Every AI agent. Protected.

# Install and enable the PreToolUse hook (blocks dangerous installs):
pip install stillrunning
stillrunning --hook-claude-code

# Now every pip/npm install is checked before execution.
# Dangerous packages are BLOCKED. Suspicious packages show warnings.
# To remove: stillrunning --uninstall-hook-claude-code

# Add to .cursorrules in your project root:

Before installing any Python or npm package, check it first:
curl "https://stillrunning.io/api/check-package?name={package}"

Only proceed if verdict is CLEAN.
Block if DANGEROUS. Ask user if SUSPICIOUS.

# Add to ~/.claude/settings.json or equivalent:

{
  "mcpServers": {
    "stillrunning": {
      "type": "url",
      "url": "https://stillrunning.io/mcp",
      "name": "stillrunning"
    }
  }
}

# Add to .github/workflows/stillrunning.yml:

name: stillrunning Security
on: [push, pull_request]
jobs:
  security:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - uses: johhnyg/stillrunning-action@v2
        with:
          token: ${{ secrets.STILLRUNNING_TOKEN }}
          block_dangerous: true
          scan_imports: true

# Doc-only: requires user to configure intercept
pip install stillrunning && stillrunning --setup --autonomous
export STILLRUNNING_TOKEN=your_token
aider

# Intercept now active for all Aider installs

# Add to .windsurfrules in project root:

Security: Before installing any package with pip or npm, verify at
https://stillrunning.io/api/check-package?name={package}

Block if DANGEROUS. Warn if SUSPICIOUS.
Token: stored in ~/stillrunning.yaml