- Home
- Skills
- Code Quality & Review
- Prism Code Review Pack
Works with the AI Agents you already use
Prism Code Review Pack
A suite of four developer tools for diff explanations, general logic reviews, performance audits, and security checks.
Free
Prism Code Review Pack
Example session with this skill installed
Review this diff for performance issues, specifically looking for database efficiency and loop complexity: [diff of a Python service updating user records in a loop]
- Read your context and instructions
- Compiled the prism code review
I have reviewed the diff for performance risks. The primary concern is an N+1 query pattern within the user update logic.
Performance Review Summary
The change introduces significant database overhead that will scale poorly as the user_ids list grows. Replacing the per-user update with a bulk operation is recommended.
Findings
| Severity | Risk | Location | Evidence | Recommended Change |
|---|---|---|---|---|
| High | N+1 Database Query | services/user_manager.py:42 | db.query() is called inside a for user_id in user_ids loop. | Use a single UPDATE ... WHERE id IN (...) query or a bulk ORM operation. |
| Medium | Unnecessary Serialization | services/user_manager.py:45 | User profile JSON is re-parsed in every iteration. | Move json.loads(profile_template) outside the loop. |
| Low | Growth Complexity | services/user_manager.py:38 | Time complexity is O(N) where N is the number of users. | The N+1 issue makes the constant factor high; bulk operations maintain O(N) but reduce IO wait. |
Measurement Plan
- Metric: Database execution time per 1,000 users.
- Workload: 10, 100, and 1,000 user IDs in the input list.
- Comparison: Measure latency of the current loop vs. a
bulk_updateimplementation.
Next steps
- Refactor
update_user_statusto accept a list of IDs for a single database transaction. - Profile the
json.loadscall to see if template caching is necessary for larger payloads.
Connects securely to your tools. The creator never sees your data.
About this skill
Prism Code Review Pack is a collection of four specialized skills designed to eliminate the friction of reviewing pull requests and unfamiliar codebases. Reviewing complex diffs is slow and prone to human error, especially when balancing performance, security, and correctness. This pack provides structured, high-signal analysis for developers who need to understand intent and catch defects without manual drudgery.
What it does
- Explain Diffs translates raw code changes into plain-language walkthroughs of intent, logic, and side effects.
- General Review identifies correctness issues, off-by-one errors, and maintainability risks with ranked severity levels.
- Performance Audit detects N+1 queries, memory leaks, and complexity bottlenecks using workload-based hypotheses.
- Security Inspection maps trust boundaries to find injection risks, broken access controls, and exposed secrets.
How it works
- Analyze Diff by processing the code changes alongside surrounding files and architecture notes.
- Apply Lens based on the chosen skill, focusing on either general logic, performance, security, or explanation.
- Validate Findings against provided context like test output, data models, or linked issues.
- Generate Report featuring structured tables or lists with severity, location, evidence, and concrete fixes.
Frameworks & tools
Works with any language or framework visible in a diff, including SQL, REST/GraphQL APIs, and cloud configurations. It integrates with development environments like Cursor or Claude Code that provide file system access.
Why this beats prompting it yourself
Writing prompts that consistently categorize severity and map trust boundaries is difficult. Prism uses established workflows to ensure you get actionable remedies and measurement plans rather than generic style suggestions.
Use cases
- Onboard Reviewers by providing a clear walkthrough of a massive, unfamiliar PR.
- Harden APIs by identifying broken access controls and injection paths before merging.
- Optimize Data Paths by catching expensive loops and missing database indexes.
- Automate PR Summaries to generate high-quality descriptions for stakeholders.
Known limitations
Does not execute code or run actual benchmarks. Security reviews are scoped to provided files and are not a substitute for full penetration testing.
No code execution or live benchmarking. Security reviews are limited to the provided diff scope. Requires surrounding file context for highest accuracy.
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 any payment.
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 forever
No account required to browse
Trust & safety
Security scan
Security scan passed 12 days ago
- Free with an Agensi account