gait-policy-test-rollout
Evaluate Gait policy changes safely with deterministic fixture tests and rollout simulation. Use when asked to validate allow or block behavior, check reason codes, compare policy outcomes, or plan enforce-by-stage rollout.
Best use case
gait-policy-test-rollout is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Evaluate Gait policy changes safely with deterministic fixture tests and rollout simulation. Use when asked to validate allow or block behavior, check reason codes, compare policy outcomes, or plan enforce-by-stage rollout.
Teams using gait-policy-test-rollout should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/gait-policy-test-rollout/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How gait-policy-test-rollout Compares
| Feature / Agent | gait-policy-test-rollout | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Evaluate Gait policy changes safely with deterministic fixture tests and rollout simulation. Use when asked to validate allow or block behavior, check reason codes, compare policy outcomes, or plan enforce-by-stage rollout.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Policy Test Rollout Execute this workflow to validate policy behavior before enforcement. ## Gait Context Gait is an offline-first runtime for AI agents that enforces tool-boundary policy, emits signed and verifiable evidence artifacts, and supports deterministic regressions. Use this skill when: - incident triage involves policy decision verification - CI gate failures require deterministic policy verdict checks - receipt/evidence generation needs policy verdict traceability Do not use this skill when: - Gait CLI is unavailable in the environment - no policy fixture inputs or Gait artifacts are available ## Workflow 1. Require both files: - `<policy.yaml>` - `<intent_fixture.json>` 2. Run deterministic policy test (required): - `gait policy test <policy.yaml> <intent_fixture.json> --json` 3. Parse and report fields: - `ok`, `policy_digest`, `intent_digest`, `verdict`, `reason_codes`, `violations`, `summary` 4. If rollout simulation is requested, run: - `gait gate eval --policy <policy.yaml> --intent <intent_fixture.json> --simulate --json` 5. Return structured rollout recommendation: - current verdict - blocking reasons or required approvals - suggested next stage (`observe`, `require_approval`, `enforce`) ## Exit Code Contract - `0`: allow - `3`: block - `4`: require approval - `6`: invalid input ## Safety Rules - Never bypass policy test by inferring results from YAML text. - For replay workflows, prefer `gait run replay` (stub mode default); require explicit unsafe flags for real tool replay. - Never claim a policy digest or verdict without command output. - Keep simulation and enforcement clearly separated in reporting. - Keep this skill wrapper-only: do not re-implement policy parsing or verdict logic. ## Determinism Rules - Always use `--json` outputs. - Report exact `reason_codes` and `violations` as emitted. - Preserve fixture-based evaluation flow for repeatability.