v4 · 17-factor risk engine · CI-aware

The release gate for the
AI-agent era.

Trailhead waits for your CI, scores every pull request across 17 weighted risk factors, checks production health, and gates AI-authored PRs before they hit main — one Release Ready check, one merge rule.

Install from GitHub MarketplaceView Trailhead Cloud pricing

Free as a GitHub Action · no API key required to start

- uses: KomatikAI/trailhead@v4
  with:
    gate-mode: release-ready
    wait-for-checks: "true"
    risk-threshold: "70"
17
weighted risk factors
1
merge check — Release Ready
3
gate modes: release-ready, advisory, risk-only
0
secrets required for the free Action
How it works

Three steps from PR open to merge decision

STEP 1

Install the Action

Add one workflow step. Trailhead reads .trailhead.yml, waits for your required CI checks, and posts a single composite check on every pull request.

STEP 2

Score risk, CI, and health

17 weighted factors — code churn, security alerts, supply-chain changes, sensitive files, CI integrity, deploy history — roll up into a 0-100 risk score, plus a production health probe.

STEP 3

Get one merge decision

allow, warn, or block. "Trailhead — Release Ready" is the one check your branch protection rule requires, so there's nothing else to reconcile across CI and risk.

Agent governance

Gate the PRs your agents open, not just the ones your team writes

Copilot, Codex, and Claude now open pull requests directly. Trailhead treats agent-authored code as a distinct risk class instead of scoring it the same as a human diff — the gap most CI tooling still ignores in 2026.

Provenance classification

Every PR is labeled human, dependabot, copilot, codex, claude, custom-bot, or unknown.

Agent-policy findings

Sensitive-path gates, required-approval counts, and code-owner review for agent-authored PRs.

Session-burst detection

Flags correlated bursts of agent commits inside a time window — a signature of runaway automation.

Trust profiles & SLAs

Per-repo strictness (baseline, elevated, strict) plus escalation status and acknowledge/resolve SLAs.

Governance policies (approval requirements, sensitive-path gates, strict unknown-provenance handling) are configured per-repo in .trailhead.yml — see the agent governance docs.

Dogfooded internally

We run Trailhead on our own agent fleet

Komatik gates every pull request its own AI agents open against KomatikAI/agents with the same submission engine, trust scoring, and risk factors shipped here — provenance classification, sensitive-path policies, and Gate 1 submission checks, running on real, autonomous, multi-agent PR traffic every day.

Pricing

Free to gate. Paid to remember.

The Action is free forever — no API key, no store. Trailhead Cloud adds hosted evaluation history, dashboards, and org-wide agent-governance policy for teams that need the record.

Free$0

Risk-only gate, local

Pro$39/mo

5,000 evals/mo + dashboard

Team$399/mo

50,000 evals/mo + org rollup + SSO

See full plan details

Ship your next PR through a real gate

Add the Action in a few minutes. Add Cloud when you need the history.

Install from GitHub MarketplaceView source on GitHub