RE

repository-cleanup-campaign

Orchestrates multi-step cleanup tasks to reduce technical debt through systematic file discovery and bounded edits.

Install

mkdir -p .claude/skills/repository-cleanup-campaign && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/18498" && unzip -o skill.zip -d .claude/skills/repository-cleanup-campaign && rm skill.zip

Installs to .claude/skills/repository-cleanup-campaign

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Plan and execute broad workspace cleanup campaigns when repository cleanup hygiene campaign or many warnings keywords appear.
125 chars✓ has a “when” trigger
Advanced

Key capabilities

  • Plan broad workspace cleanup campaigns
  • Reduce repository warnings
  • Remove stale artifacts
  • Coordinate multi-crate cleanup efforts
  • Build a dependency queue for write workers
  • Apply changes in small batches with validation

How it works

This skill converts broad cleanup intentions into a safe campaign by coordinating read-only discovery, bounded edit batches, and validation across a repository.

Inputs & outputs

You give it
Broad cleanup intent or trigger keywords
You get back
A concise plan or patch summary, file paths, symbols, commands, and validation results

When to use repository-cleanup-campaign

  • Reducing repository warnings
  • Removing stale project artifacts
  • Executing large-scale refactoring

About this skill

repository-cleanup-campaign

When to use

Use for multi-crate cleanup, warning reduction, stale artifact removal, broad quality campaigns, or requests that need several smaller cleanup skills coordinated. Also use it when another agent, review comment, CI failure, or maintainer note uses those trigger words. Load it explicitly with /skill repository-cleanup-campaign before planning substantial work, and mention it in /agent <name> missions that need the same guardrails.

Purpose

Convert broad cleanup intent into a safe campaign with read-only discovery, bounded edit batches, validation, and final consolidation. The skill is written for Copilot CLI workflow: prefer Rust code intelligence or rust-lsp for semantic questions, then glob, grep, and view for file discovery, powershell for cargo commands on Windows, sql for dependency queues, GitHub tools for PR/CI evidence, and the agent/task tool only when parallel analysis or bounded write missions add value.

Process

  1. Start with /skill task-scope-bounding and create a task slug for .work/copilot-cli/<task-slug>/ reports if artifacts are needed.
  2. Dispatch read-only /agent workers by independent area: dead code, boundaries, duplicates, naming, tests, and manifests.
  3. Build a dependency queue so write workers never collide on the same files.
  4. Apply changes in small batches and run the narrowest meaningful validation after each batch, then full canonical gates when feasible.
  5. Produce a final summary with changed files, removed artifacts, retained risks, and commands run.

Expected output

  • A concise plan or patch summary tied to the trigger that loaded this skill.
  • File paths, symbols, commands, and validation results that a reviewer can reproduce.
  • If workers were used, a consolidated handoff that names each /agent <name> mission, its report path, and any conflicts resolved.
  • Clear blockers when a command, tool, or reference is unavailable; do not turn missing evidence into a success claim.

Reference docs

  • docs/architecture/REPOSITORY_ARCHITECTURE.md
  • docs/testing/CI_GATES.md
  • docs/reference/DOCUMENTATION_STANDARD.md
  • docs/README.md
  • docs/architecture/ENGINE_OVERVIEW.md
  • docs/project/ANDROMEDA_DOCTRINE.md

Guardrails

  • Do not wander outside the requested campaign.
  • Do not combine cleanup with feature delivery.
  • Do not let parallel workers edit overlapping files.
  • Keep changes bounded to the requested owner and avoid unrelated cleanup.
  • Preserve Andromeda's typed Procedure execution path, explicit codecs, durability, security admission, and deterministic testing expectations.

Andromeda baseline

Rust 1.95.0 / Edition 2024 / resolver 3, QUIC + custom Protobuf RPC, no gRPC, no JSON native protocol, procedure-only surface, WAL-before-commit, GPU outside commit/recovery/security path.

When not to use it

  • When combining cleanup with feature delivery
  • When parallel workers are allowed to edit overlapping files
  • When wandering outside the requested campaign scope

Limitations

  • Do not wander outside the requested campaign
  • Do not combine cleanup with feature delivery
  • Do not let parallel workers edit overlapping files

How it compares

This skill provides a structured, guarded approach to large-scale repository cleanup, ensuring safety and reproducibility through dependency queues and batch validation, unlike ad-hoc cleanup efforts.

Compared to similar skills

repository-cleanup-campaign side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
repository-cleanup-campaign (this skill)02moNo flagsAdvanced
tdd-migrate16moReviewAdvanced
tmux202moReviewIntermediate
jira116moNo flagsBeginner

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry