Makes agent reasoning visible, surfaces tradeoffs, and fades help so humans build judgment. Use when reviewing or learning from agent-written code.

Trace execution paths from entry points through the code knowledge graph. Shows call chains with criticality scores and generates Mermaid flowcharts.

Search GitHub for existing implementations of a topic. Use when the user wants to find code examples, libraries, or implementation patterns. Standalone or as part of a full research session.

Analyze current context and recommend compression strategies. Use when context feels bloated, before major task transitions, or when quota is burning fast. location: user token_budget: 150. Do not use when context-optimization already invoked or simple queries.

Guide Claude through computer use API integration - screenshot capture, mouse/keyboard control, agent loops, and sandboxed desktop automation. Use when: computer use, computer control, desktop automation, screenshot capture, mouse control, keyboard control, agent loop, GUI automation, display toolkit, phantom.

Add or edit knowledge annotations. Capture tribal knowledge, business context, and rationale that cannot be inferred from code.

Generate Mermaid data flow diagrams showing how data moves between components. Use for tracing requests and API call chains.

Generate dependency graphs showing import relationships between modules or plugins

Verifies a package exists before install, defending against hallucination and slopsquatting. Use when adding, recommending, or installing a package.

Scan community discussion channels (HN, Lobsters, Reddit, tech blogs) for experience reports and opinions on a topic. Use when the user wants community perspectives on a technology or approach.

Import external documents (PDF, DOCX, PPTX, XLSX, HTML) into markdown for editing by doc-generator. Converts the source, preserves structure, and prepares for remediation. document import, conversion, ingestion, external documents Use when: user provides a document to convert and clean up DO NOT use when: web articles (use WebFetch directly), academic papers (use tome:papers), content already in markdown (use scribe:doc-generator remediation mode)

Universal document-to-markdown conversion with tiered fallback. Wraps MCP markitdown when available, degrades to native Claude Code tools, then informs the user. document conversion, markitdown, pdf, docx, pptx, format Use when: any workflow needs to convert files or URLs to markdown for LLM consumption DO NOT use when: content is already markdown or plain text

Compute DORA delivery-performance metrics from local git and GitHub API. Classify tier and surface bottleneck.

Detect friction signals during execution and graduate recurring patterns into rules. Use for session retrospectives and learning pipeline automation.

Build or incrementally update the code knowledge graph for a codebase. Uses Tree-sitter for multi-language AST parsing and stores nodes/edges in SQLite.

Post-implementation audit that justifies every change against a minimal-intervention standard. Detects AI additive bias, test-logic tampering, and unnecessary complexity. Use after completing work, before committing or creating PRs.

Probe memory/summary clarity via dual anchor questions: task progress, info gaps. Use when verifying session state or summary before handoff or compression.

Analyze and improve the improvement process itself. Implements metacognitive self-modification from the Hyperagents paper (Zhang et al., 2026): the meta agent modifies how future improvements are generated. Auto-triggers on regression detection or low effectiveness rate.

Generate visual diagrams of palace structure, knowledge topology, and synapse connectivity.

Curate the web-capture index. Use when the capture backlog grows, captures sit unprocessed at seedling/pending, or to surface stored research during work.