CI/CD Integration Guide

Integrate prompts-gpt into your CI/CD pipeline to automate code reviews, security audits, quality checks, and more.

Prerequisites

1. A project token stored as a CI secret
2. At least one provider CLI installed in the CI environment
3. Node.js 18+ available
4. A passing prompts-gpt doctor --remote report before promotion

Recommended Order

Use this exact order on remote runners:

1. Install prompts-gpt and the provider CLI.
2. Inject PROMPTS_GPT_TOKEN, provider auth, and PROMPTS_GPT_NON_INTERACTIVE=1.
3. Run prompts-gpt doctor --remote.
4. Run prompts-gpt init --token-stdin.
5. Run prompts-gpt sync --agent all.
6. Start with prompts-gpt run --dry-run or a small --json run before heavier sweeps.

GitHub Actions

Basic Setup

name: AI Code Review
on:
  pull_request:
    branches: [main]

jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-node@v4
        with:
          node-version: 22

      - name: Install prompts-gpt
        run: npm install -g prompts-gpt

      - name: Install Codex CLI
        run: npm install -g @openai/codex

      - name: Remote readiness check
        env:
          PROMPTS_GPT_TOKEN: ${{ secrets.PROMPTS_GPT_TOKEN }}
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
          PROMPTS_GPT_NON_INTERACTIVE: "1"
        run: prompts-gpt doctor --remote

      - name: Initialize project
        run: printf '%s' "${{ secrets.PROMPTS_GPT_TOKEN }}" | prompts-gpt init --token-stdin

      - name: Sync prompts
        run: prompts-gpt sync

      - name: Run code review
        run: prompts-gpt run .prompts-gpt/code-review.md --provider codex --json
        env:
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}

Batch Review

      - name: Run all checks
        run: |
          prompts-gpt run-batch \
            .prompts-gpt/security-review.md \
            .prompts-gpt/performance-check.md \
            .prompts-gpt/accessibility-audit.md \
            --provider codex --json

Sweep Audit

      - name: Deep security audit
        run: |
          prompts-gpt sweep \
            .prompts-gpt/sweeps/security-audit.md \
            -n 3 --provider codex --json

Environment Variables

Token Management

Using environment variables

export PROMPTS_GPT_TOKEN=pgpt_abc123
prompts-gpt sync  # Uses env var

Using stdin (secure)

printf '%s' "$PROMPTS_GPT_TOKEN" | prompts-gpt init --token-stdin

Using init command

prompts-gpt init --token "$PROMPTS_GPT_TOKEN"

JSON Output for CI

All execution commands support --json for machine-readable output:

result=$(prompts-gpt run review.md --provider codex --json)
echo "$result" | jq '.ok'

Exit codes:

• 0 — Success
• 1 — Execution failed
• 2 — Configuration error

Caching

Cache the .prompts-gpt/ directory to speed up CI runs:

      - uses: actions/cache@v4
        with:
          path: .prompts-gpt
          key: prompts-gpt-${{ hashFiles('.prompts-gpt/manifest.json') }}

GitLab CI

code-review:
  image: node:22
  script:
    - npm install -g prompts-gpt @openai/codex
    - prompts-gpt init --token $PROMPTS_GPT_TOKEN
    - prompts-gpt sync
    - prompts-gpt run .prompts-gpt/code-review.md --provider codex --json
  variables:
    OPENAI_API_KEY: $OPENAI_API_KEY

Troubleshooting

Doctor check in CI

prompts-gpt doctor --remote --json

Common CI issues

1. Provider not found: Ensure the provider CLI is installed in the remote image
2. Token invalid: Check PROMPTS_GPT_TOKEN secret injection
3. Interactive prompt hangs: Set CI=true or PROMPTS_GPT_NON_INTERACTIVE=1
4. Timeout: Increase --timeout for sweep commands
5. Disk space: Prune artifacts aggressively and keep the artifacts directory out of git
6. Works locally, fails remotely: the remote runner is probably missing init --token-stdin, provider auth, or both

See Also

• Authentication
• Run CLI
• Doctor CLI