claude-optimize

Full environment health check with scored dimensions

Token/context optimization - reduce bloat and waste

Reflect on usage patterns to discover new capabilities and prune unused ones

Show all optimize modes and usage

Hook analysis - health, coverage gaps, recommendations

Use this agent to analyze codebase alignment and skill quality when running /optimize:report. Returns scores for codebase_alignment, skill_quality, and claude_md_quality dimensions. Do NOT use for general coding tasks.

Use this agent for token counting and context efficiency analysis when running /optimize:context or /optimize:report. Measures CLAUDE.md token load, skill description sizes, MCP tool counts, and compaction resilience. Returns efficiency metrics. Do NOT use for general file analysis or coding tasks.

Use this agent to generate typed improvement proposals from session reflection data when running /optimize:discover. Receives usage analysis, environment inventory, codebase profile, and gap/bloat analysis. Returns prioritized proposals with security assessments, evidence citations, and implementation previews following skill-creator and hookify best practices. Do NOT use for general coding tasks or other optimize modes.

Use this agent to analyze hook health, memory hygiene, and MCP configuration when running /optimize:report. Returns scores for hook_coverage, memory_hygiene, and mcp_health dimensions. Do NOT use for general coding tasks.

Use this agent to aggregate results from multiple optimization analyses into a single weighted report when running /optimize:report. Merges scores from security, context, hook, skill, memory, and MCP dimensions with prioritized action items. Do NOT use for general reporting or coding tasks.

Use when running /optimize:init, /optimize:audit, or /optimize:context to assess CLAUDE.md quality, generate templates, and suggest improvements. Scores files using a weighted rubric covering commands, architecture, patterns, conciseness, currency, and actionability. Do NOT trigger on general file editing or coding tasks.

Use when running /optimize:init or /optimize:audit to detect project type, language, framework, package manager, CI/CD system, and existing Claude Code configuration. Triggers on codebase analysis for optimization, project detection for Claude Code setup, or environment assessment. Do NOT trigger on general coding tasks.

Use when running /optimize:context to analyze and reduce token consumption in Claude Code configuration. Measures CLAUDE.md token count, skill description bloat, MCP tool count, and recommends compaction strategies. Triggers on context optimization, token reduction, or CLAUDE.md bloat analysis. Do NOT trigger on general coding tasks.

Use when running /optimize:discover to cross-reference session usage data against the current environment inventory. Detects gaps (repeated manual work that could be automated), bloat (unused skills, MCP servers, agents), and calculates tooling balance. Do NOT trigger on general coding tasks or other optimize modes.

Use when running /optimize:hooks to analyze existing hooks for health issues, detect missing hooks based on project type, and recommend new hooks with implementations. Triggers on hook analysis, hook creation for optimization, or hook health checking. Do NOT trigger on general hook development or coding tasks.

1 hook across 1 event