002F criteria mid-run, now odd\"                             | E — mid-run changes                          |\n| \"Ended\u002Fshipped — flag now flipped to 0\u002F100 unexpectedly\"                                   | E — mid-run changes                          |\n| \"Long-term metric moves opposite from primary\"                                             | E — mid-run changes                          |\n| \"Retention metric counts users I didn't expect\"                                            | E — mid-run changes                          |\n| \"Can't convert the feature flag back to a simple (boolean) flag after the experiment ends\" | E — mid-run changes                          |\n| \"How do I restart an experiment with new variants?\"                                        | E — mid-run changes                          |\n| Metric line is rendered but the result block is empty \u002F no chance-to-win or significance   | E — mid-run changes (E13 legacy methodology) |\n\nIf the symptom is unclear, ask one clarifying question before picking. Most diagnostics have different fixes\n— do not guess.\n\n## Step 3 — Surface every diagnostic the evidence supports\n\nAfter matching the symptom in Step 2 and reading the relevant reference file(s), list each diagnostic\nthat applies before recommending an action.\n\nSurface co-occurring mechanisms independently — even when one is more salient, don't collapse them\ninto a single \"wait\" or \"fix\" recommendation. Different mechanisms have different fixes: a\n_systematic_ bias (e.g. uneven-split + Exclude) doesn't resolve by waiting; a _statistical_ pattern\n(e.g. small-sample variance) does. Bundling them leaves the bias in place after the user follows the\nbundled advice.\n\nOnly list mechanisms that have a path to verification in the project state — config (from\n`experiment-get`), snapshot data, activity log, or repo source. Config-derived mechanisms count: an\n80\u002F20 split with default `multiple_variant_handling=\"exclude\"` is visible in `experiment-get` and is\ntherefore enumerable. Naming a mechanism with no source (e.g. SRM when the snapshot shows a clean\nvariant ratio) is not.\n\n## Diagnostic groups\n\n### A — Bias & skew\n\nVariants don't look balanced, one variant looks biased, the in-app warning banner appeared, or users are\nshowing up under multiple variants. Covers the uneven-split + Exclude interaction, SRM, identity\nfragmentation, bootstrap × `\u002Fdecide` mismatch, and flag\u002Fexperiment state inconsistency.\n\n→ See [references\u002Fbias-and-skew.md](references\u002Fbias-and-skew.md)\n\n### B — Empty experiment \u002F 0 exposures \u002F \"not enough data\"\n\nA frequent pain point. Covers SDK call (wrong evaluation method, `identify()` timing, dedup),\nexposure capture (custom event missing variant property, required properties, ad-blockers), and\nexposure-criteria match (test-account filter, eligibility ordering, events firing before exposure).\n\n→ See [references\u002Fempty-experiment.md](references\u002Fempty-experiment.md)\n\n### C — Significance \u002F interpretation traps\n\nSignificance flipping, A\u002FA test showing significance, Bayesian vs Frequentist confusion, multiple\ncomparisons, low-volume variance, peeking \u002F early stopping. Includes the legacy stats issue (A\u002FA tests\nhistorically over-fired before the new Bayesian module) and how the win-probability methodology changed in\nJan 2025 (single test vs control, not control vs all variants).\n\n→ See [references\u002Finterpretation.md](references\u002Finterpretation.md)\n\n### D — Numbers don't match (PostHog vs the user's SQL \u002F raw count)\n\nThe experiment page applies an exposure scope, `$multiple` exclusion, test-account filter, and date range\nthat ad-hoc SQL almost never replicates. Covers funnel attribution (only first→last step counts for stats),\nbreakdowns (read from the exposure event, not the metric event), the \"sum of revenue\" mean-of-per-user\nconfusion, and the recordings-panel-vs-stats divergence.\n\n→ See [references\u002Fnumbers-vs-sql.md](references\u002Fnumbers-vs-sql.md)\n\n### E — Surprises after mid-run changes (incl. lifecycle and retention quirks)\n\nIncreasing rollout is safe; decreasing is caution; changing the variant split is an anti-pattern; adding\nmetrics mid-run is p-hacking; ship-variant can rewrite the flag in surprising ways; reset clears\nresults not the flag. Also covers retention-metric quirks (first-event-must-be-after-exposure design),\n\"matured users\" filtering, and long-term vs short-term metric divergence.\n\n→ See [references\u002Fmid-run-changes.md](references\u002Fmid-run-changes.md)\n\n## Step 4 — Calibrate recommendations to experiment state\n\nSurface diagnostics first (Step 3). Then recommend — but scope what you recommend to what the\nexperiment's current state permits.\n\n- **Draft** — config changes are free; recommend and apply.\n- **Running** — every change has a tradeoff. Explain the mid-run impact (anti-pattern? safe?\n  user-visible?) before recommending. See `configuring-experiment-rollout` and its reference file\n  `references\u002Fchanging-distribution-after-launch.md` for the mid-run rules.\n- **Stopped \u002F archived** — the experiment AND its feature flag represent the documented outcome of\n  the run. Recommendations are scoped to (a) interpretation of the existing data, (b) what to do for\n  the _next_ experiment, or (c) explaining what happened.\n\nOn a stopped or archived experiment, don't preemptively offer reversal of a state mutation\n(ship-variant flag rewrite, manual flag edit, reset, archive). If the user asks \"why did X happen?\",\nexplain X — don't append a \"here's how to undo it\" coda. That pattern assumes intent the user didn't\nsignal. Conditional offers like _\"if this wasn't intended, you could…\"_ or _\"want me to revert it?\"_\ncount as preemptive too — only the user explicitly naming the reversal action (\"how do I undo this?\",\n\"can I roll back ship-variant?\", \"how do I get the 50\u002F50 split back?\") is a request to surface\nreversal mechanics.\n\nUse consistent terminology: variant _split_ (between variants) is distinct from _rollout_ (overall %\nentering); the `$feature_flag_called` exposure event is distinct from a _custom exposure event_; the\n_Exclude_ \u002F _First seen_ options control multivariate handling, not exposure.\n",{"data":38,"body":39},{"name":4,"description":6},{"type":40,"children":41},"root",[42,50,62,67,125,132,145,158,321,327,340,346,797,802,808,813,833,860,866,873,886,896,902,915,924,930,935,944,950,962,971,977,982,991,997,1002,1058,1077],{"type":43,"tag":44,"props":45,"children":46},"element","h1",{"id":4},[47],{"type":48,"value":49},"text","Diagnosing experiment results",{"type":43,"tag":51,"props":52,"children":53},"p",{},[54,56],{"type":48,"value":55},"This skill answers: ",{"type":43,"tag":57,"props":58,"children":59},"strong",{},[60],{"type":48,"value":61},"My PostHog experiment results look wrong, biased, or empty — what's going on?",{"type":43,"tag":51,"props":63,"children":64},{},[65],{"type":48,"value":66},"Match the user's complaint in the dispatch table, then read the matching reference file for the\ndiagnostic.",{"type":43,"tag":51,"props":68,"children":69},{},[70,72,79,81,87,89,95,97,102,104,109,111,116,118,123],{"type":48,"value":71},"Each diagnostic in the reference files is tagged ",{"type":43,"tag":73,"props":74,"children":76},"code",{"className":75},[],[77],{"type":48,"value":78},"[HIGH]",{"type":48,"value":80},", ",{"type":43,"tag":73,"props":82,"children":84},{"className":83},[],[85],{"type":48,"value":86},"[MEDIUM]",{"type":48,"value":88},", or ",{"type":43,"tag":73,"props":90,"children":92},{"className":91},[],[93],{"type":48,"value":94},"[LOW]",{"type":48,"value":96}," based on how\nstrongly it's verified — ",{"type":43,"tag":73,"props":98,"children":100},{"className":99},[],[101],{"type":48,"value":78},{"type":48,"value":103}," is verified directly in PostHog code, ",{"type":43,"tag":73,"props":105,"children":107},{"className":106},[],[108],{"type":48,"value":86},{"type":48,"value":110}," is partially or\nteam-source verified, ",{"type":43,"tag":73,"props":112,"children":114},{"className":113},[],[115],{"type":48,"value":94},{"type":48,"value":117}," describes SDK\u002Fexternal behavior that wasn't verified here. Treat ",{"type":43,"tag":73,"props":119,"children":121},{"className":120},[],[122],{"type":48,"value":94},{"type":48,"value":124},"\nitems as hypotheses to test, not facts to assert.",{"type":43,"tag":126,"props":127,"children":129},"h2",{"id":128},"step-1-resolve-the-experiment",[130],{"type":48,"value":131},"Step 1 — Resolve the experiment",{"type":43,"tag":51,"props":133,"children":134},{},[135,137,143],{"type":48,"value":136},"If the user refers to an experiment by name or description, load the ",{"type":43,"tag":73,"props":138,"children":140},{"className":139},[],[141],{"type":48,"value":142},"finding-experiments",{"type":48,"value":144}," skill first to\nresolve it to a concrete ID.",{"type":43,"tag":51,"props":146,"children":147},{},[148,150,156],{"type":48,"value":149},"Call ",{"type":43,"tag":73,"props":151,"children":153},{"className":152},[],[154],{"type":48,"value":155},"experiment-get",{"type":48,"value":157}," and pull these fields. They are inputs for almost every diagnostic:",{"type":43,"tag":159,"props":160,"children":161},"ul",{},[162,174,185,204,229,245,299,310],{"type":43,"tag":163,"props":164,"children":165},"li",{},[166,172],{"type":43,"tag":73,"props":167,"children":169},{"className":168},[],[170],{"type":48,"value":171},"parameters.feature_flag_variants[].rollout_percentage",{"type":48,"value":173}," — the variant split",{"type":43,"tag":163,"props":175,"children":176},{},[177,183],{"type":43,"tag":73,"props":178,"children":180},{"className":179},[],[181],{"type":48,"value":182},"parameters.rollout_percentage",{"type":48,"value":184}," — the overall rollout (% of users entering the experiment)",{"type":43,"tag":163,"props":186,"children":187},{},[188,194,196,202],{"type":43,"tag":73,"props":189,"children":191},{"className":190},[],[192],{"type":48,"value":193},"exposure_criteria.multiple_variant_handling",{"type":48,"value":195}," — defaults to ",{"type":43,"tag":73,"props":197,"children":199},{"className":198},[],[200],{"type":48,"value":201},"\"exclude\"",{"type":48,"value":203}," if absent",{"type":43,"tag":163,"props":205,"children":206},{},[207,213,215,221,223],{"type":43,"tag":73,"props":208,"children":210},{"className":209},[],[211],{"type":48,"value":212},"exposure_criteria.exposure_event",{"type":48,"value":214}," — ",{"type":43,"tag":73,"props":216,"children":218},{"className":217},[],[219],{"type":48,"value":220},"null",{"type":48,"value":222}," means default ",{"type":43,"tag":73,"props":224,"children":226},{"className":225},[],[227],{"type":48,"value":228},"$feature_flag_called",{"type":43,"tag":163,"props":230,"children":231},{},[232,238,239],{"type":43,"tag":73,"props":233,"children":235},{"className":234},[],[236],{"type":48,"value":237},"exposure_criteria.filterTestAccounts",{"type":48,"value":195},{"type":43,"tag":73,"props":240,"children":242},{"className":241},[],[243],{"type":48,"value":244},"true",{"type":43,"tag":163,"props":246,"children":247},{},[248,254,256,262,264,270,271,277,278,284,286,292,293],{"type":43,"tag":73,"props":249,"children":251},{"className":250},[],[252],{"type":48,"value":253},"feature_flag.active",{"type":48,"value":255},", status (",{"type":43,"tag":73,"props":257,"children":259},{"className":258},[],[260],{"type":48,"value":261},"draft",{"type":48,"value":263}," \u002F ",{"type":43,"tag":73,"props":265,"children":267},{"className":266},[],[268],{"type":48,"value":269},"running",{"type":48,"value":263},{"type":43,"tag":73,"props":272,"children":274},{"className":273},[],[275],{"type":48,"value":276},"paused",{"type":48,"value":263},{"type":43,"tag":73,"props":279,"children":281},{"className":280},[],[282],{"type":48,"value":283},"stopped",{"type":48,"value":285},"), ",{"type":43,"tag":73,"props":287,"children":289},{"className":288},[],[290],{"type":48,"value":291},"start_date",{"type":48,"value":80},{"type":43,"tag":73,"props":294,"children":296},{"className":295},[],[297],{"type":48,"value":298},"end_date",{"type":43,"tag":163,"props":300,"children":301},{},[302,308],{"type":43,"tag":73,"props":303,"children":305},{"className":304},[],[306],{"type":48,"value":307},"feature_flag.filters.groups[].variant",{"type":48,"value":309}," — any non-null value is a forced-variant override on the\nmatched cohort (release-condition assignment, not randomized). Surfaces A7 by default.",{"type":43,"tag":163,"props":311,"children":312},{},[313,319],{"type":43,"tag":73,"props":314,"children":316},{"className":315},[],[317],{"type":48,"value":318},"stats_config",{"type":48,"value":320}," — Bayesian (default) or Frequentist",{"type":43,"tag":126,"props":322,"children":324},{"id":323},"step-15-pull-a-diagnostic-snapshot-verify-before-asking",[325],{"type":48,"value":326},"Step 1.5 — Pull a diagnostic snapshot (verify before asking)",{"type":43,"tag":51,"props":328,"children":329},{},[330,332,338],{"type":48,"value":331},"Before asking the user clarifying questions, pull the diagnostic snapshot in\n",{"type":43,"tag":333,"props":334,"children":336},"a",{"href":335},"references\u002Fdiagnostic-snapshot.md",[337],{"type":48,"value":335},{"type":48,"value":339},". Most diagnostics in this skill\ncan be confirmed or ruled out from that data without an interview.",{"type":43,"tag":126,"props":341,"children":343},{"id":342},"step-2-match-symptom-to-diagnostic",[344],{"type":48,"value":345},"Step 2 — Match symptom to diagnostic",{"type":43,"tag":347,"props":348,"children":349},"table",{},[350,369],{"type":43,"tag":351,"props":352,"children":353},"thead",{},[354],{"type":43,"tag":355,"props":356,"children":357},"tr",{},[358,364],{"type":43,"tag":359,"props":360,"children":361},"th",{},[362],{"type":48,"value":363},"User says...",{"type":43,"tag":359,"props":365,"children":366},{},[367],{"type":48,"value":368},"Diagnostic group",{"type":43,"tag":370,"props":371,"children":372},"tbody",{},[373,387,399,411,431,443,455,467,480,493,505,517,529,542,554,566,578,590,602,614,626,638,651,663,675,687,699,711,724,736,748,760,772,784],{"type":43,"tag":355,"props":374,"children":375},{},[376,382],{"type":43,"tag":377,"props":378,"children":379},"td",{},[380],{"type":48,"value":381},"\"Smaller variant looks biased\" \u002F banner says bias",{"type":43,"tag":377,"props":383,"children":384},{},[385],{"type":48,"value":386},"A — bias & skew",{"type":43,"tag":355,"props":388,"children":389},{},[390,395],{"type":43,"tag":377,"props":391,"children":392},{},[393],{"type":48,"value":394},"\"Variant ratio doesn't match my split\" \u002F SRM warning",{"type":43,"tag":377,"props":396,"children":397},{},[398],{"type":48,"value":386},{"type":43,"tag":355,"props":400,"children":401},{},[402,407],{"type":43,"tag":377,"props":403,"children":404},{},[405],{"type":48,"value":406},"\"Why isn't it 50\u002F50?\" \u002F \"users in both groups\"",{"type":43,"tag":377,"props":408,"children":409},{},[410],{"type":48,"value":386},{"type":43,"tag":355,"props":412,"children":413},{},[414,427],{"type":43,"tag":377,"props":415,"children":416},{},[417,419,425],{"type":48,"value":418},"\"Users in both control and test\" \u002F high ",{"type":43,"tag":73,"props":420,"children":422},{"className":421},[],[423],{"type":48,"value":424},"$multiple",{"type":48,"value":426}," %",{"type":43,"tag":377,"props":428,"children":429},{},[430],{"type":48,"value":386},{"type":43,"tag":355,"props":432,"children":433},{},[434,439],{"type":43,"tag":377,"props":435,"children":436},{},[437],{"type":48,"value":438},"Multi-variant exposure on a server-rendered app",{"type":43,"tag":377,"props":440,"children":441},{},[442],{"type":48,"value":386},{"type":43,"tag":355,"props":444,"children":445},{},[446,451],{"type":43,"tag":377,"props":447,"children":448},{},[449],{"type":48,"value":450},"Banner about feature-flag\u002Fexperiment state mismatch",{"type":43,"tag":377,"props":452,"children":453},{},[454],{"type":48,"value":386},{"type":43,"tag":355,"props":456,"children":457},{},[458,463],{"type":43,"tag":377,"props":459,"children":460},{},[461],{"type":48,"value":462},"\"Migrating distinct_id\" \u002F \"switching from anonymous to user_id\" mid-run",{"type":43,"tag":377,"props":464,"children":465},{},[466],{"type":48,"value":386},{"type":43,"tag":355,"props":468,"children":469},{},[470,475],{"type":43,"tag":377,"props":471,"children":472},{},[473],{"type":48,"value":474},"Metric count is much smaller than exposures (e.g. 10× or 100× gap)",{"type":43,"tag":377,"props":476,"children":477},{},[478],{"type":48,"value":479},"A — bias & skew (route here before D)",{"type":43,"tag":355,"props":481,"children":482},{},[483,488],{"type":43,"tag":377,"props":484,"children":485},{},[486],{"type":48,"value":487},"\"Experiment shows 0 \u002F not enough data\" \u002F empty",{"type":43,"tag":377,"props":489,"children":490},{},[491],{"type":48,"value":492},"B — empty experiment",{"type":43,"tag":355,"props":494,"children":495},{},[496,501],{"type":43,"tag":377,"props":497,"children":498},{},[499],{"type":48,"value":500},"\"Variant always undefined \u002F false\"",{"type":43,"tag":377,"props":502,"children":503},{},[504],{"type":48,"value":492},{"type":43,"tag":355,"props":506,"children":507},{},[508,513],{"type":43,"tag":377,"props":509,"children":510},{},[511],{"type":48,"value":512},"\"$feature_flag_called fires but no exposures show up\"",{"type":43,"tag":377,"props":514,"children":515},{},[516],{"type":48,"value":492},{"type":43,"tag":355,"props":518,"children":519},{},[520,525],{"type":43,"tag":377,"props":521,"children":522},{},[523],{"type":48,"value":524},"\"Experiment says running but exposures haven't moved in weeks\u002Fmonths\"",{"type":43,"tag":377,"props":526,"children":527},{},[528],{"type":48,"value":492},{"type":43,"tag":355,"props":530,"children":531},{},[532,537],{"type":43,"tag":377,"props":533,"children":534},{},[535],{"type":48,"value":536},"\"Significance keeps flipping as we run longer\"",{"type":43,"tag":377,"props":538,"children":539},{},[540],{"type":48,"value":541},"C — interpretation traps",{"type":43,"tag":355,"props":543,"children":544},{},[545,550],{"type":43,"tag":377,"props":546,"children":547},{},[548],{"type":48,"value":549},"\"Significance was declared, then it wasn't significant anymore\"",{"type":43,"tag":377,"props":551,"children":552},{},[553],{"type":48,"value":541},{"type":43,"tag":355,"props":555,"children":556},{},[557,562],{"type":43,"tag":377,"props":558,"children":559},{},[560],{"type":48,"value":561},"\"30\u002F16 split at 46 exposures, is this broken?\"",{"type":43,"tag":377,"props":563,"children":564},{},[565],{"type":48,"value":541},{"type":43,"tag":355,"props":567,"children":568},{},[569,574],{"type":43,"tag":377,"props":570,"children":571},{},[572],{"type":48,"value":573},"\"A\u002FA test is showing significant results\"",{"type":43,"tag":377,"props":575,"children":576},{},[577],{"type":48,"value":541},{"type":43,"tag":355,"props":579,"children":580},{},[581,586],{"type":43,"tag":377,"props":582,"children":583},{},[584],{"type":48,"value":585},"\"Many metrics — some significant, some not\"",{"type":43,"tag":377,"props":587,"children":588},{},[589],{"type":48,"value":541},{"type":43,"tag":355,"props":591,"children":592},{},[593,598],{"type":43,"tag":377,"props":594,"children":595},{},[596],{"type":48,"value":597},"\"Bayesian says 96% chance to win — should we ship?\"",{"type":43,"tag":377,"props":599,"children":600},{},[601],{"type":48,"value":541},{"type":43,"tag":355,"props":603,"children":604},{},[605,610],{"type":43,"tag":377,"props":606,"children":607},{},[608],{"type":48,"value":609},"\"Confidence intervals overlap — does that mean not significant?\"",{"type":43,"tag":377,"props":611,"children":612},{},[613],{"type":48,"value":541},{"type":43,"tag":355,"props":615,"children":616},{},[617,622],{"type":43,"tag":377,"props":618,"children":619},{},[620],{"type":48,"value":621},"\"An external tool (significance calculator or AI agent) disagrees with PostHog\"",{"type":43,"tag":377,"props":623,"children":624},{},[625],{"type":48,"value":541},{"type":43,"tag":355,"props":627,"children":628},{},[629,634],{"type":43,"tag":377,"props":630,"children":631},{},[632],{"type":48,"value":633},"\"Should I ship? Primary is up but a secondary is down\"",{"type":43,"tag":377,"props":635,"children":636},{},[637],{"type":48,"value":541},{"type":43,"tag":355,"props":639,"children":640},{},[641,646],{"type":43,"tag":377,"props":642,"children":643},{},[644],{"type":48,"value":645},"\"PostHog numbers ≠ my SQL count\"",{"type":43,"tag":377,"props":647,"children":648},{},[649],{"type":48,"value":650},"D — numbers vs SQL",{"type":43,"tag":355,"props":652,"children":653},{},[654,659],{"type":43,"tag":377,"props":655,"children":656},{},[657],{"type":48,"value":658},"\"Funnel says X% but my raw event count says Y\"",{"type":43,"tag":377,"props":660,"children":661},{},[662],{"type":48,"value":650},{"type":43,"tag":355,"props":664,"children":665},{},[666,671],{"type":43,"tag":377,"props":667,"children":668},{},[669],{"type":48,"value":670},"\"Sum of revenue looks wrong\" \u002F \"breakdown shows 'none'\"",{"type":43,"tag":377,"props":672,"children":673},{},[674],{"type":48,"value":650},{"type":43,"tag":355,"props":676,"children":677},{},[678,683],{"type":43,"tag":377,"props":679,"children":680},{},[681],{"type":48,"value":682},"\"Recordings panel doesn't match the stats\"",{"type":43,"tag":377,"props":684,"children":685},{},[686],{"type":48,"value":650},{"type":43,"tag":355,"props":688,"children":689},{},[690,695],{"type":43,"tag":377,"props":691,"children":692},{},[693],{"type":48,"value":694},"\"I applied a filter but the user count didn't change\"",{"type":43,"tag":377,"props":696,"children":697},{},[698],{"type":48,"value":650},{"type":43,"tag":355,"props":700,"children":701},{},[702,707],{"type":43,"tag":377,"props":703,"children":704},{},[705],{"type":48,"value":706},"\"I want to slice results by current person properties (as of now, not as of exposure)\"",{"type":43,"tag":377,"props":708,"children":709},{},[710],{"type":48,"value":650},{"type":43,"tag":355,"props":712,"children":713},{},[714,719],{"type":43,"tag":377,"props":715,"children":716},{},[717],{"type":48,"value":718},"\"Changed split \u002F rollout \u002F metric \u002F criteria mid-run, now odd\"",{"type":43,"tag":377,"props":720,"children":721},{},[722],{"type":48,"value":723},"E — mid-run changes",{"type":43,"tag":355,"props":725,"children":726},{},[727,732],{"type":43,"tag":377,"props":728,"children":729},{},[730],{"type":48,"value":731},"\"Ended\u002Fshipped — flag now flipped to 0\u002F100 unexpectedly\"",{"type":43,"tag":377,"props":733,"children":734},{},[735],{"type":48,"value":723},{"type":43,"tag":355,"props":737,"children":738},{},[739,744],{"type":43,"tag":377,"props":740,"children":741},{},[742],{"type":48,"value":743},"\"Long-term metric moves opposite from primary\"",{"type":43,"tag":377,"props":745,"children":746},{},[747],{"type":48,"value":723},{"type":43,"tag":355,"props":749,"children":750},{},[751,756],{"type":43,"tag":377,"props":752,"children":753},{},[754],{"type":48,"value":755},"\"Retention metric counts users I didn't expect\"",{"type":43,"tag":377,"props":757,"children":758},{},[759],{"type":48,"value":723},{"type":43,"tag":355,"props":761,"children":762},{},[763,768],{"type":43,"tag":377,"props":764,"children":765},{},[766],{"type":48,"value":767},"\"Can't convert the feature flag back to a simple (boolean) flag after the experiment ends\"",{"type":43,"tag":377,"props":769,"children":770},{},[771],{"type":48,"value":723},{"type":43,"tag":355,"props":773,"children":774},{},[775,780],{"type":43,"tag":377,"props":776,"children":777},{},[778],{"type":48,"value":779},"\"How do I restart an experiment with new variants?\"",{"type":43,"tag":377,"props":781,"children":782},{},[783],{"type":48,"value":723},{"type":43,"tag":355,"props":785,"children":786},{},[787,792],{"type":43,"tag":377,"props":788,"children":789},{},[790],{"type":48,"value":791},"Metric line is rendered but the result block is empty \u002F no chance-to-win or significance",{"type":43,"tag":377,"props":793,"children":794},{},[795],{"type":48,"value":796},"E — mid-run changes (E13 legacy methodology)",{"type":43,"tag":51,"props":798,"children":799},{},[800],{"type":48,"value":801},"If the symptom is unclear, ask one clarifying question before picking. Most diagnostics have different fixes\n— do not guess.",{"type":43,"tag":126,"props":803,"children":805},{"id":804},"step-3-surface-every-diagnostic-the-evidence-supports",[806],{"type":48,"value":807},"Step 3 — Surface every diagnostic the evidence supports",{"type":43,"tag":51,"props":809,"children":810},{},[811],{"type":48,"value":812},"After matching the symptom in Step 2 and reading the relevant reference file(s), list each diagnostic\nthat applies before recommending an action.",{"type":43,"tag":51,"props":814,"children":815},{},[816,818,824,826,831],{"type":48,"value":817},"Surface co-occurring mechanisms independently — even when one is more salient, don't collapse them\ninto a single \"wait\" or \"fix\" recommendation. Different mechanisms have different fixes: a\n",{"type":43,"tag":819,"props":820,"children":821},"em",{},[822],{"type":48,"value":823},"systematic",{"type":48,"value":825}," bias (e.g. uneven-split + Exclude) doesn't resolve by waiting; a ",{"type":43,"tag":819,"props":827,"children":828},{},[829],{"type":48,"value":830},"statistical",{"type":48,"value":832}," pattern\n(e.g. small-sample variance) does. Bundling them leaves the bias in place after the user follows the\nbundled advice.",{"type":43,"tag":51,"props":834,"children":835},{},[836,838,843,845,851,853,858],{"type":48,"value":837},"Only list mechanisms that have a path to verification in the project state — config (from\n",{"type":43,"tag":73,"props":839,"children":841},{"className":840},[],[842],{"type":48,"value":155},{"type":48,"value":844},"), snapshot data, activity log, or repo source. Config-derived mechanisms count: an\n80\u002F20 split with default ",{"type":43,"tag":73,"props":846,"children":848},{"className":847},[],[849],{"type":48,"value":850},"multiple_variant_handling=\"exclude\"",{"type":48,"value":852}," is visible in ",{"type":43,"tag":73,"props":854,"children":856},{"className":855},[],[857],{"type":48,"value":155},{"type":48,"value":859}," and is\ntherefore enumerable. Naming a mechanism with no source (e.g. SRM when the snapshot shows a clean\nvariant ratio) is not.",{"type":43,"tag":126,"props":861,"children":863},{"id":862},"diagnostic-groups",[864],{"type":48,"value":865},"Diagnostic groups",{"type":43,"tag":867,"props":868,"children":870},"h3",{"id":869},"a-bias-skew",[871],{"type":48,"value":872},"A — Bias & skew",{"type":43,"tag":51,"props":874,"children":875},{},[876,878,884],{"type":48,"value":877},"Variants don't look balanced, one variant looks biased, the in-app warning banner appeared, or users are\nshowing up under multiple variants. Covers the uneven-split + Exclude interaction, SRM, identity\nfragmentation, bootstrap × ",{"type":43,"tag":73,"props":879,"children":881},{"className":880},[],[882],{"type":48,"value":883},"\u002Fdecide",{"type":48,"value":885}," mismatch, and flag\u002Fexperiment state inconsistency.",{"type":43,"tag":51,"props":887,"children":888},{},[889,891],{"type":48,"value":890},"→ See ",{"type":43,"tag":333,"props":892,"children":894},{"href":893},"references\u002Fbias-and-skew.md",[895],{"type":48,"value":893},{"type":43,"tag":867,"props":897,"children":899},{"id":898},"b-empty-experiment-0-exposures-not-enough-data",[900],{"type":48,"value":901},"B — Empty experiment \u002F 0 exposures \u002F \"not enough data\"",{"type":43,"tag":51,"props":903,"children":904},{},[905,907,913],{"type":48,"value":906},"A frequent pain point. Covers SDK call (wrong evaluation method, ",{"type":43,"tag":73,"props":908,"children":910},{"className":909},[],[911],{"type":48,"value":912},"identify()",{"type":48,"value":914}," timing, dedup),\nexposure capture (custom event missing variant property, required properties, ad-blockers), and\nexposure-criteria match (test-account filter, eligibility ordering, events firing before exposure).",{"type":43,"tag":51,"props":916,"children":917},{},[918,919],{"type":48,"value":890},{"type":43,"tag":333,"props":920,"children":922},{"href":921},"references\u002Fempty-experiment.md",[923],{"type":48,"value":921},{"type":43,"tag":867,"props":925,"children":927},{"id":926},"c-significance-interpretation-traps",[928],{"type":48,"value":929},"C — Significance \u002F interpretation traps",{"type":43,"tag":51,"props":931,"children":932},{},[933],{"type":48,"value":934},"Significance flipping, A\u002FA test showing significance, Bayesian vs Frequentist confusion, multiple\ncomparisons, low-volume variance, peeking \u002F early stopping. Includes the legacy stats issue (A\u002FA tests\nhistorically over-fired before the new Bayesian module) and how the win-probability methodology changed in\nJan 2025 (single test vs control, not control vs all variants).",{"type":43,"tag":51,"props":936,"children":937},{},[938,939],{"type":48,"value":890},{"type":43,"tag":333,"props":940,"children":942},{"href":941},"references\u002Finterpretation.md",[943],{"type":48,"value":941},{"type":43,"tag":867,"props":945,"children":947},{"id":946},"d-numbers-dont-match-posthog-vs-the-users-sql-raw-count",[948],{"type":48,"value":949},"D — Numbers don't match (PostHog vs the user's SQL \u002F raw count)",{"type":43,"tag":51,"props":951,"children":952},{},[953,955,960],{"type":48,"value":954},"The experiment page applies an exposure scope, ",{"type":43,"tag":73,"props":956,"children":958},{"className":957},[],[959],{"type":48,"value":424},{"type":48,"value":961}," exclusion, test-account filter, and date range\nthat ad-hoc SQL almost never replicates. Covers funnel attribution (only first→last step counts for stats),\nbreakdowns (read from the exposure event, not the metric event), the \"sum of revenue\" mean-of-per-user\nconfusion, and the recordings-panel-vs-stats divergence.",{"type":43,"tag":51,"props":963,"children":964},{},[965,966],{"type":48,"value":890},{"type":43,"tag":333,"props":967,"children":969},{"href":968},"references\u002Fnumbers-vs-sql.md",[970],{"type":48,"value":968},{"type":43,"tag":867,"props":972,"children":974},{"id":973},"e-surprises-after-mid-run-changes-incl-lifecycle-and-retention-quirks",[975],{"type":48,"value":976},"E — Surprises after mid-run changes (incl. lifecycle and retention quirks)",{"type":43,"tag":51,"props":978,"children":979},{},[980],{"type":48,"value":981},"Increasing rollout is safe; decreasing is caution; changing the variant split is an anti-pattern; adding\nmetrics mid-run is p-hacking; ship-variant can rewrite the flag in surprising ways; reset clears\nresults not the flag. Also covers retention-metric quirks (first-event-must-be-after-exposure design),\n\"matured users\" filtering, and long-term vs short-term metric divergence.",{"type":43,"tag":51,"props":983,"children":984},{},[985,986],{"type":48,"value":890},{"type":43,"tag":333,"props":987,"children":989},{"href":988},"references\u002Fmid-run-changes.md",[990],{"type":48,"value":988},{"type":43,"tag":126,"props":992,"children":994},{"id":993},"step-4-calibrate-recommendations-to-experiment-state",[995],{"type":48,"value":996},"Step 4 — Calibrate recommendations to experiment state",{"type":43,"tag":51,"props":998,"children":999},{},[1000],{"type":48,"value":1001},"Surface diagnostics first (Step 3). Then recommend — but scope what you recommend to what the\nexperiment's current state permits.",{"type":43,"tag":159,"props":1003,"children":1004},{},[1005,1015,1041],{"type":43,"tag":163,"props":1006,"children":1007},{},[1008,1013],{"type":43,"tag":57,"props":1009,"children":1010},{},[1011],{"type":48,"value":1012},"Draft",{"type":48,"value":1014}," — config changes are free; recommend and apply.",{"type":43,"tag":163,"props":1016,"children":1017},{},[1018,1023,1025,1031,1033,1039],{"type":43,"tag":57,"props":1019,"children":1020},{},[1021],{"type":48,"value":1022},"Running",{"type":48,"value":1024}," — every change has a tradeoff. Explain the mid-run impact (anti-pattern? safe?\nuser-visible?) before recommending. See ",{"type":43,"tag":73,"props":1026,"children":1028},{"className":1027},[],[1029],{"type":48,"value":1030},"configuring-experiment-rollout",{"type":48,"value":1032}," and its reference file\n",{"type":43,"tag":73,"props":1034,"children":1036},{"className":1035},[],[1037],{"type":48,"value":1038},"references\u002Fchanging-distribution-after-launch.md",{"type":48,"value":1040}," for the mid-run rules.",{"type":43,"tag":163,"props":1042,"children":1043},{},[1044,1049,1051,1056],{"type":43,"tag":57,"props":1045,"children":1046},{},[1047],{"type":48,"value":1048},"Stopped \u002F archived",{"type":48,"value":1050}," — the experiment AND its feature flag represent the documented outcome of\nthe run. Recommendations are scoped to (a) interpretation of the existing data, (b) what to do for\nthe ",{"type":43,"tag":819,"props":1052,"children":1053},{},[1054],{"type":48,"value":1055},"next",{"type":48,"value":1057}," experiment, or (c) explaining what happened.",{"type":43,"tag":51,"props":1059,"children":1060},{},[1061,1063,1068,1070,1075],{"type":48,"value":1062},"On a stopped or archived experiment, don't preemptively offer reversal of a state mutation\n(ship-variant flag rewrite, manual flag edit, reset, archive). If the user asks \"why did X happen?\",\nexplain X — don't append a \"here's how to undo it\" coda. That pattern assumes intent the user didn't\nsignal. Conditional offers like ",{"type":43,"tag":819,"props":1064,"children":1065},{},[1066],{"type":48,"value":1067},"\"if this wasn't intended, you could…\"",{"type":48,"value":1069}," or ",{"type":43,"tag":819,"props":1071,"children":1072},{},[1073],{"type":48,"value":1074},"\"want me to revert it?\"",{"type":48,"value":1076},"\ncount as preemptive too — only the user explicitly naming the reversal action (\"how do I undo this?\",\n\"can I roll back ship-variant?\", \"how do I get the 50\u002F50 split back?\") is a request to surface\nreversal mechanics.",{"type":43,"tag":51,"props":1078,"children":1079},{},[1080,1082,1087,1089,1094,1096,1101,1103,1108,1110,1115,1116,1121],{"type":48,"value":1081},"Use consistent terminology: variant ",{"type":43,"tag":819,"props":1083,"children":1084},{},[1085],{"type":48,"value":1086},"split",{"type":48,"value":1088}," (between variants) is distinct from ",{"type":43,"tag":819,"props":1090,"children":1091},{},[1092],{"type":48,"value":1093},"rollout",{"type":48,"value":1095}," (overall %\nentering); the ",{"type":43,"tag":73,"props":1097,"children":1099},{"className":1098},[],[1100],{"type":48,"value":228},{"type":48,"value":1102}," exposure event is distinct from a ",{"type":43,"tag":819,"props":1104,"children":1105},{},[1106],{"type":48,"value":1107},"custom exposure event",{"type":48,"value":1109},"; the\n",{"type":43,"tag":819,"props":1111,"children":1112},{},[1113],{"type":48,"value":1114},"Exclude",{"type":48,"value":263},{"type":43,"tag":819,"props":1117,"children":1118},{},[1119],{"type":48,"value":1120},"First seen",{"type":48,"value":1122}," options control multivariate handling, not exposure.",{"items":1124,"total":1293},[1125,1142,1154,1167,1180,1193,1209,1226,1240,1255,1265,1283],{"slug":1126,"name":1126,"fn":1127,"description":1128,"org":1129,"tags":1130,"stars":1139,"repoUrl":1140,"updatedAt":1141},"analyzing-expensive-users","analyze expensive users in AI observability","Analyze the most expensive users in AI observability and explain why they cost so much. Use when the user asks about top spenders, expensive users, per-user LLM cost, user-level cost drivers, or patterns behind high AI observability spend.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1131,1132,1135,1138],{"name":18,"slug":19,"type":13},{"name":1133,"slug":1134,"type":13},"Cost Optimization","cost-optimization",{"name":1136,"slug":1137,"type":13},"Observability","observability",{"name":9,"slug":8,"type":13},35568,"https:\u002F\u002Fgithub.com\u002FPostHog\u002Fposthog","2026-07-28T05:34:11.117757",{"slug":1143,"name":1143,"fn":1144,"description":1145,"org":1146,"tags":1147,"stars":1139,"repoUrl":1140,"updatedAt":1153},"auditing-endpoints","audit PostHog project endpoints","Audit every endpoint in a PostHog project for staleness, failed materialisations, and unused materialised versions. Use when the user asks \"what endpoints can I clean up?\", \"are any of my endpoints broken?\", \"which materialised versions are still being called?\", or wants a one-shot cleanup pass over the Endpoints product. Produces a prioritised report grouped by issue type, with recommended actions but does not modify anything without explicit confirmation.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1148,1149,1152],{"name":18,"slug":19,"type":13},{"name":1150,"slug":1151,"type":13},"Audit","audit",{"name":9,"slug":8,"type":13},"2026-06-08T08:08:33.693989",{"slug":1155,"name":1155,"fn":1156,"description":1157,"org":1158,"tags":1159,"stars":1139,"repoUrl":1140,"updatedAt":1166},"auditing-warehouse-source-health","audit PostHog data warehouse source health","Audit the health of a PostHog project's data warehouse sources and syncs — find every broken or degraded source connection, sync schema, and webhook channel. Use when the user asks \"why are my imports failing?\", \"what's broken with my sources?\", \"why is my warehouse data stale?\", or wants a one-shot triage of source\u002Fsync health before deciding where to dig in. Produces a prioritized report grouped by severity, with recommended next steps. For materialized-view health use `auditing-warehouse-view-health`; for a single failing sync use `diagnosing-failed-warehouse-syncs`.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1160,1161,1164,1165],{"name":1150,"slug":1151,"type":13},{"name":1162,"slug":1163,"type":13},"Data Warehouse","data-warehouse",{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-06-18T08:22:57.67984",{"slug":1168,"name":1168,"fn":1169,"description":1170,"org":1171,"tags":1172,"stars":1139,"repoUrl":1140,"updatedAt":1179},"auditing-warehouse-view-health","audit PostHog materialized view health","Audit the health of a PostHog project's materialized views (saved queries) — find every failed materialization and flag unused or stale materialized views that cost storage and compute. Use when the user asks \"which of my views are broken?\", \"why is this materialized view failing?\", \"are any of my views wasting compute?\", or wants a one-shot triage of view health. For source\u002Fsync health use `auditing-warehouse-source-health`.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1173,1174,1175,1178],{"name":1150,"slug":1151,"type":13},{"name":1162,"slug":1163,"type":13},{"name":1176,"slug":1177,"type":13},"Performance","performance",{"name":9,"slug":8,"type":13},"2026-06-18T08:25:10.936787",{"slug":1181,"name":1181,"fn":1182,"description":1183,"org":1184,"tags":1185,"stars":1139,"repoUrl":1140,"updatedAt":1192},"authoring-error-tracking-alerts","author PostHog error tracking alerts","Author error tracking alerts that fire when an issue is created, reopened, or starts spiking. Use when the user asks to set up error notifications, route exceptions to Slack\u002Fwebhook\u002FLinear, or evaluate which error events are worth alerting on. Covers trigger-event selection, integration choice, dedup against existing alerts, and shipping with the canonical message body shape.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1186,1189,1190,1191],{"name":1187,"slug":1188,"type":13},"Alerting","alerting",{"name":24,"slug":25,"type":13},{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-06-18T08:24:40.318583",{"slug":1194,"name":1194,"fn":1195,"description":1196,"org":1197,"tags":1198,"stars":1139,"repoUrl":1140,"updatedAt":1208},"authoring-log-alerts","author log alerts in PostHog","Author useful, low-noise log alerts on services in a PostHog project. Use when the user asks to set up alerts for their logs, suggest alerts they should add, or evaluate whether a service is worth monitoring. Covers service triage, baseline characterisation, threshold drafting, back-testing via simulate, and shipping with a notification destination.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1199,1200,1203,1204,1207],{"name":18,"slug":19,"type":13},{"name":1201,"slug":1202,"type":13},"Monitoring","monitoring",{"name":1136,"slug":1137,"type":13},{"name":1205,"slug":1206,"type":13},"Operations","operations",{"name":9,"slug":8,"type":13},"2026-07-18T05:10:54.430898",{"slug":1210,"name":1210,"fn":1211,"description":1212,"org":1213,"tags":1214,"stars":1139,"repoUrl":1140,"updatedAt":1225},"building-workflows","build and edit PostHog workflows","Build, edit, test, enable, and monitor PostHog workflows over MCP. Author the action\u002Fedge graph so it runs and opens cleanly in the visual editor, then change drafts surgically with patch operations. Use when asked to build, set up, automate, change, fix, or debug a workflow, campaign, broadcast, drip sequence, or event-triggered automation in the workflows product.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1215,1218,1221,1222],{"name":1216,"slug":1217,"type":13},"Automation","automation",{"name":1219,"slug":1220,"type":13},"MCP","mcp",{"name":9,"slug":8,"type":13},{"name":1223,"slug":1224,"type":13},"Workflow Automation","workflow-automation","2026-07-28T05:34:12.167015",{"slug":1227,"name":1227,"fn":1228,"description":1229,"org":1230,"tags":1231,"stars":1139,"repoUrl":1140,"updatedAt":1239},"check-posthog-loading","inspect PostHog SDK loading across URLs","Inspect how the PostHog JavaScript SDK is loaded across a list of URLs. Use to confirm consistent installation across pages, find pages missing the snippet, detect mismatched API keys or hosts between pages, and verify the load method (head snippet vs deferred vs array.js).\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1232,1233,1234,1237,1238],{"name":18,"slug":19,"type":13},{"name":24,"slug":25,"type":13},{"name":1235,"slug":1236,"type":13},"Frontend","frontend",{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-05-07T05:56:19.828048",{"slug":1241,"name":1241,"fn":1242,"description":1243,"org":1244,"tags":1245,"stars":1139,"repoUrl":1140,"updatedAt":1254},"consuming-endpoints-from-client-code","integrate PostHog endpoints into client applications","Wire a PostHog endpoint into a client app or SDK. Covers fetching the OpenAPI spec, generating a typed client with openapi-generator or @hey-api\u002Fopenapi-ts, sending the right auth header, shaping the variables payload (HogQL code_name vs insight breakdown property), handling rate-limit and materialised-endpoint error responses. Use when the user says \"how do I call my endpoint\", \"generate a client for this\", or \"what auth header do I use\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1246,1249,1250,1251],{"name":1247,"slug":1248,"type":13},"API Development","api-development",{"name":1235,"slug":1236,"type":13},{"name":9,"slug":8,"type":13},{"name":1252,"slug":1253,"type":13},"SDK","sdk","2026-06-08T08:08:34.929454",{"slug":1256,"name":1256,"fn":1257,"description":1258,"org":1259,"tags":1260,"stars":1139,"repoUrl":1140,"updatedAt":1264},"copying-endpoints-across-projects","copy PostHog endpoints across projects","Copy a PostHog endpoint (a saved HogQL\u002Finsight query exposed as an API route) to another project in the same organization, or duplicate it under a new name in the same project. Use when the user wants to duplicate an endpoint, promote an endpoint from staging to production, replicate an endpoint's query\u002Fvariables\u002Ffreshness config in another workspace, or clone an endpoint to iterate on it. Unlike feature flags and experiments, endpoints have NO native cross-project copy tool — this skill covers the read-then-recreate flow (endpoint-get then endpoint-create), the active-project switching it requires, name-collision checks, and the safe defaults (land unmaterialised in the target, verify with endpoint-run). Does not cover editing endpoint versions (see managing-endpoint-versions) or authoring a brand-new endpoint from scratch (see creating-an-endpoint).\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1261,1262,1263],{"name":1247,"slug":1248,"type":13},{"name":1205,"slug":1206,"type":13},{"name":9,"slug":8,"type":13},"2026-07-15T05:29:58.442727",{"slug":1266,"name":1266,"fn":1267,"description":1268,"org":1269,"tags":1270,"stars":1139,"repoUrl":1140,"updatedAt":1282},"creating-ai-subscription","schedule recurring AI-generated PostHog reports","Create a recurring AI-generated PostHog report — schedule a free-text prompt to run on a cron, with the LLM-synthesized markdown delivered to email or Slack on each tick. Use when the user wants a recurring AI summary of X on any cadence (daily, weekly, monthly, yearly) rather than a one-off report. (To attach an AI summary to an existing insight\u002Fdashboard subscription instead of a free-text prompt, see `managing-subscriptions` and its `summary_enabled` option.)\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1271,1272,1275,1276,1279],{"name":1216,"slug":1217,"type":13},{"name":1273,"slug":1274,"type":13},"Email","email",{"name":9,"slug":8,"type":13},{"name":1277,"slug":1278,"type":13},"Reporting","reporting",{"name":1280,"slug":1281,"type":13},"Slack","slack","2026-06-09T07:32:27.935712",{"slug":1284,"name":1284,"fn":1285,"description":1286,"org":1287,"tags":1288,"stars":1139,"repoUrl":1140,"updatedAt":1292},"creating-an-endpoint","create PostHog API endpoints","Create a PostHog endpoint with the right shape on the first try — covers query kind choice, name conventions, what to expose as variables (HogQL code_name vs insight breakdown), data_freshness_seconds, and whether to materialise on day one. Use when the user says \"create an endpoint\", \"expose this query as an API\", \"turn this insight into an endpoint\", or asks for help structuring a new endpoint. Steers away from common mistakes: materialising a query with cohort breakdowns or compare mode, inline-only variables on a materialised endpoint, unbounded date ranges, ambiguous names.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1289,1290,1291],{"name":18,"slug":19,"type":13},{"name":1247,"slug":1248,"type":13},{"name":9,"slug":8,"type":13},"2026-06-08T08:08:29.624498",231,{"items":1295,"total":1390},[1296,1313,1329,1343,1351,1363,1374],{"slug":1297,"name":1297,"fn":1298,"description":1299,"org":1300,"tags":1301,"stars":26,"repoUrl":27,"updatedAt":1312},"account-handover","draft sales account handover notes","Draft structured handover notes for transitioning a PostHog account from one TAM or CSM to another. Use this skill when a TAM needs to hand over an account, prepare a transition briefing, write handover notes, create an account summary for a new owner, or any request involving account transitions between TAMs or CSMs. Triggers on \"hand over this account\", \"transition account to\", \"draft handover notes\", \"account briefing for new TAM\", \"prepare account transition\", or when a TAM names an account and says they're leaving or reassigning it.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1302,1305,1308,1309],{"name":1303,"slug":1304,"type":13},"Communications","communications",{"name":1306,"slug":1307,"type":13},"CRM","crm",{"name":9,"slug":8,"type":13},{"name":1310,"slug":1311,"type":13},"Sales","sales","2026-04-16T05:13:00.172732",{"slug":1314,"name":1314,"fn":1315,"description":1316,"org":1317,"tags":1318,"stars":26,"repoUrl":27,"updatedAt":1328},"auditing-warehouse-data-health","audit PostHog data warehouse health","Audit the health of a PostHog project's data warehouse — find every broken or degraded pipeline item across sources, sync schemas, materialized views, batch exports, and transformations. Use when the user asks \"what's broken in my warehouse?\", \"give me a health check\", \"audit my data pipeline\", \"why are some dashboards stale?\", or wants a one-shot triage summary before deciding where to spend time. Produces a prioritized report of issues grouped by severity and type, with recommended next steps.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1319,1320,1323,1326,1327],{"name":1150,"slug":1151,"type":13},{"name":1321,"slug":1322,"type":13},"Data Engineering","data-engineering",{"name":1324,"slug":1325,"type":13},"Data Quality","data-quality",{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-06-21T08:19:05.85849",{"slug":1330,"name":1330,"fn":1331,"description":1332,"org":1333,"tags":1334,"stars":26,"repoUrl":27,"updatedAt":1342},"copying-flags-across-projects","copy feature flags across PostHog projects","Copy a feature flag from one PostHog project to one or more target projects in the same organization. Use when the user wants to duplicate a flag, promote a flag from staging to production, sync flags across projects, or replicate a flag configuration in a different workspace. Covers cohort remapping, scheduled-change handling, encrypted payloads, and the safe defaults (disabled in target, no scheduled changes).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1335,1338,1341],{"name":1336,"slug":1337,"type":13},"Deployment","deployment",{"name":1339,"slug":1340,"type":13},"Feature Flags","feature-flags",{"name":9,"slug":8,"type":13},"2026-05-04T05:56:44.484909",{"slug":4,"name":4,"fn":5,"description":6,"org":1344,"tags":1345,"stars":26,"repoUrl":27,"updatedAt":28},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1346,1347,1348,1349,1350],{"name":21,"slug":22,"type":13},{"name":18,"slug":19,"type":13},{"name":15,"slug":16,"type":13},{"name":24,"slug":25,"type":13},{"name":9,"slug":8,"type":13},{"slug":1352,"name":1352,"fn":1353,"description":1354,"org":1355,"tags":1356,"stars":26,"repoUrl":27,"updatedAt":1362},"diagnosing-missing-recordings","diagnose missing PostHog session recordings","Diagnoses why a session recording is missing or was not captured. Use when a user asks why a session has no replay, why recordings aren't appearing, or wants to troubleshoot session replay capture issues for a specific session ID or across their project. Covers SDK diagnostic signals, project settings, sampling, triggers, ad blockers, and quota\u002Fbilling scenarios.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1357,1358,1359,1360,1361],{"name":18,"slug":19,"type":13},{"name":24,"slug":25,"type":13},{"name":1235,"slug":1236,"type":13},{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-04-22T05:06:51.989772",{"slug":1364,"name":1364,"fn":1365,"description":1366,"org":1367,"tags":1368,"stars":26,"repoUrl":27,"updatedAt":1373},"diagnosing-sdk-health","diagnose PostHog SDK health","Diagnoses the health of a project's PostHog SDK integrations — which SDKs are out of date and how to fix them. Use when a user asks about PostHog SDK versions, outdated SDKs, upgrade recommendations, \"SDK health\", \"SDK doctor\" (the former name), or when events or features seem off and it might be due to an old SDK.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1369,1370,1371,1372],{"name":18,"slug":19,"type":13},{"name":24,"slug":25,"type":13},{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-04-27T05:46:14.554016",{"slug":1375,"name":1375,"fn":1376,"description":1377,"org":1378,"tags":1379,"stars":26,"repoUrl":27,"updatedAt":1389},"error-tracking-android","track Android errors with PostHog","PostHog error tracking for Android",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1380,1383,1384,1387,1388],{"name":1381,"slug":1382,"type":13},"Android","android",{"name":24,"slug":25,"type":13},{"name":1385,"slug":1386,"type":13},"Mobile","mobile",{"name":1136,"slug":1137,"type":13},{"name":9,"slug":8,"type":13},"2026-04-06T18:46:26.982494",110]