[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-posthog-exploring-mcp-tool-quality":3,"mdc-woh6sw-key":49,"related-org-posthog-exploring-mcp-tool-quality":1025,"related-repo-posthog-exploring-mcp-tool-quality":1190},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":44,"sourceUrl":47,"mdContent":48},"exploring-mcp-tool-quality","investigate PostHog MCP tool quality","Investigate the quality of PostHog MCP tool calls — error rates, latency, reach, and which tools are failing or slow. Use when the user asks \"which MCP tool has the highest error rate?\", \"what's the slowest tool?\", \"which tools fail most often?\", \"how reliable is tool X?\", wants a tool-quality matrix, or pastes an MCP analytics tool-quality \u002F dashboard URL and asks what it shows.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"posthog","PostHog","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fposthog.png",[12,16,17,20],{"name":13,"slug":14,"type":15},"Observability","observability","tag",{"name":9,"slug":8,"type":15},{"name":18,"slug":19,"type":15},"MCP","mcp",{"name":21,"slug":22,"type":15},"Debugging","debugging",35568,"https:\u002F\u002Fgithub.com\u002FPostHog\u002Fposthog","2026-07-24T05:39:42.81578",null,2977,[29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"ab-testing","ai-analytics","analytics","cdp","data-warehouse","experiments","feature-flags","javascript","product-analytics","python","react","session-replay","surveys","typescript","web-analytics",{"repoUrl":24,"stars":23,"forks":27,"topics":45,"description":46},[29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"🦔 PostHog is an all-in-one developer platform for building successful products. We offer product analytics, web analytics, session replay, error tracking, feature flags, experimentation, surveys, data warehouse, a CDP, and an AI product assistant to help debug your code, ship features faster, and keep all your usage and customer data in one stack.","https:\u002F\u002Fgithub.com\u002FPostHog\u002Fposthog\u002Ftree\u002FHEAD\u002Fproducts\u002Fmcp_analytics\u002Fskills\u002Fexploring-mcp-tool-quality","---\nname: exploring-mcp-tool-quality\ndescription: >\n  Investigate the quality of PostHog MCP tool calls — error rates, latency,\n  reach, and which tools are failing or slow. Use when the user asks \"which\n  MCP tool has the highest error rate?\", \"what's the slowest tool?\", \"which\n  tools fail most often?\", \"how reliable is tool X?\", wants a tool-quality\n  matrix, or pastes an MCP analytics tool-quality \u002F dashboard URL and asks\n  what it shows.\n---\n\n# Exploring MCP tool quality\n\nAny MCP server instrumented with PostHog's MCP analytics SDK emits a\n`$mcp_tool_call` event on the shared `events` table every time an agent invokes a\ntool. There is **no dedicated ClickHouse table** — every field lives as a\n`$mcp_*` property on `events`, and every tool-quality metric (error rate, latency\npercentiles, reach) is an aggregation over this one event. This is the data\nbehind the MCP analytics dashboard and tool-quality screens.\n\n**For a single tool, prefer the typed tools** — `posthog:query-mcp-tool-stats` (calls,\nerrors, p50\u002Fp95, users, sessions, intents), `posthog:query-mcp-tool-failures` (top error\nmessages by harness), and `posthog:query-mcp-tool-daily-stats` (day-by-day trend). Each\ntakes a `toolName` + `dateRange`, runs the same query runner as the tool-detail\nUI, and is gated behind the `mcp-analytics` flag — no hand-written SQL needed.\n\n**HogQL via `posthog:execute-sql` is the path for cross-tool questions** — the\n\"which tool errors most\" ranking below has no typed tool, so rank with SQL, then\ndrill into the worst tool with `posthog:query-mcp-tool-stats` and\n`posthog:query-mcp-tool-failures`. The full\nproperty schema and the canonical query recipes live in the shared MCP data\nreference:\n[`products\u002Fposthog_ai\u002Fskills\u002Fquerying-posthog-data\u002Freferences\u002Fmodels-mcp.md`](..\u002F..\u002F..\u002Fposthog_ai\u002Fskills\u002Fquerying-posthog-data\u002Freferences\u002Fmodels-mcp.md).\nThat reference is the single source of truth for the `$mcp_*` schema and the\neffective-tool-name idiom used below — this skill inlines only the headline\n\"which tool errors most\" query for convenience; pull the matrix, latency, and\nharness recipes from the reference rather than re-deriving them. Read it before\nwriting queries.\n\n## The two rules that matter most\n\n- **Always use the effective tool name.** New-SDK events wrap the real tool in\n  a single-exec call, so grouping on raw `$mcp_tool_name` collapses everything\n  under the wrapper. Use:\n\n  ```sql\n  coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name))\n  ```\n\n- **Always read `$mcp_is_error` via `toBool(...)`** and cast\n  `$mcp_duration_ms` via `toFloat(...)`. The properties are strings.\n\nAlways set a time range — these queries scan `events` otherwise.\n\n## Workflow: which tool has the highest error rate\n\nThis is the canonical \"which tool errors most\" question. Rank tools by error\nrate, but guard against small-sample noise with a `HAVING` floor on call volume:\n\n```sql\nposthog:execute-sql\nSELECT\n    coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) AS tool,\n    count() AS total_calls,\n    countIf(toBool(properties.$mcp_is_error)) AS errors,\n    round(countIf(toBool(properties.$mcp_is_error)) * 100.0 \u002F count(), 1) AS error_rate_pct\nFROM events\nWHERE event = '$mcp_tool_call'\n    AND coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) != ''\n    AND timestamp >= now() - INTERVAL 30 DAY\nGROUP BY tool\nHAVING total_calls >= 20\nORDER BY error_rate_pct DESC, total_calls DESC\nLIMIT 20\n```\n\nReport both **rate and volume** — a 100% error rate over 3 calls is rarely the\nreal story; a 12% rate over 50,000 calls is. Offer to pull the top\n`$mcp_error_message` values for the worst tool (see below).\n\n## Workflow: tool-quality matrix\n\nOne row per tool with error rate, latency percentiles, and reach — mirrors the\ntool-quality screen. The ready-to-run query is in\n[models-mcp.md](..\u002F..\u002F..\u002Fposthog_ai\u002Fskills\u002Fquerying-posthog-data\u002Freferences\u002Fmodels-mcp.md)\nunder \"Tool-quality matrix\".\n\n## Workflow: why is a tool failing\n\nFor one tool's top failure buckets (grouped by harness), call\n`posthog:query-mcp-tool-failures` with the `toolName` — it's the typed equivalent of the\nquery below. Failures come from the **same source as the error rate**: errored\n`$mcp_tool_call` events (`$mcp_is_error`), scoped by the effective tool name. Failures are\ngrouped by `$mcp_error_type` (a semantic bucket: `internal`, `validation`, `api_4xx`,\n`api_5xx`, `permission`, `timeout`, `rate_limited`, `missing_context`) and the HTTP\n`$mcp_error_status` when present. To see individual errored calls inside a bucket — with\nthe captured `$mcp_error_message`, session id, harness, and intent — pass the bucket's raw\n`error_type`\u002F`error_status` to `posthog:query-mcp-tool-failure-occurrences`\n(`$mcp_error_message` is empty on events captured before message capture shipped):\n\n```sql\nposthog:execute-sql\nSELECT\n    concat(\n        coalesce(nullIf(toString(properties.$mcp_error_type), ''), 'unknown'),\n        if(empty(coalesce(toString(properties.$mcp_error_status), '')), '',\n           concat(' (HTTP ', coalesce(toString(properties.$mcp_error_status), ''), ')'))\n    ) AS failure,\n    count() AS n\nFROM events\nWHERE event = '$mcp_tool_call'\n    AND toBool(properties.$mcp_is_error)\n    AND coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) = '\u003Ctool>'\n    AND timestamp >= now() - INTERVAL 30 DAY\nGROUP BY failure ORDER BY n DESC LIMIT 10\n```\n\n`$mcp_error_type` is only populated on newer SDK\u002Fserver paths — a chunk of errored calls\ncarry neither type nor status and fall into the `unknown` bucket.\n\n## Workflow: slowest tools\n\nSwap the aggregate for latency percentiles\n(`quantile(0.95)(toFloat(properties.$mcp_duration_ms))`) and order by `p95_ms`.\nThe matrix query already returns `p50_ms` \u002F `p95_ms`.\n\n## Constructing UI links\n\n- **Dashboard**: `https:\u002F\u002Fapp.posthog.com\u002Fproject\u002F\u003Cproject_id>\u002Fmcp-analytics\u002Fdashboard`\n- **Tool quality**: `https:\u002F\u002Fapp.posthog.com\u002Fproject\u002F\u003Cproject_id>\u002Fmcp-analytics\u002Ftool-quality`\n\nAlways surface a UI link so the user can verify visually.\n\n## Tips\n\n- Report error rate **and** call volume together; a `HAVING total_calls >= N`\n  floor stops tools with very few calls from topping the list spuriously\n- Exclude errored calls from latency percentiles only when asked — failed calls\n  are often the slow ones, and dropping them hides the problem\n- `$mcp_client_name` lets you cut quality by harness (Claude Code vs Cursor vs\n  …); the canonical bucketing `multiIf` is in\n  [models-mcp.md](..\u002F..\u002F..\u002Fposthog_ai\u002Fskills\u002Fquerying-posthog-data\u002Freferences\u002Fmodels-mcp.md)\n- Harness bucketing is resolved **server-side** by\n  `products\u002Fmcp_analytics\u002Fbackend\u002Fmcp_harness.py` — that's the source of truth,\n  and `posthog:query-mcp-harness-breakdown` runs it. If your hand-written SQL\n  disagrees with the screen, your bucketing has drifted from `mcp_harness.py`;\n  prefer the typed tool over re-deriving it\n\n## Related skills\n\n- [`exploring-mcp-sessions`](..\u002Fexploring-mcp-sessions\u002FSKILL.md) — drill into a\n  single agent run and its tool sequence\n- [`exploring-mcp-intent-clusters`](..\u002Fexploring-mcp-intent-clusters\u002FSKILL.md) —\n  group agent goals and see which intents drive the errors\n",{"data":50,"body":51},{"name":4,"description":6},{"type":52,"children":53},"root",[54,62,108,166,218,225,308,320,326,339,471,491,497,509,515,668,782,800,806,842,848,882,887,893,980,986,1019],{"type":55,"tag":56,"props":57,"children":58},"element","h1",{"id":4},[59],{"type":60,"value":61},"text","Exploring MCP tool quality",{"type":55,"tag":63,"props":64,"children":65},"p",{},[66,68,75,77,83,85,91,93,99,101,106],{"type":60,"value":67},"Any MCP server instrumented with PostHog's MCP analytics SDK emits a\n",{"type":55,"tag":69,"props":70,"children":72},"code",{"className":71},[],[73],{"type":60,"value":74},"$mcp_tool_call",{"type":60,"value":76}," event on the shared ",{"type":55,"tag":69,"props":78,"children":80},{"className":79},[],[81],{"type":60,"value":82},"events",{"type":60,"value":84}," table every time an agent invokes a\ntool. There is ",{"type":55,"tag":86,"props":87,"children":88},"strong",{},[89],{"type":60,"value":90},"no dedicated ClickHouse table",{"type":60,"value":92}," — every field lives as a\n",{"type":55,"tag":69,"props":94,"children":96},{"className":95},[],[97],{"type":60,"value":98},"$mcp_*",{"type":60,"value":100}," property on ",{"type":55,"tag":69,"props":102,"children":104},{"className":103},[],[105],{"type":60,"value":82},{"type":60,"value":107},", and every tool-quality metric (error rate, latency\npercentiles, reach) is an aggregation over this one event. This is the data\nbehind the MCP analytics dashboard and tool-quality screens.",{"type":55,"tag":63,"props":109,"children":110},{},[111,116,118,124,126,132,134,140,142,148,150,156,158,164],{"type":55,"tag":86,"props":112,"children":113},{},[114],{"type":60,"value":115},"For a single tool, prefer the typed tools",{"type":60,"value":117}," — ",{"type":55,"tag":69,"props":119,"children":121},{"className":120},[],[122],{"type":60,"value":123},"posthog:query-mcp-tool-stats",{"type":60,"value":125}," (calls,\nerrors, p50\u002Fp95, users, sessions, intents), ",{"type":55,"tag":69,"props":127,"children":129},{"className":128},[],[130],{"type":60,"value":131},"posthog:query-mcp-tool-failures",{"type":60,"value":133}," (top error\nmessages by harness), and ",{"type":55,"tag":69,"props":135,"children":137},{"className":136},[],[138],{"type":60,"value":139},"posthog:query-mcp-tool-daily-stats",{"type":60,"value":141}," (day-by-day trend). Each\ntakes a ",{"type":55,"tag":69,"props":143,"children":145},{"className":144},[],[146],{"type":60,"value":147},"toolName",{"type":60,"value":149}," + ",{"type":55,"tag":69,"props":151,"children":153},{"className":152},[],[154],{"type":60,"value":155},"dateRange",{"type":60,"value":157},", runs the same query runner as the tool-detail\nUI, and is gated behind the ",{"type":55,"tag":69,"props":159,"children":161},{"className":160},[],[162],{"type":60,"value":163},"mcp-analytics",{"type":60,"value":165}," flag — no hand-written SQL needed.",{"type":55,"tag":63,"props":167,"children":168},{},[169,182,184,189,191,196,198,209,211,216],{"type":55,"tag":86,"props":170,"children":171},{},[172,174,180],{"type":60,"value":173},"HogQL via ",{"type":55,"tag":69,"props":175,"children":177},{"className":176},[],[178],{"type":60,"value":179},"posthog:execute-sql",{"type":60,"value":181}," is the path for cross-tool questions",{"type":60,"value":183}," — the\n\"which tool errors most\" ranking below has no typed tool, so rank with SQL, then\ndrill into the worst tool with ",{"type":55,"tag":69,"props":185,"children":187},{"className":186},[],[188],{"type":60,"value":123},{"type":60,"value":190}," and\n",{"type":55,"tag":69,"props":192,"children":194},{"className":193},[],[195],{"type":60,"value":131},{"type":60,"value":197},". The full\nproperty schema and the canonical query recipes live in the shared MCP data\nreference:\n",{"type":55,"tag":199,"props":200,"children":202},"a",{"href":201},"..\u002F..\u002F..\u002Fposthog_ai\u002Fskills\u002Fquerying-posthog-data\u002Freferences\u002Fmodels-mcp.md",[203],{"type":55,"tag":69,"props":204,"children":206},{"className":205},[],[207],{"type":60,"value":208},"products\u002Fposthog_ai\u002Fskills\u002Fquerying-posthog-data\u002Freferences\u002Fmodels-mcp.md",{"type":60,"value":210},".\nThat reference is the single source of truth for the ",{"type":55,"tag":69,"props":212,"children":214},{"className":213},[],[215],{"type":60,"value":98},{"type":60,"value":217}," schema and the\neffective-tool-name idiom used below — this skill inlines only the headline\n\"which tool errors most\" query for convenience; pull the matrix, latency, and\nharness recipes from the reference rather than re-deriving them. Read it before\nwriting queries.",{"type":55,"tag":219,"props":220,"children":222},"h2",{"id":221},"the-two-rules-that-matter-most",[223],{"type":60,"value":224},"The two rules that matter most",{"type":55,"tag":226,"props":227,"children":228},"ul",{},[229,269],{"type":55,"tag":230,"props":231,"children":232},"li",{},[233,238,240,246,248],{"type":55,"tag":86,"props":234,"children":235},{},[236],{"type":60,"value":237},"Always use the effective tool name.",{"type":60,"value":239}," New-SDK events wrap the real tool in\na single-exec call, so grouping on raw ",{"type":55,"tag":69,"props":241,"children":243},{"className":242},[],[244],{"type":60,"value":245},"$mcp_tool_name",{"type":60,"value":247}," collapses everything\nunder the wrapper. Use:",{"type":55,"tag":249,"props":250,"children":255},"pre",{"className":251,"code":252,"language":253,"meta":254,"style":254},"language-sql shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name))\n","sql","",[256],{"type":55,"tag":69,"props":257,"children":258},{"__ignoreMap":254},[259],{"type":55,"tag":260,"props":261,"children":264},"span",{"class":262,"line":263},"line",1,[265],{"type":55,"tag":260,"props":266,"children":267},{},[268],{"type":60,"value":252},{"type":55,"tag":230,"props":270,"children":271},{},[272,291,293,299,300,306],{"type":55,"tag":86,"props":273,"children":274},{},[275,277,283,285],{"type":60,"value":276},"Always read ",{"type":55,"tag":69,"props":278,"children":280},{"className":279},[],[281],{"type":60,"value":282},"$mcp_is_error",{"type":60,"value":284}," via ",{"type":55,"tag":69,"props":286,"children":288},{"className":287},[],[289],{"type":60,"value":290},"toBool(...)",{"type":60,"value":292}," and cast\n",{"type":55,"tag":69,"props":294,"children":296},{"className":295},[],[297],{"type":60,"value":298},"$mcp_duration_ms",{"type":60,"value":284},{"type":55,"tag":69,"props":301,"children":303},{"className":302},[],[304],{"type":60,"value":305},"toFloat(...)",{"type":60,"value":307},". The properties are strings.",{"type":55,"tag":63,"props":309,"children":310},{},[311,313,318],{"type":60,"value":312},"Always set a time range — these queries scan ",{"type":55,"tag":69,"props":314,"children":316},{"className":315},[],[317],{"type":60,"value":82},{"type":60,"value":319}," otherwise.",{"type":55,"tag":219,"props":321,"children":323},{"id":322},"workflow-which-tool-has-the-highest-error-rate",[324],{"type":60,"value":325},"Workflow: which tool has the highest error rate",{"type":55,"tag":63,"props":327,"children":328},{},[329,331,337],{"type":60,"value":330},"This is the canonical \"which tool errors most\" question. Rank tools by error\nrate, but guard against small-sample noise with a ",{"type":55,"tag":69,"props":332,"children":334},{"className":333},[],[335],{"type":60,"value":336},"HAVING",{"type":60,"value":338}," floor on call volume:",{"type":55,"tag":249,"props":340,"children":342},{"className":251,"code":341,"language":253,"meta":254,"style":254},"posthog:execute-sql\nSELECT\n    coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) AS tool,\n    count() AS total_calls,\n    countIf(toBool(properties.$mcp_is_error)) AS errors,\n    round(countIf(toBool(properties.$mcp_is_error)) * 100.0 \u002F count(), 1) AS error_rate_pct\nFROM events\nWHERE event = '$mcp_tool_call'\n    AND coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) != ''\n    AND timestamp >= now() - INTERVAL 30 DAY\nGROUP BY tool\nHAVING total_calls >= 20\nORDER BY error_rate_pct DESC, total_calls DESC\nLIMIT 20\n",[343],{"type":55,"tag":69,"props":344,"children":345},{"__ignoreMap":254},[346,354,363,372,381,390,399,408,417,426,435,444,453,462],{"type":55,"tag":260,"props":347,"children":348},{"class":262,"line":263},[349],{"type":55,"tag":260,"props":350,"children":351},{},[352],{"type":60,"value":353},"posthog:execute-sql\n",{"type":55,"tag":260,"props":355,"children":357},{"class":262,"line":356},2,[358],{"type":55,"tag":260,"props":359,"children":360},{},[361],{"type":60,"value":362},"SELECT\n",{"type":55,"tag":260,"props":364,"children":366},{"class":262,"line":365},3,[367],{"type":55,"tag":260,"props":368,"children":369},{},[370],{"type":60,"value":371},"    coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) AS tool,\n",{"type":55,"tag":260,"props":373,"children":375},{"class":262,"line":374},4,[376],{"type":55,"tag":260,"props":377,"children":378},{},[379],{"type":60,"value":380},"    count() AS total_calls,\n",{"type":55,"tag":260,"props":382,"children":384},{"class":262,"line":383},5,[385],{"type":55,"tag":260,"props":386,"children":387},{},[388],{"type":60,"value":389},"    countIf(toBool(properties.$mcp_is_error)) AS errors,\n",{"type":55,"tag":260,"props":391,"children":393},{"class":262,"line":392},6,[394],{"type":55,"tag":260,"props":395,"children":396},{},[397],{"type":60,"value":398},"    round(countIf(toBool(properties.$mcp_is_error)) * 100.0 \u002F count(), 1) AS error_rate_pct\n",{"type":55,"tag":260,"props":400,"children":402},{"class":262,"line":401},7,[403],{"type":55,"tag":260,"props":404,"children":405},{},[406],{"type":60,"value":407},"FROM events\n",{"type":55,"tag":260,"props":409,"children":411},{"class":262,"line":410},8,[412],{"type":55,"tag":260,"props":413,"children":414},{},[415],{"type":60,"value":416},"WHERE event = '$mcp_tool_call'\n",{"type":55,"tag":260,"props":418,"children":420},{"class":262,"line":419},9,[421],{"type":55,"tag":260,"props":422,"children":423},{},[424],{"type":60,"value":425},"    AND coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) != ''\n",{"type":55,"tag":260,"props":427,"children":429},{"class":262,"line":428},10,[430],{"type":55,"tag":260,"props":431,"children":432},{},[433],{"type":60,"value":434},"    AND timestamp >= now() - INTERVAL 30 DAY\n",{"type":55,"tag":260,"props":436,"children":438},{"class":262,"line":437},11,[439],{"type":55,"tag":260,"props":440,"children":441},{},[442],{"type":60,"value":443},"GROUP BY tool\n",{"type":55,"tag":260,"props":445,"children":447},{"class":262,"line":446},12,[448],{"type":55,"tag":260,"props":449,"children":450},{},[451],{"type":60,"value":452},"HAVING total_calls >= 20\n",{"type":55,"tag":260,"props":454,"children":456},{"class":262,"line":455},13,[457],{"type":55,"tag":260,"props":458,"children":459},{},[460],{"type":60,"value":461},"ORDER BY error_rate_pct DESC, total_calls DESC\n",{"type":55,"tag":260,"props":463,"children":465},{"class":262,"line":464},14,[466],{"type":55,"tag":260,"props":467,"children":468},{},[469],{"type":60,"value":470},"LIMIT 20\n",{"type":55,"tag":63,"props":472,"children":473},{},[474,476,481,483,489],{"type":60,"value":475},"Report both ",{"type":55,"tag":86,"props":477,"children":478},{},[479],{"type":60,"value":480},"rate and volume",{"type":60,"value":482}," — a 100% error rate over 3 calls is rarely the\nreal story; a 12% rate over 50,000 calls is. Offer to pull the top\n",{"type":55,"tag":69,"props":484,"children":486},{"className":485},[],[487],{"type":60,"value":488},"$mcp_error_message",{"type":60,"value":490}," values for the worst tool (see below).",{"type":55,"tag":219,"props":492,"children":494},{"id":493},"workflow-tool-quality-matrix",[495],{"type":60,"value":496},"Workflow: tool-quality matrix",{"type":55,"tag":63,"props":498,"children":499},{},[500,502,507],{"type":60,"value":501},"One row per tool with error rate, latency percentiles, and reach — mirrors the\ntool-quality screen. The ready-to-run query is in\n",{"type":55,"tag":199,"props":503,"children":504},{"href":201},[505],{"type":60,"value":506},"models-mcp.md",{"type":60,"value":508},"\nunder \"Tool-quality matrix\".",{"type":55,"tag":219,"props":510,"children":512},{"id":511},"workflow-why-is-a-tool-failing",[513],{"type":60,"value":514},"Workflow: why is a tool failing",{"type":55,"tag":63,"props":516,"children":517},{},[518,520,525,527,532,534,539,541,546,548,553,555,561,563,569,571,577,578,584,586,592,593,599,600,606,607,613,614,620,622,628,630,635,637,643,645,651,653,659,661,666],{"type":60,"value":519},"For one tool's top failure buckets (grouped by harness), call\n",{"type":55,"tag":69,"props":521,"children":523},{"className":522},[],[524],{"type":60,"value":131},{"type":60,"value":526}," with the ",{"type":55,"tag":69,"props":528,"children":530},{"className":529},[],[531],{"type":60,"value":147},{"type":60,"value":533}," — it's the typed equivalent of the\nquery below. Failures come from the ",{"type":55,"tag":86,"props":535,"children":536},{},[537],{"type":60,"value":538},"same source as the error rate",{"type":60,"value":540},": errored\n",{"type":55,"tag":69,"props":542,"children":544},{"className":543},[],[545],{"type":60,"value":74},{"type":60,"value":547}," events (",{"type":55,"tag":69,"props":549,"children":551},{"className":550},[],[552],{"type":60,"value":282},{"type":60,"value":554},"), scoped by the effective tool name. Failures are\ngrouped by ",{"type":55,"tag":69,"props":556,"children":558},{"className":557},[],[559],{"type":60,"value":560},"$mcp_error_type",{"type":60,"value":562}," (a semantic bucket: ",{"type":55,"tag":69,"props":564,"children":566},{"className":565},[],[567],{"type":60,"value":568},"internal",{"type":60,"value":570},", ",{"type":55,"tag":69,"props":572,"children":574},{"className":573},[],[575],{"type":60,"value":576},"validation",{"type":60,"value":570},{"type":55,"tag":69,"props":579,"children":581},{"className":580},[],[582],{"type":60,"value":583},"api_4xx",{"type":60,"value":585},",\n",{"type":55,"tag":69,"props":587,"children":589},{"className":588},[],[590],{"type":60,"value":591},"api_5xx",{"type":60,"value":570},{"type":55,"tag":69,"props":594,"children":596},{"className":595},[],[597],{"type":60,"value":598},"permission",{"type":60,"value":570},{"type":55,"tag":69,"props":601,"children":603},{"className":602},[],[604],{"type":60,"value":605},"timeout",{"type":60,"value":570},{"type":55,"tag":69,"props":608,"children":610},{"className":609},[],[611],{"type":60,"value":612},"rate_limited",{"type":60,"value":570},{"type":55,"tag":69,"props":615,"children":617},{"className":616},[],[618],{"type":60,"value":619},"missing_context",{"type":60,"value":621},") and the HTTP\n",{"type":55,"tag":69,"props":623,"children":625},{"className":624},[],[626],{"type":60,"value":627},"$mcp_error_status",{"type":60,"value":629}," when present. To see individual errored calls inside a bucket — with\nthe captured ",{"type":55,"tag":69,"props":631,"children":633},{"className":632},[],[634],{"type":60,"value":488},{"type":60,"value":636},", session id, harness, and intent — pass the bucket's raw\n",{"type":55,"tag":69,"props":638,"children":640},{"className":639},[],[641],{"type":60,"value":642},"error_type",{"type":60,"value":644},"\u002F",{"type":55,"tag":69,"props":646,"children":648},{"className":647},[],[649],{"type":60,"value":650},"error_status",{"type":60,"value":652}," to ",{"type":55,"tag":69,"props":654,"children":656},{"className":655},[],[657],{"type":60,"value":658},"posthog:query-mcp-tool-failure-occurrences",{"type":60,"value":660},"\n(",{"type":55,"tag":69,"props":662,"children":664},{"className":663},[],[665],{"type":60,"value":488},{"type":60,"value":667}," is empty on events captured before message capture shipped):",{"type":55,"tag":249,"props":669,"children":671},{"className":251,"code":670,"language":253,"meta":254,"style":254},"posthog:execute-sql\nSELECT\n    concat(\n        coalesce(nullIf(toString(properties.$mcp_error_type), ''), 'unknown'),\n        if(empty(coalesce(toString(properties.$mcp_error_status), '')), '',\n           concat(' (HTTP ', coalesce(toString(properties.$mcp_error_status), ''), ')'))\n    ) AS failure,\n    count() AS n\nFROM events\nWHERE event = '$mcp_tool_call'\n    AND toBool(properties.$mcp_is_error)\n    AND coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) = '\u003Ctool>'\n    AND timestamp >= now() - INTERVAL 30 DAY\nGROUP BY failure ORDER BY n DESC LIMIT 10\n",[672],{"type":55,"tag":69,"props":673,"children":674},{"__ignoreMap":254},[675,682,689,697,705,713,721,729,737,744,751,759,767,774],{"type":55,"tag":260,"props":676,"children":677},{"class":262,"line":263},[678],{"type":55,"tag":260,"props":679,"children":680},{},[681],{"type":60,"value":353},{"type":55,"tag":260,"props":683,"children":684},{"class":262,"line":356},[685],{"type":55,"tag":260,"props":686,"children":687},{},[688],{"type":60,"value":362},{"type":55,"tag":260,"props":690,"children":691},{"class":262,"line":365},[692],{"type":55,"tag":260,"props":693,"children":694},{},[695],{"type":60,"value":696},"    concat(\n",{"type":55,"tag":260,"props":698,"children":699},{"class":262,"line":374},[700],{"type":55,"tag":260,"props":701,"children":702},{},[703],{"type":60,"value":704},"        coalesce(nullIf(toString(properties.$mcp_error_type), ''), 'unknown'),\n",{"type":55,"tag":260,"props":706,"children":707},{"class":262,"line":383},[708],{"type":55,"tag":260,"props":709,"children":710},{},[711],{"type":60,"value":712},"        if(empty(coalesce(toString(properties.$mcp_error_status), '')), '',\n",{"type":55,"tag":260,"props":714,"children":715},{"class":262,"line":392},[716],{"type":55,"tag":260,"props":717,"children":718},{},[719],{"type":60,"value":720},"           concat(' (HTTP ', coalesce(toString(properties.$mcp_error_status), ''), ')'))\n",{"type":55,"tag":260,"props":722,"children":723},{"class":262,"line":401},[724],{"type":55,"tag":260,"props":725,"children":726},{},[727],{"type":60,"value":728},"    ) AS failure,\n",{"type":55,"tag":260,"props":730,"children":731},{"class":262,"line":410},[732],{"type":55,"tag":260,"props":733,"children":734},{},[735],{"type":60,"value":736},"    count() AS n\n",{"type":55,"tag":260,"props":738,"children":739},{"class":262,"line":419},[740],{"type":55,"tag":260,"props":741,"children":742},{},[743],{"type":60,"value":407},{"type":55,"tag":260,"props":745,"children":746},{"class":262,"line":428},[747],{"type":55,"tag":260,"props":748,"children":749},{},[750],{"type":60,"value":416},{"type":55,"tag":260,"props":752,"children":753},{"class":262,"line":437},[754],{"type":55,"tag":260,"props":755,"children":756},{},[757],{"type":60,"value":758},"    AND toBool(properties.$mcp_is_error)\n",{"type":55,"tag":260,"props":760,"children":761},{"class":262,"line":446},[762],{"type":55,"tag":260,"props":763,"children":764},{},[765],{"type":60,"value":766},"    AND coalesce(nullIf(toString(properties.$mcp_exec_tool_call_name), ''), toString(properties.$mcp_tool_name)) = '\u003Ctool>'\n",{"type":55,"tag":260,"props":768,"children":769},{"class":262,"line":455},[770],{"type":55,"tag":260,"props":771,"children":772},{},[773],{"type":60,"value":434},{"type":55,"tag":260,"props":775,"children":776},{"class":262,"line":464},[777],{"type":55,"tag":260,"props":778,"children":779},{},[780],{"type":60,"value":781},"GROUP BY failure ORDER BY n DESC LIMIT 10\n",{"type":55,"tag":63,"props":783,"children":784},{},[785,790,792,798],{"type":55,"tag":69,"props":786,"children":788},{"className":787},[],[789],{"type":60,"value":560},{"type":60,"value":791}," is only populated on newer SDK\u002Fserver paths — a chunk of errored calls\ncarry neither type nor status and fall into the ",{"type":55,"tag":69,"props":793,"children":795},{"className":794},[],[796],{"type":60,"value":797},"unknown",{"type":60,"value":799}," bucket.",{"type":55,"tag":219,"props":801,"children":803},{"id":802},"workflow-slowest-tools",[804],{"type":60,"value":805},"Workflow: slowest tools",{"type":55,"tag":63,"props":807,"children":808},{},[809,811,817,819,825,827,833,835,840],{"type":60,"value":810},"Swap the aggregate for latency percentiles\n(",{"type":55,"tag":69,"props":812,"children":814},{"className":813},[],[815],{"type":60,"value":816},"quantile(0.95)(toFloat(properties.$mcp_duration_ms))",{"type":60,"value":818},") and order by ",{"type":55,"tag":69,"props":820,"children":822},{"className":821},[],[823],{"type":60,"value":824},"p95_ms",{"type":60,"value":826},".\nThe matrix query already returns ",{"type":55,"tag":69,"props":828,"children":830},{"className":829},[],[831],{"type":60,"value":832},"p50_ms",{"type":60,"value":834}," \u002F ",{"type":55,"tag":69,"props":836,"children":838},{"className":837},[],[839],{"type":60,"value":824},{"type":60,"value":841},".",{"type":55,"tag":219,"props":843,"children":845},{"id":844},"constructing-ui-links",[846],{"type":60,"value":847},"Constructing UI links",{"type":55,"tag":226,"props":849,"children":850},{},[851,867],{"type":55,"tag":230,"props":852,"children":853},{},[854,859,861],{"type":55,"tag":86,"props":855,"children":856},{},[857],{"type":60,"value":858},"Dashboard",{"type":60,"value":860},": ",{"type":55,"tag":69,"props":862,"children":864},{"className":863},[],[865],{"type":60,"value":866},"https:\u002F\u002Fapp.posthog.com\u002Fproject\u002F\u003Cproject_id>\u002Fmcp-analytics\u002Fdashboard",{"type":55,"tag":230,"props":868,"children":869},{},[870,875,876],{"type":55,"tag":86,"props":871,"children":872},{},[873],{"type":60,"value":874},"Tool quality",{"type":60,"value":860},{"type":55,"tag":69,"props":877,"children":879},{"className":878},[],[880],{"type":60,"value":881},"https:\u002F\u002Fapp.posthog.com\u002Fproject\u002F\u003Cproject_id>\u002Fmcp-analytics\u002Ftool-quality",{"type":55,"tag":63,"props":883,"children":884},{},[885],{"type":60,"value":886},"Always surface a UI link so the user can verify visually.",{"type":55,"tag":219,"props":888,"children":890},{"id":889},"tips",[891],{"type":60,"value":892},"Tips",{"type":55,"tag":226,"props":894,"children":895},{},[896,916,921,944],{"type":55,"tag":230,"props":897,"children":898},{},[899,901,906,908,914],{"type":60,"value":900},"Report error rate ",{"type":55,"tag":86,"props":902,"children":903},{},[904],{"type":60,"value":905},"and",{"type":60,"value":907}," call volume together; a ",{"type":55,"tag":69,"props":909,"children":911},{"className":910},[],[912],{"type":60,"value":913},"HAVING total_calls >= N",{"type":60,"value":915},"\nfloor stops tools with very few calls from topping the list spuriously",{"type":55,"tag":230,"props":917,"children":918},{},[919],{"type":60,"value":920},"Exclude errored calls from latency percentiles only when asked — failed calls\nare often the slow ones, and dropping them hides the problem",{"type":55,"tag":230,"props":922,"children":923},{},[924,930,932,938,940],{"type":55,"tag":69,"props":925,"children":927},{"className":926},[],[928],{"type":60,"value":929},"$mcp_client_name",{"type":60,"value":931}," lets you cut quality by harness (Claude Code vs Cursor vs\n…); the canonical bucketing ",{"type":55,"tag":69,"props":933,"children":935},{"className":934},[],[936],{"type":60,"value":937},"multiIf",{"type":60,"value":939}," is in\n",{"type":55,"tag":199,"props":941,"children":942},{"href":201},[943],{"type":60,"value":506},{"type":55,"tag":230,"props":945,"children":946},{},[947,949,954,956,962,964,970,972,978],{"type":60,"value":948},"Harness bucketing is resolved ",{"type":55,"tag":86,"props":950,"children":951},{},[952],{"type":60,"value":953},"server-side",{"type":60,"value":955}," by\n",{"type":55,"tag":69,"props":957,"children":959},{"className":958},[],[960],{"type":60,"value":961},"products\u002Fmcp_analytics\u002Fbackend\u002Fmcp_harness.py",{"type":60,"value":963}," — that's the source of truth,\nand ",{"type":55,"tag":69,"props":965,"children":967},{"className":966},[],[968],{"type":60,"value":969},"posthog:query-mcp-harness-breakdown",{"type":60,"value":971}," runs it. If your hand-written SQL\ndisagrees with the screen, your bucketing has drifted from ",{"type":55,"tag":69,"props":973,"children":975},{"className":974},[],[976],{"type":60,"value":977},"mcp_harness.py",{"type":60,"value":979},";\nprefer the typed tool over re-deriving it",{"type":55,"tag":219,"props":981,"children":983},{"id":982},"related-skills",[984],{"type":60,"value":985},"Related skills",{"type":55,"tag":226,"props":987,"children":988},{},[989,1004],{"type":55,"tag":230,"props":990,"children":991},{},[992,1002],{"type":55,"tag":199,"props":993,"children":995},{"href":994},"..\u002Fexploring-mcp-sessions\u002FSKILL.md",[996],{"type":55,"tag":69,"props":997,"children":999},{"className":998},[],[1000],{"type":60,"value":1001},"exploring-mcp-sessions",{"type":60,"value":1003}," — drill into a\nsingle agent run and its tool sequence",{"type":55,"tag":230,"props":1005,"children":1006},{},[1007,1017],{"type":55,"tag":199,"props":1008,"children":1010},{"href":1009},"..\u002Fexploring-mcp-intent-clusters\u002FSKILL.md",[1011],{"type":55,"tag":69,"props":1012,"children":1014},{"className":1013},[],[1015],{"type":60,"value":1016},"exploring-mcp-intent-clusters",{"type":60,"value":1018}," —\ngroup agent goals and see which intents drive the errors",{"type":55,"tag":1020,"props":1021,"children":1022},"style",{},[1023],{"type":60,"value":1024},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":1026,"total":1189},[1027,1041,1053,1065,1078,1091,1107,1122,1136,1151,1161,1179],{"slug":1028,"name":1028,"fn":1029,"description":1030,"org":1031,"tags":1032,"stars":23,"repoUrl":24,"updatedAt":1040},"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},[1033,1035,1038,1039],{"name":1034,"slug":31,"type":15},"Analytics",{"name":1036,"slug":1037,"type":15},"Cost Optimization","cost-optimization",{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-07-28T05:34:11.117757",{"slug":1042,"name":1042,"fn":1043,"description":1044,"org":1045,"tags":1046,"stars":23,"repoUrl":24,"updatedAt":1052},"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},[1047,1048,1051],{"name":1034,"slug":31,"type":15},{"name":1049,"slug":1050,"type":15},"Audit","audit",{"name":9,"slug":8,"type":15},"2026-06-08T08:08:33.693989",{"slug":1054,"name":1054,"fn":1055,"description":1056,"org":1057,"tags":1058,"stars":23,"repoUrl":24,"updatedAt":1064},"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},[1059,1060,1062,1063],{"name":1049,"slug":1050,"type":15},{"name":1061,"slug":33,"type":15},"Data Warehouse",{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-06-18T08:22:57.67984",{"slug":1066,"name":1066,"fn":1067,"description":1068,"org":1069,"tags":1070,"stars":23,"repoUrl":24,"updatedAt":1077},"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},[1071,1072,1073,1076],{"name":1049,"slug":1050,"type":15},{"name":1061,"slug":33,"type":15},{"name":1074,"slug":1075,"type":15},"Performance","performance",{"name":9,"slug":8,"type":15},"2026-06-18T08:25:10.936787",{"slug":1079,"name":1079,"fn":1080,"description":1081,"org":1082,"tags":1083,"stars":23,"repoUrl":24,"updatedAt":1090},"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},[1084,1087,1088,1089],{"name":1085,"slug":1086,"type":15},"Alerting","alerting",{"name":21,"slug":22,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-06-18T08:24:40.318583",{"slug":1092,"name":1092,"fn":1093,"description":1094,"org":1095,"tags":1096,"stars":23,"repoUrl":24,"updatedAt":1106},"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},[1097,1098,1101,1102,1105],{"name":1034,"slug":31,"type":15},{"name":1099,"slug":1100,"type":15},"Monitoring","monitoring",{"name":13,"slug":14,"type":15},{"name":1103,"slug":1104,"type":15},"Operations","operations",{"name":9,"slug":8,"type":15},"2026-07-18T05:10:54.430898",{"slug":1108,"name":1108,"fn":1109,"description":1110,"org":1111,"tags":1112,"stars":23,"repoUrl":24,"updatedAt":1121},"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},[1113,1116,1117,1118],{"name":1114,"slug":1115,"type":15},"Automation","automation",{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":1119,"slug":1120,"type":15},"Workflow Automation","workflow-automation","2026-07-28T05:34:12.167015",{"slug":1123,"name":1123,"fn":1124,"description":1125,"org":1126,"tags":1127,"stars":23,"repoUrl":24,"updatedAt":1135},"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},[1128,1129,1130,1133,1134],{"name":1034,"slug":31,"type":15},{"name":21,"slug":22,"type":15},{"name":1131,"slug":1132,"type":15},"Frontend","frontend",{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-05-07T05:56:19.828048",{"slug":1137,"name":1137,"fn":1138,"description":1139,"org":1140,"tags":1141,"stars":23,"repoUrl":24,"updatedAt":1150},"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},[1142,1145,1146,1147],{"name":1143,"slug":1144,"type":15},"API Development","api-development",{"name":1131,"slug":1132,"type":15},{"name":9,"slug":8,"type":15},{"name":1148,"slug":1149,"type":15},"SDK","sdk","2026-06-08T08:08:34.929454",{"slug":1152,"name":1152,"fn":1153,"description":1154,"org":1155,"tags":1156,"stars":23,"repoUrl":24,"updatedAt":1160},"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},[1157,1158,1159],{"name":1143,"slug":1144,"type":15},{"name":1103,"slug":1104,"type":15},{"name":9,"slug":8,"type":15},"2026-07-15T05:29:58.442727",{"slug":1162,"name":1162,"fn":1163,"description":1164,"org":1165,"tags":1166,"stars":23,"repoUrl":24,"updatedAt":1178},"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},[1167,1168,1171,1172,1175],{"name":1114,"slug":1115,"type":15},{"name":1169,"slug":1170,"type":15},"Email","email",{"name":9,"slug":8,"type":15},{"name":1173,"slug":1174,"type":15},"Reporting","reporting",{"name":1176,"slug":1177,"type":15},"Slack","slack","2026-06-09T07:32:27.935712",{"slug":1180,"name":1180,"fn":1181,"description":1182,"org":1183,"tags":1184,"stars":23,"repoUrl":24,"updatedAt":1188},"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},[1185,1186,1187],{"name":1034,"slug":31,"type":15},{"name":1143,"slug":1144,"type":15},{"name":9,"slug":8,"type":15},"2026-06-08T08:08:29.624498",231,{"items":1191,"total":1241},[1192,1199,1205,1212,1219,1226,1234],{"slug":1028,"name":1028,"fn":1029,"description":1030,"org":1193,"tags":1194,"stars":23,"repoUrl":24,"updatedAt":1040},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1195,1196,1197,1198],{"name":1034,"slug":31,"type":15},{"name":1036,"slug":1037,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},{"slug":1042,"name":1042,"fn":1043,"description":1044,"org":1200,"tags":1201,"stars":23,"repoUrl":24,"updatedAt":1052},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1202,1203,1204],{"name":1034,"slug":31,"type":15},{"name":1049,"slug":1050,"type":15},{"name":9,"slug":8,"type":15},{"slug":1054,"name":1054,"fn":1055,"description":1056,"org":1206,"tags":1207,"stars":23,"repoUrl":24,"updatedAt":1064},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1208,1209,1210,1211],{"name":1049,"slug":1050,"type":15},{"name":1061,"slug":33,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},{"slug":1066,"name":1066,"fn":1067,"description":1068,"org":1213,"tags":1214,"stars":23,"repoUrl":24,"updatedAt":1077},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1215,1216,1217,1218],{"name":1049,"slug":1050,"type":15},{"name":1061,"slug":33,"type":15},{"name":1074,"slug":1075,"type":15},{"name":9,"slug":8,"type":15},{"slug":1079,"name":1079,"fn":1080,"description":1081,"org":1220,"tags":1221,"stars":23,"repoUrl":24,"updatedAt":1090},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1222,1223,1224,1225],{"name":1085,"slug":1086,"type":15},{"name":21,"slug":22,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},{"slug":1092,"name":1092,"fn":1093,"description":1094,"org":1227,"tags":1228,"stars":23,"repoUrl":24,"updatedAt":1106},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1229,1230,1231,1232,1233],{"name":1034,"slug":31,"type":15},{"name":1099,"slug":1100,"type":15},{"name":13,"slug":14,"type":15},{"name":1103,"slug":1104,"type":15},{"name":9,"slug":8,"type":15},{"slug":1108,"name":1108,"fn":1109,"description":1110,"org":1235,"tags":1236,"stars":23,"repoUrl":24,"updatedAt":1121},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1237,1238,1239,1240],{"name":1114,"slug":1115,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":1119,"slug":1120,"type":15},61]