Index of every page on this site: llms.txt
All skills
Run it yourself

Emit BuildManifest telemetry to the portal after any non-trivial build operation. Invoke when you've completed a feature, fix, or refactor and need to notify the SystemStateEngine of what changed.

  • 1 file
  • 4 KB
  • Updated 3 months ago
  • GitHub
Use this Skill: https://skilld.dev/gh/colaberryintern/colaberryenterprise_ai_leadershipaccelerator/telemetry-emission

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈54 tokens for metadata: the name and description. ≈952 when used: this file.

Description uses 2.7% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈217 characters, or 2.7%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Telemetry Emission — BuildManifest Contract

When to invoke

After Claude Code completes a build (feature, fix, refactor) that changed files, schema, APIs, routes, or UI components. The portal consumes these manifests to maintain authoritative state maps. Without a manifest, the portal falls back to repo heuristics, which are less reliable.

The contract

Manifest schema: system/intelligence/manifests/build_manifest.schema.json Full contract: system/intelligence/contracts/BUILD_MANIFEST_CONTRACT.md

Required fields

  • manifest_version: "1.0", telemetry_version: "1.0"
  • task_id, project_id (UUIDs); bp_id if the build targeted a specific BP
  • execution_timestamp (ISO-8601)
  • Whatever changed (use the relevant arrays, omit empty):
    • files_created, files_modified, files_deleted
    • database_changes
    • apis_added, apis_modified
    • frontend_routes_added, ui_components_added, ui_components_modified
    • tests_added
    • validation_results (whichever ran: tsc, jest, playwright, build, lint)
    • dependencies_added
  • system_impacts (qualitative deltas)
  • Optionally: decision_trace (why the build chose this approach; powers the "Why is this next?" panel)

How to emit

POST to /api/portal/project/telemetry. The portal validates, ingests, and triggers a state rebuild — fire-and-forget.

Strict rules

  • Manifest authority. When telemetry exists, the engine prefers it over repo heuristics. Don't assume the engine will "figure it out from the file tree." Emit the manifest.
  • No secrets. The validator rejects AWS keys, GitHub PATs, OpenAI keys, JWTs, and private key blocks with HTTP 400. Redact before emitting.
  • No .. traversal in paths. Repo-relative POSIX only. The validator rejects leading / and ../.
  • Append-only. Never modify or delete an emitted manifest. To correct stale state, emit a new one with a later execution_timestamp — the resolver picks the later one as the winner.
  • Validation results are first-class. Emit validation_results whenever a check ran. Missing validation results contribute to the missing_validation_telemetry health score.
  • Database, API, UI changes flow through the manifest. Don't assume out-of-band documentation will catch up — the manifest IS the documentation.
  • Decision tracing. When the build's choices were non-obvious (alternative approach considered, trade-off accepted, dependency surfaced), include a decision_trace block, even a short one.

What the portal owns (do NOT duplicate)

  • The state graph: system/intelligence/state_graph.json — auto-maintained by graphSynchronizer. Manual edits are overwritten.
  • The DB map: system/database/database_map.json — auto-maintained by databaseSynchronizer.
  • The UI map: system/ui/ui_map.json — auto-maintained by uiSynchronizer.
  • The queue: read via GET /api/portal/project/system-state — never invent a parallel queue.

What Claude Code owns

  • BuildManifest emission for every non-trivial build.
  • Updating docs/sessions/CC-<id>.md (this session's human-readable log).
  • Updating directives in /directives when scope changes.
  • Updating CLAUDE.md itself for new operational rules.

Reading state

If you need to understand current project state during a build:

  • GET /api/portal/project/system-state — full state
  • GET /api/portal/project/system-state/explain/:taskId — "Why is this task next?"
  • GET /api/portal/project/telemetry — recent manifests
  • GET /api/portal/project/telemetry/health — health summary
  • GET /api/portal/project/graph — state graph
  • GET /api/portal/project/database-map / /ui-map — declared topology

Don't re-read the codebase to derive state when an endpoint already answers the question.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 7f168b8. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 3 months ago
user-invocable
true

README badge

README badge for colaberryintern/colaberryenterprise_ai_leadershipaccelerator/telemetry-emission