Simplify the focus of devops health workflows#521
Conversation
There was a problem hiding this comment.
Pull request overview
This PR narrows the DevOps health workflow suite to focus on repository infrastructure signals (pipelines, infra configuration, and resource usage), removing skill-quality and PR/review tracking from the orchestrator and investigation playbooks.
Changes:
- Removed “quality” and “PR” health checks/playbooks, updating workflow scope text and dispatch rules accordingly.
- Reduced required GitHub permissions/toolsets for the orchestrator and groomer to match the new infra-only focus.
- Updated the shared compiled knowledge (
devops-*.lock.md) to reflect the simplified categories and fingerprints.
Show a summary per file
| File | Description |
|---|---|
| docs/agentic-workflows.md | Updates docs to describe the simplified infra-focused health workflow scope/categories. |
| .github/workflows/devops-health-check.md | Refocuses the orchestrator workflow to pipeline/infra/resource signals; removes PR/quality scope and related permissions/toolsets. |
| .github/workflows/devops-health-investigate.md | Limits investigation worker inputs/playbook routing to pipeline/infra/resource categories. |
| .github/workflows/devops-health-groom.md | Removes PR permissions/toolsets now that PR/review health is out of scope. |
| .github/aw/shared/devops-investigate.lock.md | Removes quality/PR playbooks and strengthens pipeline/infra/resource investigation guidance. |
| .github/aw/shared/devops-health.lock.md | Removes quality/PR fingerprinting/severity rules and updates dispatch rules/budgeting for the reduced scope. |
Copilot's findings
Tip
Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comments suppressed due to low confidence (1)
docs/agentic-workflows.md:24
- The architecture diagram still says the orchestrator “Dispatches investigation workers (up to 10)”, but the updated dispatch rules now cap investigations at 2 per run. Please update this line to avoid conflicting guidance.
devops-health-check (Orchestrator) ─── runs daily
├─ Collects health signals from 3 categories:
│ Pipeline · Infrastructure · Resources
├─ Fingerprints each finding for stable diff tracking
├─ Classifies: 🆕 NEW · 📌 EXISTING · ✅ RESOLVED
├─ Updates pinned health dashboard issue
└─ Dispatches investigation workers (up to 10)
- Files reviewed: 6/6 changed files
- Comments generated: 4
There was a problem hiding this comment.
Pull request overview
This PR narrows the “DevOps health” agentic workflows to an infrastructure-focused scope (pipeline, infra configuration, and resource usage), removing skill-quality and PR-review health tracking. It also updates the shared playbooks/fingerprinting guidance and regenerates the compiled .lock.yml workflows accordingly.
Changes:
- Updated the health check + investigation documentation/playbooks to only cover
pipeline,infra, andresourcefindings (removing quality/PR categories). - Adjusted workflow permissions/toolsets for the orchestrator and groomer to drop PR access, and regenerated compiled lock files with the newer gh-aw compiler.
- Updated shared “compiled knowledge” docs (
devops-health.lock.md,devops-investigate.lock.md) to match the new scope and dispatch rules.
Show a summary per file
| File | Description |
|---|---|
| docs/agentic-workflows.md | Updates the high-level documentation to reflect infra-only health checks (but still has a stale dispatch limit reference). |
| .github/workflows/devops-health-check.md | Refocuses the orchestrator prompt and rules on pipeline/infra/resource signals; removes PR toolset usage. |
| .github/workflows/devops-health-check.lock.yml | Regenerated compiled workflow reflecting updated scope/toolsets and gh-aw version bump. |
| .github/workflows/devops-health-investigate.md | Updates worker workflow input semantics/docs to remove quality/PR categories. |
| .github/workflows/devops-health-investigate.lock.yml | Regenerated compiled worker workflow with updated metadata and toolchain versions. |
| .github/workflows/devops-health-groom.md | Removes PR permissions/toolset from grooming workflow. |
| .github/workflows/devops-health-groom.lock.yml | Regenerated compiled groom workflow reflecting updated toolsets and gh-aw version bump. |
| .github/workflows/close-stale-prs.agent.lock.yml | Regenerated compiled workflow due to compiler/toolchain updates. |
| .github/aw/shared/devops-investigate.lock.md | Removes quality/PR playbooks and refines pipeline/infra/resource investigation guidance. |
| .github/aw/shared/devops-health.lock.md | Removes quality/PR fingerprinting and dispatch rules; updates infra/resource catalog and dispatch budget guidance. |
| .github/aw/actions-lock.json | Updates the pinned action entries to include actions/github-script@v9 and github/gh-aw-actions/setup@v0.68.1. |
Copilot's findings
Tip
Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comments suppressed due to low confidence (1)
docs/agentic-workflows.md:24
- The architecture diagram still says the orchestrator dispatches investigation workers "(up to 10)", but the updated health check workflow now caps dispatches at 2 per run. Please update this line so the docs match the new dispatch budget/guardrail.
devops-health-check (Orchestrator) ─── runs daily
├─ Collects health signals from 3 categories:
│ Pipeline · Infrastructure · Resources
├─ Fingerprints each finding for stable diff tracking
├─ Classifies: 🆕 NEW · 📌 EXISTING · ✅ RESOLVED
├─ Updates pinned health dashboard issue
└─ Dispatches investigation workers (up to 10)
- Files reviewed: 11/11 changed files
- Comments generated: 1
| **First run note:** On the first run all findings are 🆕 NEW. This means ALL critical findings MUST be dispatched. | ||
|
|
||
| **Budget:** Maximum **2** dispatches per run (limited to avoid investigation runs cancelling each other due to a shared agent concurrency group — see [gh-aw#20187](https://github.com/github/gh-aw/issues/20187)). If more than 2 qualify, prioritize by: | ||
| 1. Severity descending (🔴 first) | ||
| 2. Pipeline findings first | ||
| 3. Quality findings second | ||
| 3. Infrastructure findings second |
There was a problem hiding this comment.
The dispatch budget here is capped at 2, but the workflow frontmatter still allows safe-outputs.dispatch-workflow.max: 5. Since safe-outputs is the enforcement layer, consider lowering the safe-output max to 2 as well so the workflow cannot dispatch more investigations than intended (and risk hitting the shared concurrency cancellation issue).
Motivation
The size and scope of the repo increased - so limiting the scope of the devops workflows to just the infra stuff