dead-code-sweep
This skill should be used when cleaning up codebases that have accumulated dead code, redundant implementations, and orphaned artifacts — especially codebases maintained by coding agents. Triggers on "find dead code", "clean up unused code", "remove redundant code", "prune this codebase", "dead code sweep", "code cleanup", or when a codebase has gone through multiple agent-driven refactors and likely contains overlooked remnants. Systematically identifies cruft, categorizes findings, and removes confirmed dead code with user approval.
What this skill does
# Dead Code Sweep Systematic identification and removal of dead code, redundant implementations, and orphaned artifacts in codebases — particularly those maintained by coding agents with limited context windows. ## Why Agent-Maintained Codebases Accumulate Cruft Coding agents operate within narrow context windows. When refactoring, they often: - Re-implement functionality without finding and removing the original - Leave compatibility shims for interfaces that no longer exist - Abandon helper functions after inlining their logic - Create new files without deleting the ones they replace - Duplicate type definitions across module boundaries - Leave imports for symbols they stopped using mid-refactor This cruft compounds. Each orphaned artifact misleads the next agent (or human), who wastes context budget reading code that does nothing. ## Workflow ### Phase 1: Scope and Inventory Determine the analysis boundary. **If the user provides a scope** (directory, file pattern, module), constrain all analysis to that scope. **If no scope is provided**, analyze the full codebase. Start by reading the project structure: 1. Identify the primary language(s) and framework(s) 2. Map entry points (main files, route definitions, exported modules, test runners) 3. Note the build system and dependency configuration 4. Check for monorepo structure — analyze each package as a unit Produce a brief inventory: ``` Language(s): TypeScript, Python Entry points: src/index.ts, src/cli.ts Build: esbuild via package.json scripts Packages: 1 (single package) Estimated LOC: ~4,200 ``` ### Phase 2: Detection Launch parallel sub-agents to scan for different categories of dead code. Read `references/cruft-patterns.md` for the full catalog of detection patterns. Organize detection into these parallel tracks: | Track | What It Finds | |-------|--------------| | **Orphaned files** | Files not imported, required, or referenced by any other file | | **Unused exports** | Exported symbols (functions, classes, types, constants) never imported elsewhere | | **Redundant implementations** | Multiple functions/classes doing the same thing under different names | | **Stale compatibility code** | Shims, adapters, wrappers, and re-exports that bridge interfaces that no longer differ | | **Dead branches** | Conditional paths that can never execute (always-true/false guards, unreachable returns) | | **Orphaned tests** | Test files testing functions or modules that no longer exist | | **Orphaned dependencies** | Packages in dependency manifests not imported anywhere in source | For each track, the sub-agent should: 1. Read `references/cruft-patterns.md` for detection strategies specific to that track 2. Search the codebase using Grep and Glob 3. Verify each candidate by tracing references — a symbol is only dead if **zero live code paths** reach it 4. Record findings with file path, line range, and evidence **Verification is critical.** Common false positives to watch for: - Symbols used via dynamic dispatch, reflection, or string-based lookups - Framework-magic exports (e.g., Next.js page components, pytest fixtures, Rails conventions) - Public API surface intended for external consumers - Conditional imports behind feature flags or environment checks - Decorator-registered or plugin-registered handlers - CLI entry points referenced in package.json `bin` fields - CSS class names referenced in templates or JSX as dynamic strings When uncertain, mark as "needs review" rather than "confirmed dead." ### Phase 3: Report Consolidate all findings into a structured report at `.claude/dead-code-report.md`. Organize findings by confidence level, then by category: ```markdown # Dead Code Sweep Report **Scope:** [full codebase | specific path] **Date:** [date] **Estimated removable lines:** [count] ## Confirmed Dead (high confidence) ### Orphaned Files - `src/utils/old-parser.ts` — Not imported anywhere. Superseded by `src/parser/index.ts`. - ... ### Unused Exports - `formatDate()` in `src/helpers.ts:42-58` — Exported but zero imports across codebase. - ... [...other categories...] ## Needs Review (uncertain) ### Possibly Dynamic - `handleLegacyEvent()` in `src/events.ts:91` — No static imports, but may be registered dynamically. - ... ``` For each finding, include: - **File path and line range** - **What it is** (function, class, file, type, constant, dependency) - **Why it appears dead** (no imports, no references, superseded by X) - **Confidence** (confirmed / needs review) ### Phase 4: Cleanup with Approval Present the report summary to the user and ask for approval before removing anything. Use AskUserQuestion to present findings by category: ``` Found 12 confirmed dead items and 3 needing review. Confirmed dead by category: - 3 orphaned files (~280 lines) - 5 unused exports (~120 lines) - 2 redundant implementations (~90 lines) - 2 orphaned dependencies Which categories should I clean up? ``` Options: Remove all confirmed / Select categories / Review each item / Skip cleanup For each approved category: 1. Remove the dead code 2. Clean up any imports that referenced the removed code 3. Remove empty files left after cleanup 4. Remove orphaned dependencies from the manifest 5. Run the project's lint/typecheck/build commands to verify nothing broke If any removal causes a build or type error, immediately revert that specific removal and move the item to "needs review." After cleanup, update the report with what was removed and what was kept. ## Detection Principles ### Trace from entry points, not from suspects Start from known entry points and trace what's reachable, rather than starting from a suspect symbol and trying to prove it's used. The reachability approach has fewer false negatives. ### Respect the module boundary A symbol exported from a package's public API may be consumed by external code not visible in this repository. When analyzing libraries or packages with external consumers, only flag internal (non-public-API) dead code as "confirmed." Flag public API dead code as "needs review." ### Look for clusters, not just individuals Agent-generated cruft tends to cluster. When one dead function is found, examine its neighbors — the agent likely abandoned the entire section during a refactor. A dead file often has sibling dead files created in the same commit. ### Use git history as a signal When available, check when suspect code was last meaningfully modified. Code untouched across several refactor commits is more likely dead. Use `git log --follow` to trace renames and detect superseded files.
Related in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.