ptp-requesting-code-review

Reviews code diffs against contract documents and emits structured findings.

Updated Jun 8, 2026
One-click install
npx skills add https://github.com/AlmogMaayan/ptp --skill ptp-requesting-code-review-almogmaayan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ptp-requesting-code-review
Source: https://github.com/AlmogMaayan/ptp/tree/main/skills/ptp-requesting-code-review
Command: npx skills add https://github.com/AlmogMaayan/ptp --skill ptp-requesting-code-review-almogmaayan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Code reviews often drift into subjective commentary or miss whether the work actually matches the agreed specification. This Skill enforces a disciplined review pass that checks contract compliance first and code quality second, then emits findings in a strict, machine-consumable record format. ## Core Features & Use Cases - Contract-first review: Assesses whether the diff matches proposal.md, design.md, tasks.md, and spec deltas before judging code quality. - Structured finding records: Emits each finding with exactly six fields (key, severity, location, defect, evidence, remedy) and no extra prose. - Severity discipline: Uses only the four labels Critical, High, Medium, Low, and requires quoted evidence from the current file content for every finding. - Use Case: During a PtP review loop, supply the change contract and diff; the Skill returns findings that downstream skills like review-loop and review-severity can filter, key, and threshold. ## Quick Start Ask the AI to review the supplied contract and diff for contract compliance and emit findings in the six-field record format.

Frequently Asked Questions about ptp-requesting-code-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a contract-first code review on a diff?

Supply the contract documents (proposal.md, design.md, tasks.md, spec deltas) together with the diff, then assess contract compliance before code quality. Emit each issue as a finding record with key, severity, location, defect, evidence, and remedy fields.

What format should code review findings use for automation?

Each finding is a single block with exactly six fields: key, severity, location, defect, evidence, and remedy. No strengths section, summary essay, or extra narrative fields are allowed, so downstream tooling can parse and filter findings deterministically.

What severity levels does the review use?

Findings use exactly four severity labels: Critical, High, Medium, and Low. Every finding is emitted at its true severity; filtering below-threshold findings is the consumer's responsibility, not the reviewer's.

Can the review consider conversation history or prior sessions?

No. The review is restricted to the supplied contract and diff only, never session history or conversation transcripts. It also dispatches nothing, since model and effort selection belong to the orchestrating skills.

Why must every finding include quoted evidence?

The evidence field must quote the file's current content at the cited location, and a finding without such evidence is not emitted. This keeps findings verifiable and prevents speculative or stale reports.