Pepe39
13 verified skills
.claude/skills/init
Boot the local stack for the code-review extension. Starts the Flask API on :5002 in background, installs Python deps if missing, verifies /status, and reports the log path. Use when the user says /init, "arranca todo", "boot", "start the api/server/stack".
tools
.claude/skills/run
Single entry point for code review labeling tasks. Creates new tasks or resumes existing ones.
development
.claude/skills/step-01-parse-inputs
# step-01-parse-inputs ## What it does Parses the task variables pasted by the user into `inputs.md` and populates `task_info.md` with structured data. ## Prerequisites - Task directory created by `/run` - User has pasted task variables into `inputs.md` ## Context > See `docs/steps/step0.md` for reference on input fields. ## Arguments - `id` (required): Task ID. E.g.: 2937204136 ## Instructions ### 1. Recover context Find the task: ``` find tasks/ -maxdepth 2 -type d -name "{id}" ``` Read
documentation
.claude/skills/step-02-analyze-pr
# step-02-analyze-pr ## What it does Opens the PR on GitHub, reads the title and description, and gathers high-level context about what the PR changes and why. This context is essential for evaluating the review comment accurately. ## Prerequisites - Step 01 completed (inputs parsed into `task_info.md`) ## Context > See `docs/steps/step1.md` and `docs/steps/step3.md` for reference. ## Arguments - `id` (required): Task ID ## Instructions ### 1. Recover context Find the task directory and r
documentation
.claude/skills/step-03-analyze-comment
# step-03-analyze-comment ## What it does Goes to the discussion URL, verifies the comment matches the body field, reviews the diff in depth, and optionally browses the repo for additional context. Produces a thorough understanding of what the comment is pointing out and whether its claims are correct. ## Prerequisites - Step 02 completed (PR analyzed, diff saved) ## Context > See `docs/steps/step2.md`, `docs/steps/step3.md`, `docs/steps/step4.md`, and `docs/steps/step5.md` for reference. ##
development
.claude/skills/step-04-label-quality
# step-04-label-quality ## What it does Labels Axis 1 — Quality. Determines whether the review comment is helpful, unhelpful, or wrong. ## Prerequisites - Step 03 completed (comment fully analyzed) ## Context > See `docs/axis-1-quality.md` for definitions, evaluation criteria, and examples. > See `docs/steps/step6.md` for the step-by-step process. > See `DOCUMENTATION.md` sections 8 (FAQ), 9 (Common Mistakes), and 10 (Tips) for edge cases and pitfalls. ## Arguments - `id` (required): Task ID
testing
.claude/skills/step-06-label-context-scope
# step-06-label-context-scope ## What it does Labels Axis 3 — Context Scope. Determines what level of context a reviewer would need to confidently make this comment and documents all evidence used. ## Prerequisites - Step 05 completed (Severity labeled) ## Context > See `docs/axis-3-context-scope.md` for definitions, evaluation criteria, and examples. > See `docs/steps/step8.md` for the step-by-step process. > See `DOCUMENTATION.md` sections 8 (FAQ), 9 (Common Mistakes), and 10 (Tips) for edg
documentation
.claude/skills/step-07-label-advanced
# step-07-label-advanced ## What it does Labels Axis 4 — Advanced. Derives the Advanced label from Context Scope using a deterministic mapping. ## Prerequisites - Step 06 completed (Context Scope labeled) ## Context > See `docs/axis-4-advanced.md` for the mapping rule and definitions. > See `DOCUMENTATION.md` section 7 (Axis 4) for the rationale. ## Arguments - `id` (required): Task ID ## Instructions ### 1. Recover context Read `task_info.md` — specifically the Context Scope label from s
testing
.claude/skills/step-08-generate-output
# step-08-generate-output ## What it does Compiles all four axis labels into the final JSON output, validates consistency, and generates the deliverable file ready for submission. ## Prerequisites - Steps 04-07 completed (all four axes labeled) ## Context > See `docs/steps/step10.md` for the submission checklist. ## Arguments - `id` (required): Task ID ## Instructions ### 1. Recover context Read `task_info.md` — extract all four labels and their reasoning from the Labels section. Update
development
.claude/skills/step-09-recheck
step-09-recheck
testing
.claude/skills/fix-recheck
fix-recheck
testing
.claude/skills/review
Validate deliverables for a completed task. Re-evaluates labels, reasoning, and format.
development
.claude/skills/step-05-label-severity
# step-05-label-severity ## What it does Labels Axis 2 — Severity. Assesses how severe the issue is that the review comment points out. ## Prerequisites - Step 04 completed (Quality labeled) ## Context > See `docs/axis-2-severity.md` for definitions, evaluation criteria, and examples. > See `docs/steps/step7.md` for the step-by-step process. > See `DOCUMENTATION.md` sections 8 (FAQ), 9 (Common Mistakes), and 10 (Tips) for edge cases and pitfalls. **Critical reminder:** Severity measures the
testing