amirbena / amirbena/code-review-skill

Add integrity validation for policies, references, and packaged Skills

Open
#46 1 comment 0 reactions 0 assignees View on GitHub
area:packaging-portability priority:P1 type:quality
Dominant language
Python
Stars
7
Forks
2
Avg merge
27m
Merged PRs (30d)
188

Description

## Type

Quality

## Area

Packaging / Portability

## Priority

P1 — High

## Problem

Broken cross-references, missing packaged resources, and accidental repository-only dependencies are now largely caught by automated checks (#77, #78, #81 — closed). What remains is documenting the prohibited repository-only path set so it can't silently drift from the enforcement code (#79), and detecting near-duplicate normative-rule text across policy sources, which has no automated check at all (#80).

## Goal

The prohibited repository-only dependency set is documented alongside its enforcement, and canonical-rule duplication risk is automatically detected.

## Scope

Child issues:

- [x] #77 — Validate Markdown and policy references (closed)
- [x] #78 — Validate packaged resource references (closed)
- [ ] #79 — Document the prohibited repository-only path/basename set
- [ ] #80 — Detect canonical-rule duplication risks
- [x] #81 — Add archive contract tests (closed)

## Non-Goals

- Not a documentation style linter.
- Does not reformat content.
- Does not change the instruction architecture itself.

## Acceptance Criteria

- [ ] #79 and #80 are closed.
- [ ] Each child links back to this issue.

## Dependencies

Depends on: none
Blocks: none
Parent: #39

Independent foundation: no dependency on the other Golden items.

## Validation

- Each remaining check fails on a seeded defect and passes on a clean tree.
- Checks run in standard validation.

Contributor guide

Open the contributing guide

Research direction

Start by reading child issues #79 and #80, then trace the existing standard validation referenced by this issue. The payload names no files or test entry points; done means both remaining checks fail on seeded defects, pass on a clean tree, run in standard validation, and link back to #46.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
testing, tooling
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.