Repository navigation
Conversation
The posthog.com library pages for python, node, go, rust, ios, android, flutter, django and flask move to the capture v1 majors on each SDK's release day. The legacy integration variants now read copies taken on 2026-11-02 from context/docs-snapshots/legacy instead of the live pages, so they keep describing the majors they install. A docs_urls entry without a scheme is a repo file; a missing one fails the build. The generated references are byte-identical to today's build. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
🧙 Wizard CIRun the Wizard CI and test your changes against wizard-workbench example apps by replying with a GitHub comment using one of the following commands: Test all apps:
Test all apps in a directory:
Test an individual app:
Show more apps
Test against a wizard branch:
Add Results will be posted here when complete. |
The aggregated integration skills fetch the same pages; after the docs flip they would describe the v1 majors while a legacy-pinned run can still pick them from the integration category. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The error-tracking and feature-flags steps, the standalone error-tracking, feature-flags, AI observability and MCP analytics skills and the third omnibus skill fetched the same pages live. The dev server now watches the snapshots and rebuilds the skills that read an edited one. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
| @@ -1,5 +1,7 @@ | |||
| # Integration skills - example-based (code + docs) | |||
| # `framework` = the detection id a variant serves; `default: true` marks the family default. Both are emitted into skill-menu.json. | |||
| # Legacy SDK variants read `context/docs-snapshots/legacy/*.md` (copies of the library pages) instead of the live pages, | |||
There was a problem hiding this comment.
[nit] #453 also adds a comment at line 3 of this file, so whichever stack lands second will conflict here.
| tags: [android, java, kotlin] | ||
| docs_urls: | ||
| - https://posthog.com/docs/libraries/android.md | ||
| - context/docs-snapshots/legacy/android.md |
There was a problem hiding this comment.
[suggestion] #453 says an untagged variant serves either mode, so a v1 run could get these 3.x docs. Could we tag android, swift and flutter capture_mode: legacy?
| tags: [flutter, dart, mobile] | ||
| docs_urls: | ||
| - https://posthog.com/docs/libraries/flutter.md | ||
| - context/docs-snapshots/legacy/flutter.md |
There was a problem hiding this comment.
[question] #479 bumps the android/swift/flutter example apps to 4.0/6.0 while these variants read the 3.x/5.x snapshots, so agents would get 3.x docs next to 4.0 code. We could keep 3.x/5.x copies of the example apps for the legacy variants, or add capture_mode: v1 variants that read the live pages and the 4.0 examples. Which do we prefer?
|
Closing this stack. The design comes back after the agent DAG refactor, where the wizard picks each SDK's major, and the skills written for it, from project traits and context-mill tags. The branch stays. |
Problem
Many skills build their references from the live posthog.com library pages for python, node, go, rust, ios, android, flutter, django and flask. Those pages move to the capture v1 majors on each SDK's release day, and the next build would turn every legacy skill v1-shaped while the wizard still installs the legacy majors.
Changes
context/docs-snapshots/legacy/: copies of the thirteen pages, taken 2026-11-02, with a README listing each source and when to refresh.docs_urlsentry without a scheme is a repo file. A missing one fails the build. The dev server watches the snapshots and rebuilds the skills that read an edited one.Must be released before the docs pages change.
Test plan
Tests for the local read, the missing and out-of-repo paths, that every referenced snapshot exists, that no config fetches one of the pages live, and the file naming.
pnpm test.LLM context
Written with Claude Code (Claude Fable 5.1).
🤖 Generated with Claude Code