Gc Config Init

Set up a GitHub Copilot Coding Agent configuration for this project.

Step 0 — Recall learnings

If .github/copilot-learnings.md exists, read all entries and apply them silently to inform this run. The [skill-name] tag on each entry is provenance only — all entries apply regardless of which skill wrote them. Do not announce that learnings were loaded.

If the file does not exist, proceed without mention.

Step 1 — Gather context

Before creating any files:

1. If .github/copilot-instructions.md already exists, stop and suggest running /gc-config-optimize instead.
2. Scan for toolchain clues: package.json, Cargo.toml, go.mod, pyproject.toml, requirements.txt, Makefile, README.md, and similar.
3. Check if .github/hooks/hooks.json already exists (skip Step 2 if present).
4. Note which formatters are present: Prettier (.prettierrc*, prettier in package.json), ruff (ruff.toml, [tool.ruff] in pyproject.toml), rustfmt (rustfmt.toml), gofmt (any Go project), php-cs-fixer (.php-cs-fixer.php).
5. If no project description was provided in $ARGUMENTS and the toolchain cannot be determined, ask: what does this project produce and what stack is involved?

Step 2 — Create .github/hooks/hooks.json

Skip this step if .github/hooks/hooks.json already exists.

If a formatter was detected in Step 1, create .github/hooks/hooks.json:

{
  "version": 1,
  "hooks": {
    "postToolUse": [
      {
        "type": "command",
        "bash": ".github/hooks/scripts/format.sh",
        "cwd": ".",
        "timeoutSec": 30
      }
    ]
  }
}

Also create .github/hooks/scripts/format.sh. The script receives the hook payload as JSON on stdin — extract the edited file path from it:

#!/usr/bin/env bash
set -euo pipefail

INPUT=$(cat)
FILE=$(echo "$INPUT" | jq -r '.toolInput.path // .toolInput.file_path // empty')

if [[ -z "$FILE" || ! -f "$FILE" ]]; then
  exit 0
fi

Then append the formatter call for the detected ecosystem:

• JS/TS/Markdown (Prettier): npx prettier --write "$FILE" || true
• Python (ruff): ruff format "$FILE" || true
• Rust (rustfmt): rustfmt "$FILE" || true
• Go (gofmt): gofmt -w "$FILE" || true
• PHP (php-cs-fixer): php-cs-fixer fix "$FILE" || true

Make the script executable: run chmod +x .github/hooks/scripts/format.sh after creating it.

PostToolUse scripts must exit 0. They run after the tool completes and cannot block it — use || true on the formatter command so a formatter error never crashes the hook.

Optional — preToolUse guard for sensitive files:

Ask the user: "Would you like a preToolUse hook that blocks writes to .env and secrets/ files? This is the Copilot equivalent of permissions.deny."

If yes, add a preToolUse entry to hooks.json:

"preToolUse": [
  {
    "type": "command",
    "bash": ".github/hooks/scripts/guard.sh",
    "cwd": ".",
    "timeoutSec": 10
  }
]

And create .github/hooks/scripts/guard.sh:

#!/usr/bin/env bash
INPUT=$(cat)
TARGET=$(echo "$INPUT" | jq -r '.toolInput.path // .toolInput.file_path // empty')

if [[ -n "$TARGET" ]] && [[ "$TARGET" =~ (^|/)\\.env($|\\.) || "$TARGET" =~ (^|/)secrets/ ]]; then
  echo "Blocked: writes to .env and secrets/ are not allowed" >&2
  exit 1
fi
exit 0

Make the script executable: chmod +x .github/hooks/scripts/guard.sh

PreToolUse scripts block by exiting 1. The script reads a JSON payload on stdin with .toolName and .toolInput. Exiting 0 allows the tool call to proceed.

If no formatter was detected in Step 1, skip hooks.json creation and note it in the Step 7 summary.

Step 3 — Create .github/copilot-instructions.md

Create the file with this structure (keep total under ~8,000 characters):

# <Project Name>

<One-line description and stack summary.>

## Commands

- Build: `<command or TODO>`
- Test: `<command or TODO>`
- Lint: `<command or TODO>`

## Architecture

<Key directories and patterns — only non-obvious parts. Omit if too new to know.>

## Conventions

<Concrete rules that deviate from defaults or that Copilot commonly gets wrong.
Never standard language conventions. Omit if the project is too new.>

## Don't

- Don't commit secrets or credentials to git
- Don't use `--force` git flags — fix the underlying issue instead

Fill in what the toolchain scan revealed. Use TODO placeholders for commands that cannot be determined yet.

Step 4 — Create path-specific instruction files (optional)

Offer to create .github/instructions/*.instructions.md files when the project has distinct subsystems with different conventions (frontend/backend, tests, API layer). Each file requires an applyTo frontmatter field:

---
applyTo: "src/frontend/**"
---

<Frontend-specific conventions here.>

Do not create these if the project is too new or the subsystems are not yet clear. Explain that they can be created later.

Step 5 — Create copilot-setup-steps.yml (optional)

Create .github/workflows/copilot-setup-steps.yml only when a package manager or build tool is detected. The job name must be exactly copilot-setup-steps and runtime must stay under 59 minutes. Always include dependency caching — without it, every agent session reinstalls all dependencies from scratch.

Example for a Node.js project:

name: Copilot Setup Steps
on: workflow_call

jobs:
  copilot-setup-steps:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: "20"
          cache: "npm"
      - run: npm ci

Adapt caching to the detected ecosystem:

• Python: actions/setup-python@v5 with cache: 'pip'
• Go: actions/setup-go@v5 with cache: true
• Rust: actions/cache@v4 targeting ~/.cargo and target/

Step 6 — Create AGENTS.md (optional)

Create AGENTS.md only when there is evidence that other AI coding tools are used (.claude/, .gemini/, .codex/ directories, or Cursor config files present). AGENTS.md is the vendor-neutral standard read by Codex, Amp, Cursor, Copilot, and others.

Keep it focused on universal concerns: setup commands, architecture boundaries, code style, testing conventions, and safety rules. Do not include Copilot-specific syntax.

Step 7 — Summary

List every file created, note any TODO placeholders that still need filling in, and include:

• Fill in TODO commands once the build/test/lint commands are known.
• MCP servers: For Copilot CLI, MCP servers can be configured in ~/.copilot/mcp-config.json. For the Coding Agent, use GitHub repository Settings → Copilot → MCP servers.
• Run /gc-config-optimize once the project has more content for a full audit.
• Learnings: Project-specific discoveries are automatically stored in .github/copilot-learnings.md and recalled on future runs of either skill. Run /gc-config-optimize periodically to promote recurring patterns into the configuration.

Store learnings

Before responding, review this run against the notice criteria below. For each entry that qualifies, append one line to .github/copilot-learnings.md (create the file if it does not exist), tagged [gc-config-init]. Skip any entry that duplicates one already in the file. If nothing qualifies, do not create or modify the file — no user notification either way.

Qualifies:

• Something about this repo that differs from what this skill assumes on a generic project
• A suggestion the user explicitly accepted or rejected that deviates from skill defaults
• A constraint or fact discovered that would change how this skill behaves next time

Does not qualify:

• Standard skill behavior applied without deviation
• Facts already present in AGENTS.md, CLAUDE.md, or other config files
• Anything a reader could determine from the repo without this skill having run

Entry format:

[gc-config-init] <concise fact about this project>

Did this output meet your expectations? If not, describe what was off and Copilot will log the correction to .github/copilot-learnings.md.

Note: Learnings are automatically recalled on the next run of either skill. Run /gc-config-optimize periodically to promote recurring patterns into the configuration.