More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    PR Blast Radius Auditor

    1

    Trace what a pull request can break and produce targeted merge evidence.

    $19.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    PR Blast Radius Auditor

    PR Blast Radius Auditor

    Example session with this skill installed

    Audit this PR before merge. It changes packages/auth/token.ts and services/api/auth-middleware.ts. The shared parser will now reject tokens with our legacy issuer. services/refund-worker imports the same auth package, but it is absent from the changed files. We have API unit tests, no worker replay test, no issuer telemetry, and rollback currently means redeploying the previous auth package. Produce the dependency cone, material findings, targeted test matrix, and verdict.

    • Read your context and instructions
    • Compiled the pr blast radius

    Verdict — TARGETED-VERIFICATION-REQUIRED

    The change crosses a shared authentication contract and can affect a refund worker that is not visible in the diff. The blast radius is bounded enough to test, but merge evidence is incomplete because legacy-issuer traffic and the worker consumer are not covered.

    Changed-surface ledger

    Changed pathClassificationBoundary
    packages/auth/token.tsshared runtime logicissuer acceptance contract
    services/api/auth-middleware.tsAPI integrationauthenticated request behavior

    Dependency and boundary map

    token.ts → shared auth package → API middleware and refund worker. The material boundary is token issuer acceptance. Rejecting a previously accepted issuer is externally observable even if the parser signature is unchanged.

    Material finding

    HIGH — Refund worker may reject valid legacy tokens. Evidence: the shared import is observed; the volume of legacy issuers is unknown. Failure mechanism: the worker receives the same tokens but lacks a replay test. Required test: replay current and legacy issuer fixtures through both API and worker consumers. Owner question: which service owns issuer retirement? Rollback: retain the previous auth package as a deployable artifact until production issuer evidence is available.

    Must-test matrix

    1. Current issuer through API and refund worker — proves the retained path.
    2. Legacy issuer through both consumers — directly discriminates the compatibility risk.
    3. Mixed-version deploy — checks whether old and new workers can coexist.
    4. Rollback smoke test using the previous package — proves reversal is operational, not theoretical.

    Unknowns

    Legacy-issuer frequency, mobile or external token issuers, worker replay coverage, and issuer-specific telemetry remain unknown. Absence of a failing test is not evidence of safety.

    Validator boundary

    The structured demonstration contains all six coverage domains and passes the bundled schema validator. This validates completeness of the record only; no repository or runtime test was executed in this demonstration.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    A local diff can break a distant system. Changed lines rarely show the complete impact of a pull request: shared parsers, event payloads, configuration keys, persisted formats, build artifacts, and deployment assumptions can affect workers and clients that never appear in the diff.

    PR Blast Radius Auditor converts repository evidence into a bounded dependency cone. It classifies changed surfaces, traces direct and indirect consumers, identifies runtime, contract, state, build, observability, and rollback risks, then chooses the smallest tests that can confirm or falsify each material finding.

    Built for backend, platform, monorepo, and release engineers reviewing shared libraries, APIs, events, migrations, background jobs, and cross-service pull requests. It profiles the repository's own framework, test runner, package graph, job adapter, ORM, and deployment conventions before applying the audit.

    What the buyer receives

    • Merge-evidence verdict
    • Changed-surface ledger
    • Dependency and boundary map
    • Severity-ranked findings with observed, inferred, and unknown evidence separated
    • Targeted must-test matrix
    • Owner questions and unresolved evidence
    • Rollback proof plan
    • Deterministic JSON validation

    The package includes the complete Skill source, domain method, worked audit, validator, mutation tests, expected-output example, trigger tests, and a six-page Buyer Guide PDF.

    Boundaries

    The Skill analyzes buyer-supplied or locally visible artifacts. It does not edit code, approve, merge, push, deploy, or contact owners. A passing validator proves record coverage, not runtime safety.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.

    Requires buyer-supplied diffs, changed-file sets, repository manifests, or file trees for dependency mapping. Findings remain bounded by visible evidence. It does not execute tests, approve merges, or guarantee runtime safety; the bundled validator checks report completeness only.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 2 months ago

    Permissions

    No special permissions declared or detected

    Listed2 months ago

    What's inside

    Frequently Asked Questions

    Unlike standard diff tools, this skill identifies 'silent' dependencies like event payloads and shared parsers that don't appear in the changed lines but could fail in production.

    The skill is language-agnostic as it profiles your specific framework, ORM, and package graph to understand how your unique system architecture handles dependencies.

    You receive the complete Skill source, the domain method, a six-page Buyer Guide, and a validation suite to ensure the auditor is correctly mapping your repository.

    No, the skill operates in a read-only capacity; it analyzes artifacts and produces evidence-based reports without ever editing, pushing, or merging code.

    This skill is designed for any AI agent capable of repository analysis and file-system access, such as AutoGPT, OpenDevin, or custom enterprise LLM workflows.

    The 'must-test matrix' identifies the smallest possible subset of your existing tests required to validate the PR, saving time by avoiding unnecessary full-suite runs.

    New in Code Quality & Review