- Home
- Skills
- Code Quality & Review
- severity rated code review
Works with the AI Agents you already use
severity rated code review
Logic-first code review agent that ranks actionable security and performance defects by severity.
Free
severity rated code review
Example session with this skill installed
Review the changes in auth_provider.py. Focus on security and reliability.
- Read your context and instructions
- Compiled the severity rated code
I have reviewed the changes in auth_provider.py against the current main branch. I identified two actionable findings related to session management and error handling.
Findings
[HIGH] Potential Session Fixation in Token Refresh
services/auth_provider.py:42-45
The refresh_session function updates the expiry timestamp but fails to rotate the underlying session identifier. An attacker with access to a hijacked session token could maintain persistent access even after a user triggers a refresh. Recommend calling self._rotate_id() during the refresh flow to generate a new UUID.
[MEDIUM] Unhandled Exception in OAuth Callback
services/auth_provider.py:89-92
The callback handler does not catch timeout errors from the identity provider. A slow network response will trigger a 500 Internal Server Error, potentially leaking stack traces to the client. Recommend wrapping the request in a try-except block with a specific ProviderTimeout handler and returning a structured 503 error.
Checks
pytest tests/test_auth.py: Passed (existing tests do not cover token rotation).flake8 services/auth_provider.py: No relevant linting errors.- Manual trace of
session_storeinteraction: Verified ID persistence.
Next steps
- Implement session ID rotation in the
refresh_sessionmethod. - Add a try-except block to the OAuth callback to handle external API timeouts.
- Update the test suite to include a test case for session ID changes post-refresh.
Connects securely to your tools. The creator never sees your data.
About this skill
Tired of code reviews that focus on variable naming while missing critical race conditions? This skill transforms your agent into a skeptical, security-first reviewer that ignores bike-shedding to hunt for logic flaws, data leaks, and performance bottlenecks. It delivers high-signal reports that prioritize system integrity over stylistic preferences.
What it does
- Severity-ranked reporting Organizes findings into Critical, High, Medium, and Low buckets based on real-world impact.
- Evidence-based auditing Maps every finding to exact file paths and line ranges with concrete failure scenarios.
- Security-first analysis Scrutinizes authorization boundaries, untrusted input handling, and credential exposure.
- Behavioral tracing Evaluates state transitions, error paths, and persistence logic rather than just syntax.
- Actionable remediation Provides specific technical directions for fixing identified defects without bloating the codebase.
How it works
- Scope Definition The agent analyzes the PR, diff, or repository to establish the language context and affected boundaries.
- Defect Hunting It traces inputs and state transitions to find plausible paths to failure or exploitation.
- Validation It runs project-native tests or static analysis to verify risks before reporting them.
- Structured Reporting It outputs a sorted list of findings using a standardized
[SEVERITY] Titleformat followed by line-level evidence.
Frameworks & tools
Works with any language or framework visible to your coding agent. It utilizes project-native CLI tools, linters, and test runners already present in your repository for verification.
Why this beats prompting it yourself
Standard prompts often produce "hallucinated" style critiques or generic advice. This skill enforces a strict evidence-based quality bar, requiring a plausible execution path for every finding and preventing the agent from reporting unchanged code or stylistic nitpicks.
Use cases
- Auditing pull requests for high-stakes security vulnerabilities or data loss risks.
- Performing deep-dive reviews of complex state management or concurrency logic.
- Verifying that new code changes haven't introduced regressions in authorization flows.
- Cleaning up technical debt by identifying resource leaks and performance bottlenecks.
Known limitations
Does not report stylistic or formatting issues unless requested. Requires access to the base revision or full repository context to accurately identify regressions.
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.
Focuses on logic and security; ignores style/formatting. Effectiveness depends on the agent's ability to access the full repository context or diff history.
How to use
- 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
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 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
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 13 days ago
- Free with an Agensi account