Configuration reference

---
name: test-specialist                  # optional; defaults to the file name
description: Improves test coverage without changing production code   # REQUIRED
target: github-copilot                 # vscode | github-copilot; omit for both
tools: ["read", "search", "edit", "execute", "github/*"]   # omit or ["*"] = all; [] = none
model: gpt-5.4                         # optional; inherits the default if unset
disable-model-invocation: false        # true = never auto-selected (and not callable as a subagent)
user-invocable: true                   # false = hidden from manual selection, programmatic only
mcp-servers:                           # this agent only; ignored by IDE custom agents
  coverage:
    type: local
    command: npx
    args: ["coverage-mcp@latest"]
    tools: ["get_report"]
metadata:
  owner: qa-platform
---

You are a testing specialist. Analyse coverage gaps and add unit and integration tests.
Only change files under tests/. Never delete or skip an existing test.

# .github/copilot-instructions.md — repository-wide, applied to every request
This is a Go service with a Ruby client.

## Build and test
- Build: make build
- Test:  make test      (run before every commit)
- Full CI check: make ci

## Conventions
- Run make fmt before committing.
- Changes under proto/ require make proto.

---
applyTo: "**/*.ts,**/*.tsx"
excludeAgent: "code-review"      # or "cloud-agent"; omit to apply to both
---

Use strict TypeScript. Prefer named exports. Every component needs a test.

---
name: github-actions-failure-debugging          # lowercase, hyphenated; matches the folder
description: Debug failing GitHub Actions workflows. Use when asked to fix a failing CI run.
allowed-tools: shell                            # optional pre-approval — trusted skills only
---

1. Use list_workflow_runs to find the failing run for the PR.
2. Use summarize_job_log_failures to understand what failed.
3. Reproduce the failure locally, fix it, and confirm before committing.

jobs:
  copilot-setup-steps:        # exact name required; file must be on the default branch
    runs-on: ubuntu-latest    # larger, self-hosted (ARC) or Windows labels allowed
    timeout-minutes: 30       # <= 59
    permissions:
      contents: read
    services:                 # e.g. a database for integration tests
      postgres:
        image: postgres:17
        env:
          POSTGRES_PASSWORD: postgres
    steps:
      - uses: actions/checkout@v7
      - uses: actions/setup-node@v7
        with:
          node-version: "24"
          cache: "npm"
      - run: npm ci

{
  "mcpServers": {
    "sentry": {
      "type": "local",                                   // local | stdio | http | sse
      "command": "npx",
      "args": ["@sentry/mcp-server@latest"],
      "tools": ["get_issue_details"],                     // REQUIRED — allowlist tools
      "env": { "SENTRY_ACCESS_TOKEN": "$COPILOT_MCP_SENTRY_TOKEN" }
    },
    "cloudflare": {
      "type": "sse",
      "url": "https://docs.mcp.cloudflare.com/sse",
      "tools": ["*"]
    }
  }
}

// ~/.copilot/mcp-config.json — Copilot CLI user configuration
{
  "mcpServers": {
    "github": {
      "type": "http",
      "url": "https://api.githubcopilot.com/mcp/readonly"
    }
  }
}

{
  "version": 1,
  "hooks": {
    "sessionStart":        [{ "type": "command", "bash": "./scripts/log-start.sh", "timeoutSec": 10 }],
    "userPromptSubmitted": [{ "type": "command", "bash": "./scripts/log-prompt.sh" }],
    "preToolUse":          [{ "type": "command", "matcher": "bash", "bash": "./scripts/policy.sh", "timeoutSec": 15 }],
    "postToolUse":         [{ "type": "http", "url": "https://audit.contoso.com/copilot" }],
    "errorOccurred":       [{ "type": "command", "bash": "./scripts/alert.sh" }],
    "sessionEnd":          [{ "type": "command", "bash": "./scripts/cleanup.sh", "timeoutSec": 60 }]
  }
}

{"permissionDecision": "deny", "permissionDecisionReason": "Force pushes are not allowed"}

---
on:
  schedule: daily                 # any GitHub Actions trigger
permissions:
  contents: read                  # read-only unless you grant more
  issues: read
  pull-requests: read
  copilot-requests: write         # bill Copilot usage to the organization via GITHUB_TOKEN
engine: copilot                   # default; claude, codex and gemini also supported
network: defaults                 # firewalled egress
tools:
  github:
    toolsets: [default]
safe-outputs:
  create-issue:                   # the ONLY write this workflow may perform
    title-prefix: "[repo-status] "
    labels: [report]
max-ai-credits: 300               # per-run cap (default 1,000 AIC)
---

# Daily repository status

Summarize the last 24 hours of issues, pull requests and commits.
Highlight blockers and recommend next steps. Keep it short and link everything.

gh extension install github/gh-aw     # install the extension
gh aw init                               # add authoring skills, instructions and an agent
gh aw compile                            # markdown -> hardened .lock.yml (commit both)
gh aw run daily-status                   # run a workflow
gh aw logs                               # recent runs: duration, tokens, AIC estimates
gh aw audit 1234567890                   # inspect one run in detail

on:
  pull_request:
    branches: [main]
  workflow_dispatch:

permissions:
  contents: read                              # least privilege by default

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

jobs:
  plan:
    if: github.event_name == 'pull_request'   # defensive gating: PR-only logic
    runs-on: ubuntu-latest
    outputs:
      risk: ${{ steps.plan.outputs.risk }}    # job output = contract for later jobs
    steps:
      - id: plan
        run: echo "risk=low" >> "$GITHUB_OUTPUT"   # step output

  deploy:
    needs: plan
    if: needs.plan.outputs.risk == 'low'
    runs-on: ubuntu-latest
    environment: production                   # required reviewers, scoped secrets
    env:
      REGION: ${{ vars.DEPLOY_REGION }}       # vars.* = centrally managed configuration
    steps:
      - run: echo "Deploying to $REGION"

<!-- .github/pull_request_template.md -->
## Plan (required)
- **Goal:**
- **Scope (paths):**
- **Steps:**
- **Success criteria:** [ ] required checks pass  [ ] security signals reviewed
- **Risks and mitigations:**
- **Rollback / escalation:**

## Evidence
- Workflow run(s):
- Scan results:

## Decisions and handoffs
- Decision / rationale / next owner: