A collection of skills for AI coding agents. Skills are packaged instructions, scripts, and reference material that extend an agent's capabilities for common software engineering workflows.
Skills in this repository follow the Agent Skills format.
Run detailed CodeRabbit-powered code reviews for local diffs, branches, pull requests, or specific repository directories. The skill checks CLI prerequisites, selects an appropriate review scope, captures CodeRabbit's plain-text output to a report file, and presents human-readable findings in review form.
Use when:
- Reviewing local code changes or branch changes
- Running CodeRabbit on a repository or specific directory
- Checking code quality, security, performance, or bug risk
- Summarizing CodeRabbit findings for a pull request
- Fixing actionable issues reported by CodeRabbit
Covers:
- CodeRabbit CLI and authentication checks
- Working tree, committed, base branch, and base commit review scopes
--dirreviews for specific Git repositories- Plain-text detailed review output by default, with
--agentreserved for explicit machine-readable workflows - Full review capture to a text artifact before summarization to avoid truncated terminal streams
- Severity-based findings with code-review output guidance
- Safety notes for secret exposure, authentication tokens, and untrusted output
Included files:
- SKILL.md - agent instructions and workflow
Create, validate, rewrite, and explain Git commit messages using an opinionated Conventional Commits format. The skill can stage repository changes, inspect the staged diff, derive a commit message, validate the header, and create exactly one commit.
Use when:
- Committing current repository changes
- Validating an existing commit message
- Rewriting a commit message to match Conventional Commits
- Explaining why a commit message does or does not satisfy the local convention
Covers:
- Typed Conventional Commit headers
- Optional scopes and breaking-change markers
- Body and footer guidance
- Issue footer placement
- Merge, revert, and initial commit special cases
- Optional deterministic header validation with Python 3
Included files:
- SKILL.md - agent instructions and workflow
- scripts/validate_commit_header.py - header validator
- references/convention.md - detailed examples and versioning notes
Review unmerged branch changes against the repository default branch for KISS and DRY issues. The skill focuses on overcomplicated code, unnecessary abstraction, duplicated changing knowledge, and repeated implementation details that can drift before a branch is merged.
Use when:
- Reviewing branch or pull request changes for KISS and DRY concerns
- Checking unmerged changes against the repository default branch
- Finding over-engineered code, avoidable indirection, or premature abstraction
- Finding duplicated business rules, mappings, constants, validations, or test setup
- Producing a maintainability-focused code review with actionable fixes
Covers:
- Git-based default branch discovery or an explicit base branch
- KISS checks for unnecessary complexity, control flow, public surface area, configuration, and test setup
- DRY checks for duplicated change-prone knowledge across code, tests, schemas, build logic, and documentation
- Guardrails against recommending premature abstraction when local duplication is clearer
- Separate KISS and DRY passes for large or risky diffs
- Validation and prioritization guidance to reduce speculative findings
- Severity-based review output with file and line references
Included files:
- SKILL.md - agent instructions and workflow
Draft, improve, rewrite, and review pull request or merge request descriptions from repository changes, branch diffs, commit lists, provided patches, or existing drafts. The skill uses an opinionated template that separates summary, motivation, modifications, result, and validation.
Use when:
- Drafting a PR or MR description from local changes
- Improving an existing PR or MR message
- Turning a diff, branch, or commit list into reviewer-friendly context
- Reviewing whether a PR/MR template has the right sections
Covers:
- Diff and commit inspection guidance
- Motivation, implementation, result, and validation separation
- Optional screenshots, risks, rollout, and follow-up sections
- Copy-ready output guidance
- Template variants for documentation-only and small changes
Included files:
- SKILL.md - agent instructions and workflow
- references/template.md - template variants and examples
Install this skill collection with the Agent Skills CLI:
npx skills add ankitgoyalio/code-skillsSkills are automatically available once installed. The agent will load the full skill instructions only when a relevant task is detected.
Run a detailed CodeRabbit review of my uncommitted changes and summarize the captured report
Run a detailed CodeRabbit review of this branch compared with main and include the report file path
Run CodeRabbit against ~/Developer/fsm-ios and give me a human-readable review from the saved report
Commit my changes
Validate this commit message: feat(api): add user search
Rewrite this commit message using our convention
Review my unmerged changes for KISS and DRY issues
Check this branch for over-engineering and duplicated logic before I open a PR
Draft a pull request message from my current changes
Improve this merge request description
Each skill directory contains:
SKILL.md- required instructions and metadata for the agentscripts/- optional helper scripts for deterministic checks or automationreferences/- optional supporting documentation loaded only when needed
Keep SKILL.md focused and move detailed examples, long rule lists, and
background material into references/ so agents can use context efficiently.
When adding or changing a skill, update this README in the same change so the skill catalog, installation notes, and usage examples stay accurate.
MIT