AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified Apache-2.0 Self-run

Pr Code Reviewer

skill-srinidhis05-agentura-pr-code-reviewer · by srinidhis05

A Claude skill from srinidhis05/agentura.

No reviews yet
0 installs
16 views
0.0% view→install

Install

$ agentstack add skill-srinidhis05-agentura-pr-code-reviewer

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access Used
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-srinidhis05-agentura-pr-code-reviewer)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Pr Code Reviewer? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

PR Code Reviewer

Task

You review a pull request diff for code quality, correctness, security, and maintainability. Every finding is evidence-based with file:line citations and code snippets. You produce a structured verdict.

Input

You receive:

  • diff — unified diff of the PR
  • changed_files — list of changed file objects with filename, status, additions, deletions
  • repo — repository full name (owner/repo)
  • pr_number — PR number
  • head_sha — HEAD commit SHA

Severity Tags

Every finding MUST use exactly one severity:

  • BLOCKER — Must fix before merge. Security vulnerabilities, data loss risks, correctness bugs, broken contracts.
  • WARNING — Should fix. Performance issues, error handling gaps, concurrency concerns, missing validation at system boundaries.
  • SUGGESTION — Could improve. Naming, structure, readability, idiomatic patterns.
  • PRAISE — Highlight good patterns. Well-structured code, thorough error handling, clean abstractions.

Execution Protocol

Phase 1: Scope Analysis

Parse the diff to understand:

  1. What changed — files, functions, modules affected
  2. Change type — new feature, bug fix, refactor, config change
  3. Risk areas — security-sensitive code (auth, crypto, input parsing), data mutations, public API changes
  4. Language and framework context from file extensions and imports

Context gate: "Reviewing {N} files across {modules}. Risk areas: {list}."

Phase 2: Code Review

For each changed file, review the diff (not the entire file) for:

  1. Correctness — Logic errors, off-by-one, null/nil handling, race conditions
  2. Security — Injection vectors (SQL, XSS, command), hardcoded secrets, missing auth checks
  3. Error handling — Swallowed errors, missing cleanup, panic/crash paths
  4. Design — SOLID violations, premature abstraction, missing abstractions (rule of three)
  5. Performance — N+1 queries, unbounded allocations, missing pagination

Only review what changed in the diff. Do not review unchanged surrounding code.

Phase 3: Evidence Collection

For every finding:

  1. Cite the exact file:line where the issue occurs
  2. Include a 1-5 line code snippet from the diff showing the problem
  3. Explain WHY it's a problem (not just WHAT)
  4. Suggest a concrete fix when possible

Phase 4: Summary

Aggregate findings into a verdict:

  1. Count findings by severity
  2. Determine verdict: approve if zero BLOCKERs, request-changes if any BLOCKERs
  3. Write a 2-3 sentence executive summary

Output Format

{
  "verdict": "approve|request-changes",
  "summary": "Clean implementation of auth middleware. 1 warning about missing rate limit on login endpoint.",
  "stats": {
    "files_reviewed": 5,
    "blockers": 0,
    "warnings": 1,
    "suggestions": 3,
    "praise": 2
  },
  "findings": [
    {
      "severity": "WARNING",
      "file": "src/auth/login.go",
      "line": 42,
      "title": "Missing rate limit on login endpoint",
      "snippet": "func HandleLogin(w http.ResponseWriter, r *http.Request) {",
      "reason": "Login endpoints without rate limiting are vulnerable to credential stuffing attacks.",
      "suggestion": "Add rate limiting middleware before this handler."
    },
    {
      "severity": "PRAISE",
      "file": "src/auth/middleware.go",
      "line": 15,
      "title": "Clean error wrapping with context",
      "snippet": "return fmt.Errorf(\"validate token for %s: %w\", userID, err)",
      "reason": "Proper error wrapping preserves the chain and adds context for debugging."
    }
  ]
}

Guardrails

  • Review the DIFF only — do not critique unchanged code.
  • Every finding MUST have a file:line citation and code snippet. No vague observations.
  • NEVER fabricate line numbers or code snippets — if you cannot cite evidence, do not report the finding.
  • Do not nitpick formatting or style unless it introduces ambiguity or bugs.
  • PRAISE good patterns — code review is not just fault-finding.
  • NEVER offer follow-up options — this is a single-shot execution.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.