install-hooks
/install-hooks
Install or repair hooks that check for secrets on each commit, push, or AI assistant action. Use when the user wants leak prevention, when credential-handling work has no hook protection, or when agent scans duplicate a hook that already exists. Use scan-secrets for a one-off scan or a leak investigation.
- Author
- gitguardian ↗
- Repository
- agent-skills ↗
- Licence
- MIT
- Last repository change
- 06.11.2026
Assessment and security
- Files in this package
- 12
- Scripts
- 5
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 06.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
- 5 scripts in this package. Review them before installing, as Claude may execute them.
Scripts: evals/files/already-has-hook/setup.sh, evals/files/ambiguous-asks-family/setup.sh, evals/files/global-needs-consent/setup.sh, evals/files/install-claude-code/setup.sh, evals/files/install-precommit-local/setup.sh
What it does and when to use it
Install or repair hooks that check for secrets on each commit, push, or AI assistant action. Use when the user wants leak prevention, when credential-handling work has no hook protection, or when agent scans duplicate a hook that already exists. Use scan-secrets for a one-off scan or a leak investigation.
Original author description
Install or repair hooks that check for secrets on each commit, push, or AI assistant action. Use when the user wants leak prevention, when credential-handling work has no hook protection, or when agent scans duplicate a hook that already exists. Use scan-secrets for a one-off scan or a leak investigation.
How it works
- Overview
- When to Use
- Onboarding (first use)
- Commands
- Best Practices
- Troubleshooting
SKILL.md preview
--- name: install-hooks description: Install or repair hooks that check for secrets on each commit, push, or AI assistant action. Use when the user wants leak prevention, when credential-handling work has no hook protection, or when agent scans duplicate a hook that already exists. Use scan-secrets for a one-off scan or a leak investigation. license: MIT compatibility: Requires the ggshield CLI installed and authenticated, version 1.49.0 or later for AI-assistant hooks and 1.51.0 or later for codex. Git hooks need git; AI-assistant hooks need the target tool installed. metadata: version: "0.8.0" # x-release-please-version --- # Install secret-scanning hooks ## Overview Hooks do the recurring checks. The agent does not. Never run a scan yourself, and never write a script or other tool to find secrets. Check the existing protection once per context. Reuse protection that works. Offer a missing installation or a repair one time. If the user declines, do not ask again, and do not run scans instead. ## When to Use | Requested protection | Hook | Content that the hook checks | |---|---|---| | Before a commit | `pre-commit` | Staged changes | | Before a push | `pre-push` | Outgoing commits, including intermediate versions | | During AI assistant actions | AI hook for the named or active assistant | Supported prompt and tool events | Use the stage, the assistant, and the scope that the user specified. Ask only for a missing choice. For a bare "install hooks" request, ask which hook family the user wants. An explicit installation request authorizes its stated scope. A recommendation request or a scan request does not authorize an installation. A global installation needs explicit authorization. ## Onboarding (first use) ### Prerequisites Before an installation, check: - `ggshield --version` - the `--help` output of the relevant command - `ggshield api-status` for authentication - the target assistant, when the request names one Do not repeat a check that already passed in this context. ### Setup If the CLI is missing or not authenticated, read [ggshield-cli-setup.md](references/ggshield-cli-setup.md). Otherwise, continue. ## Commands Read the relevant section of [hook-selection.md](references/hook-selection.md) when you check, install, or repair protection. It covers the effective Git hook paths, existing hook managers, and scoped AI hook installation. Do not read it again for a routine operation that it already covered. After an installation, check the configuration. Do not run a test scan. Do not create a test secret, a test commit, or a test push. Report "configured" and "ran and passed" as two different states. The next authorized operation runs the hook and gives the "ran and passed" result. ## Best Practices - Keep existing hook managers and existing secret protection. A hook that only runs a linter is not secret protection. - Do not add manual scans next to a hook. Do not bypass a hook failure. Do not report a skipped or failed check as clean. For an explicit scan, use `scan-secrets` when it is available. - When a hook finds a secret, report the secret type and location. Do not show the secret value. Show the organization's remediation message first, when one exists. Determine if the secret was exposed. A blocked commit or push does not prove that the secret never leaked elsewhere. If the secret is only local, remove it, including from unpushed commits. If the secret was exposed, coordinate a rotation. The same hook checks the next authorized retry. If `scan-secrets` is installed, it contains detailed remediation guidance. ## Troubleshooting If a hook does not run, check the effective hook paths, the hook manager dispatch, the enabled events, the authentication, and the project trust settings. The hook-selection reference describes each check. An AI scan error can let an unscanned action continue. To uninstall, remove only the ggshield entries, through the hook manager that owns them. Keep the other hooks.
Files in this package (12)
- SKILL.md 4 KB
- evals/evals.json 10 KB
- evals/targets.json 1 KB
- evals/files/README.md 4 KB
- evals/files/already-has-hook/setup.sh 2 KB
- evals/files/ambiguous-asks-family/setup.sh 1 KB
- evals/files/global-needs-consent/setup.sh 1 KB
- evals/files/install-claude-code/setup.sh 1 KB
- evals/files/install-precommit-local/setup.sh 1 KB
- references/ggshield-cli-setup.md 14 KB
- references/gitguardian-platform.md 6 KB
- references/hook-selection.md 6 KB
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · gitguardian/agent-skills | 90 | MIT |
| rvdbreemen/adr-kit | 100 | MIT |
| gitguardian/agent-skills | 100 | MIT |
Database snapshot: 13.11.2026 · package version 7c78a011c274 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 90security-secrets-scan
Scans repo and build artifacts for leaked secrets (keys, tokens, certs) and proposes a prevention policy — pre-commit hooks, CI check, rotation plan. Output: findings report, remediation steps (rotate / revoke / invalidate), and hook/CI proposals. Pulled by the `security`… - 100scan-secrets
Scan files, staged changes, commits, history, Docker images, or packages for hardcoded secrets. Use when the user asks for a scan, during an audit, or when the user asks you to investigate a suspected leak or a finding. Use install-hooks for recurring checks on edits, commits,… - 100specstory-guard
Install a pre-commit hook that scans .specstory/history for secrets before commits. Run when user says "set up secret scanning", "install specstory guard", "protect my history", or "check for secrets". - 100secrets-management
Find hardcoded secrets in code, config and git history (gitleaks, trufflehog, detect-secrets), plan and run credential rotation after a leak, and move secrets into Vault, AWS Secrets Manager, GCP Secret Manager, Azure Key Vault or KMS-encrypted files (SOPS, Sealed Secrets,… - 100secrets-scan
Detect hardcoded credentials, API keys, tokens, and secrets in source code and configuration files. Use when reviewing code for leaked secrets before commit/merge, auditing a repository for credential exposure, or setting up secret detection. - 100synthesis-git-hooks
Commit-time protection: blocks credentials, private keys and credential files in every repo, and unapproved disclosures where outsiders read it (class read from push remotes, with a disclosure ledger). Use to install, configure or debug the commit check, or when a commit is refused.