optimlight/AgentProof

Local QA gate for AI-generated code and agent PRs.

View on GitHub

Trust Signals

Scorecard Score: not yet scored
Maintenance Recency: Activelast commit Jun 2, 2026
License: MIT

Pinned Snippet

workflow.ymlSHA-pinned

uses: optimlight/agentproof@8fc248d21a1c7710dc207b6d557f0d0ba7c1be5c # v1

tags can be moved; commit SHAs can't. why a SHA?

Inputs

name	description	required	default
path	Project path to audit.	no	.
fail-under	Minimum score required to pass.	no	80
observe-only	Keep reports and verdicts, but do not fail the workflow.	no	false
no-run-scripts	Skip package scripts.	no	false
changed	Scan only changed files.	no	false
base	Base ref for changed-file scans.	no	origin/main
profile	Risk profile to apply: relaxed, standard, or strict.	no	standard
design	Enable DesignGuard for a product type: landing, saas, dashboard, marketplace, ecommerce, webapp, or mobile-app.	no	""
design-only	Run only DesignGuard. Keeps path for source UI checks but excludes technical/docs/security gates.	no	false
design-url	URL to render for DesignGuard DOM, CSS, viewport, and screenshot proof.	no	""
design-routes	Comma-separated routes to audit with DesignGuard.	no	""
design-viewports	Comma-separated DesignGuard viewports: mobile, tablet, desktop.	no	""
design-ai	Run optional AI design review.	no	false
design-ai-provider	AI provider for DesignGuard, either openai-api, codex-auth, or auto.	no	""
design-ai-model	AI model for DesignGuard, for example gpt-5-mini or gpt-5.3-codex-spark.	no	""
sarif	Optional SARIF output path.	no	""
badge	Optional SVG badge output path.	no	""
html	Optional HTML report path.	no	""
pr-comment	Optional compact Markdown PR comment output path.	no	""
receipt	Optional portable JSON evidence receipt path.	no	""
summary	Optional compact JSON summary path for bots and dashboards.	no	""
history	Optional JSONL history file to append this run to.	no	""
claim	Optional final agent claim text or file path to audit.	no	""
baseline	Optional baseline file to exclude known findings from the active score.	no	""
update-baseline	Optional baseline file to write from current findings.	no	""
github-summary	Append the compact AgentProof summary to GITHUB_STEP_SUMMARY.	no	true
github-annotations	Emit GitHub log annotations for active findings.	no	true

Outputs

name	description
score	AgentProof score from 0 to 100.
verdict	AgentProof verdict, for example SHIP or DO NOT SHIP.
report-path	Markdown report path.
receipt-path	Receipt path when the receipt input is set.
summary-path	Compact summary path when the summary input is set.
findings	Active finding count.
technical-score	TechnicalGate score from 0 to 100.
technical-verdict	TechnicalGate verdict.
design-score	DesignGuard score from 0 to 100 when enabled.
design-verdict	DesignGuard verdict when enabled.
design-findings	Active DesignGuard finding count when enabled.