PR Evidence Pack Generator

$9 one-time

Prove your AI code review before merge.

AI coding agents generate PRs fast. But teams need evidence — not just claims — that changes were reviewed, tests were run, and risk was assessed. The PR Evidence Pack Generator produces structured, shareable evidence reports from any diff.

Risk Analysis

Automatic risk scoring (0–100) and flagging: dependency changes, config changes, secrets in diffs, auth/payment paths, source-without-tests, and infrastructure changes. Now configurable per team.

GitHub PR Comment

Generate ready-to-paste GitHub PR comments with emoji risk indicators, collapsible metadata, and review checklists. Just --format github-comment.

Configurable Severity NEW v1.2

Custom risk thresholds, flag weights, file patterns, sensitive paths, and secret detection via JSON config. --severity-config team.json

Reviewer Checklist & Sign-Off

Tailored reviewer checklists based on detected risk flags. Author, reviewer, and security reviewer sign-off requirements based on risk level.

Test Coverage

Parses pytest, jest, go test, and generic test output. Shows pass/fail counts and test commands directly in your evidence report.

Quick start

      # Generate Markdown evidence report

      python3 src/pr_evidence_pack.py --diff my-pr.diff

      # GitHub PR comment — paste directly into GitHub

      python3 src/pr_evidence_pack.py \\

        --diff my-pr.diff \\

        --test-output pytest-output.txt \\

        --metadata pr-meta.json \\

        --format github-comment \\

        --output pr-comment.md

      # Full report with test output + metadata → HTML

      python3 src/pr_evidence_pack.py \\

        --diff my-pr.diff \\

        --test-output pytest-output.txt \\

        --metadata pr-meta.json \\

        --format html \\

        --output evidence.html

      # Use in CI: exit 1 on high risk → block merge

      python3 src/pr_evidence_pack.py --diff pr-changes.diff

      echo $?

      # Custom severity config for your team

      python3 src/pr_evidence_pack.py \\

        --diff my-pr.diff \\

        --severity-config team-config.json \\

        --format markdown

What's in the report

Sample Evidence Pack Output

Section	Content
Metadata	PR title, URL, author, reviewer, branch, timestamp
Risk Summary	Score 0–100, level HIGH / MEDIUM / low, verdict
Risk Flags	DEPENDENCY_CHANGE, SENSITIVE_AREA_CHANGE, POSSIBLE_SECRET_LITERAL, etc.
Changed Files	Per-file: path, +lines, -lines, risk flags
Test Coverage	Tests run, passed, failed, commands
Reviewer Checklist	Tailored checklist based on risk flags
Sign-Off	Author / Reviewer / Security Reviewer requirements

What's included

CLI tool — run from terminal, CI, or pre-commit hook
4 output formats — Markdown, HTML, JSON, GitHub PR comment
CI exit codes — exit 1 on high risk (block merge)
Python API — import and use in your own tools
GitHub PR comments — emoji risk indicators, collapsible metadata, review checklists
Test output parser — pytest, jest, go test, generic
Risk analysis — 7 risk flag categories, scoring, recommendations
Reviewer checklists — tailored to risk level
Sign-off tracking — role-based requirements
Configurable severity — custom thresholds, weights, file patterns, sensitive paths via JSON config
Sample files — diff, test output, metadata JSON, severity config, GitHub comment output
58 unit tests — tested and verified
Works standalone — no external dependencies (Python stdlib only)
Integrates with Audit Kit — enhanced risk analysis when both installed

Who is this for?

Tech leads reviewing AI-generated PRs who need evidence of review
Agencies shipping client work with AI tools who need sign-off records
Solo developers using Cursor, Claude Code, or Codex who want structured self-review
CI/CD pipeline builders who want risk gates for AI-generated changes

Evidence-backed — no fake claims

58 passing unit tests. py_compile validated. Sample fixtures that produce real output. This is a working CLI tool, not a prompt pack or PDF.

Ran 58 tests in 0.019s — OK
py_compile: OK
ZIP SHA256: verified

Honest limitations

Does not enforce sign-off — it generates reports and checklists for your team to use
Diff analysis is pattern-based, not a full AST parser
No network calls — fully local and offline
No fake testimonials or inflated metrics

Also available

Pair with the AI Agent Change Risk Audit Kit for enhanced risk analysis, or the CI Gate Templates Pack for drop-in CI/CD risk gates.

Buy PR Evidence Pack Generator — $9