Skip to content

Documentation

Understand the diff. Inspect the test candidates. Write the checks that matter.
Documentation Portal

Impact Gate starts with deterministic diff analysis.

Review pull requests, release branches, hotfixes, and shipped tags. Inspect existing spec associations and their provenance before choosing which scenarios to author or run first.

Primary workflow

Review → Inspect → Test

node dist/cli.js review —since origin/mainnode dist/cli.js review —since origin/main —scenarios-output scenarios.jsonnode dist/cli.js gate —since origin/main —threshold 80

Build the current source, export a reviewable scenario plan, and check spec-mapping policy. A passing gate does not establish measured coverage or release safety.

Detailed concepts

How the evidence layer works

Learn route families, confidence scoring, plan artifacts, and where AI fits into the workflow without becoming the workflow.

Commercial comparison

See where it fits against commercial platforms

Compare Impact Gate honestly against hosted optimization, AI-first E2E, no-code automation, and monitoring-oriented tools.

Release review

Inspect the changes since the previous release

Compare the current candidate to the last shipped tag and turn that delta into review context while retaining the required regression suite.

Hallucination guardrails

Ground generation against what your repo already knows

Review generation uses local API context. Unaccepted proposals stay in quarantine; acceptance currently requires direct local-source mutation evidence. Browser application tests remain unverified by that verifier.

Experimental browser QA

Explore one running user journey

Give the agent a running app URL and a bounded goal. Inspect the resulting findings and reproduction steps before authoring tests.

Experimental AI

Layer generation and healing on after the plan is useful

Configure providers, inspect artifacts, and add autonomous workflows only after the deterministic path is already trustworthy.

Operational guides

CI, troubleshooting, and cost control

Use focused playbooks for PR gating, release diffs, AI safety, budget controls, and rollout troubleshooting.