skills.rest
seal-harness
seal-harnessOFFICIAL
Explore all skills by this author →
GitHub

self-reflect

Extract reusable learnings from PR reviews and session patterns into a knowledge base.

1|Updated Jun 23, 2026
Search Tags:#code review#github#coderabbit#knowledge base#pull requests#self-reflection#learnings extraction
One-click install
npx skills add https://github.com/seal-harness/seal-harness --skill self-reflect-seal-harness
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: self-reflect
Source: https://github.com/seal-harness/seal-harness/tree/main/.agents/skills/self-reflect
Command: npx skills add https://github.com/seal-harness/seal-harness --skill self-reflect-seal-harness

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Valuable insights from PR reviews, CodeRabbit comments, and debugging sessions are lost after each pull request closes, forcing teams to relearn the same lessons repeatedly. ## Core Features & Use Cases - PR Comment Mining: Fetches review comments via the GitHub CLI and extracts structured learnings from CodeRabbit output. - Quality Filtering: Evaluates each candidate learning with accept/reject/transform criteria, deduplication checks, and confidence scoring before storage. - Knowledge Base Storage: Canonicalizes approved learnings into typed JSONL entries (pattern, gotcha, decision, security, performance) and writes summary reports to docs/learnings/. - Use Case: After a sprint with multiple merged PRs, run a reflection pass to convert recurring reviewer feedback into durable coding rules that future agent sessions can apply. ## Quick Start Analyze recent PR reviews and session history, then propose candidate learnings for me to approve and add to the knowledge base.

Frequently Asked Questions about self-reflect

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

FAQPage Schema
How do I extract learnings from GitHub PR review comments?▼

Use the GitHub CLI to fetch PR comments and reviews as JSON, then filter CodeRabbit's structured 'Learning:' lines with jq and grep. Evaluate each candidate against quality criteria before adding it to the knowledge base.

What makes a good knowledge base learning from code reviews?▼

Good learnings are actionable, codebase-specific or clearly universal, and include the reasoning behind the rule. Reject PR-specific anecdotes, personality observations, and obvious advice that any developer already knows.

Does this require the GitHub CLI to be installed?▼

Yes, PR comment analysis uses the gh CLI to list pull requests and export reviews and comments as JSON. The conversation mining and config reflection phases work without it since they only inspect local context and files.

How are duplicate learnings handled in the knowledge base?▼

Before adding a new fact, existing JSONL knowledge files are checked for semantic duplicates. Conflicts are resolved by superseding, merging, or asking the user which version is correct.

When should I not capture a learning from a review comment?▼

Skip learnings that are too specific to one PR, not actionable by an agent, trivially obvious, or likely to confuse a future agent lacking the original context. Quality matters more than quantity.

Related Skills

Explore More Software Engineering Skills→
GenerativeXGenerativeX

GH Review Comment Response

Automate pull request review comment resolution using the GitHub CLI.

Official
Intermediate
psd401psd401
2

compound

Capture development session learnings into structured documents under docs/learnings.

Official
Intermediate
ajsai47ajsai47
11

learn

Manage cross-session project learnings with search, pruning, and markdown export.

Community
Intermediate
stummanstumman

update-critical-review

Apply validated review ledger feedback to update local critical-review skill documentation.

Community
Advanced