gait-incident-to-regression
Convert a Gait run artifact into a deterministic regression workflow. Use when asked to initialize fixtures from run_id or runpack path, run graders, produce CI-friendly outputs, or summarize drift and failures.
Best use case
gait-incident-to-regression is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Convert a Gait run artifact into a deterministic regression workflow. Use when asked to initialize fixtures from run_id or runpack path, run graders, produce CI-friendly outputs, or summarize drift and failures.
Teams using gait-incident-to-regression 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-incident-to-regression/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How gait-incident-to-regression Compares
| Feature / Agent | gait-incident-to-regression | 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?
Convert a Gait run artifact into a deterministic regression workflow. Use when asked to initialize fixtures from run_id or runpack path, run graders, produce CI-friendly outputs, or summarize drift and failures.
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
# Incident To Regression Execute this workflow to transform an observed run into repeatable CI checks. ## Gait Context Gait is the offline-first policy-as-code runtime for AI agent tool calls. It enforces tool-boundary policy, emits signed and verifiable evidence artifacts, and supports deterministic regressions. Use this skill when: - incident triage needs repeatable fixture creation - CI gate failures require deterministic grader reruns - receipt/evidence generation depends on regression outputs Do not use this skill when: - Gait CLI is unavailable in the environment - no Gait run/pack artifact or run identifier is available as input ## Workflow 1. Resolve source run artifact: - use `<run_id>` or `<runpack_path>` 2. Initialize fixture deterministically (required): - explicit path: `gait capture --from <run_id_or_path> --json` - then `gait regress add --from ./gait-out/capture.json --json` - legacy fallback: `gait regress init --from <run_id_or_path> --json` 3. Parse and report: - `ok`, `run_id`, `fixture_name`, `fixture_dir`, `config_path`, `next_commands` 4. Run regression suite (required): - `gait regress run --json` 5. If CI output is requested, add JUnit: - `gait regress run --json --junit junit.xml` 6. Return concise summary: - source run - fixture path - pass/fail status - failed graders count - output paths ## Safety Rules - Keep replay deterministic defaults. - For replay workflows, prefer `gait run replay` (stub mode default); require explicit unsafe flags for real tool replay. - Do not pass `--allow-nondeterministic` unless explicitly requested. - Treat non-zero regress run exits as regressions, not soft warnings. - Keep this skill wrapper-only: no inline grading logic and no policy-evaluator behavior outside CLI calls. ## Determinism Rules - Always create a deterministic fixture before `regress run` for new incidents. - Always consume `--json` output fields for decisions. - Keep fixture names stable and explicit when user provides naming constraints.