tylerburleigh

Task preparation skill for spec-driven workflows. Reads specifications, identifies next actionable tasks, and creates detailed execution plans. Use when ready to implement a task from an existing spec - bridges the gap between planning and coding.

Validate SDD JSON specs, auto-fix common issues, generate detailed reports, and analyze dependencies.

Targeted query capabilities for machine-readable codebase documentation with cross-reference tracking, call graph analysis, and workflow automation. Enables fast lookups of classes, functions, dependencies, and function relationships without parsing source code.

LLM-powered documentation generation for narrative architecture docs, tutorials, and developer guides. Uses AI consultation to create contextual, human-readable documentation from code analysis and spec data.

Plan-first development methodology that creates detailed specifications before coding. Use when building features, refactoring code, or implementing complex changes. Creates structured plans with phases, file-level details, and verification steps to prevent drift and ensure production-ready code.

AI-powered PR creation after spec completion. Analyzes spec metadata, git diffs, commit history, and journal entries to generate comprehensive PR descriptions with user approval before creation.

Comprehensive pytest testing and debugging framework. Use when running tests, debugging failures, fixing broken tests, or investigating test errors. Includes systematic investigation workflow with external AI tool consultation and verification strategies.

Review implementation fidelity against specifications by comparing actual code to spec requirements. Identifies deviations, assesses impact, and generates compliance reports for tasks, phases, or entire specs.

Quick preliminary research on a supplement or question. Searches for promising evidence that might warrant full deep research. Outputs a structured scouting brief to wiki/scouting/ — kept separate from the main wiki until promoted through deep research and ingest. Use when: "is X worth researching deeper?", "what's the preliminary evidence for X?", "$scout taurine cognitive", "should I research quercetin?". Do NOT use for comprehensive research (use foundry-research deep-research instead) or for wiki ingestion (use wiki-ingest).

Ingest a source into the supplements wiki, or re-audit a past ingest. Use for foundry-research reports such as "ingest the sulforaphane research" or "$wiki-ingest sulforaphane"; arbitrary sources in raw/ such as "ingest the source at raw/foo.pdf" or "$wiki-ingest raw/foo.md"; and audit-only mode such as "$wiki-ingest --audit-only sulforaphane" or "what did we miss in the sulforaphane ingest?" Orchestrates staging, pre-check, repo-local extractor and auditor role instructions, and gap-report append. Does not commit to git; the human reviews first.

Health-check the supplements wiki structurally and conceptually. Use when the user says "lint the wiki", asks whether the vault is clean, asks for maintenance findings, or wants low-risk structural fixes. Runs deterministic lint first, then performs targeted judgment-dependent review. Does not commit to git.

Answer substantive questions from the supplements wiki and optionally file reusable answers into wiki/queries. Use for practical supplement questions, comparisons, stack decisions, evidence questions, mechanism questions, safety checks, and "what does the wiki say about..." prompts. Does not commit to git.

Make a scoped corrective edit to existing supplement wiki pages when the problem is already known. Use for audit findings, attribution mismatches, broken source links, stale status fields, scope qualifiers, metadata drift, or other narrow page issues that do not require a full re-ingest. Does not commit to git.

Run a periodic supplements wiki review. Use when the user says "run a wiki review", "monthly review", "quarterly review", "review the supplements vault", or asks what needs attention beyond a single ingest/query.

Ingest a source into the supplements wiki, or re-audit a past ingest. Use for two source types: (1) foundry-research reports — "ingest the sulforaphane research", "/wiki-ingest sulforaphane"; (2) arbitrary sources in raw/ — "ingest the source at raw/foo.pdf", "/wiki-ingest raw/foo.md". Also handles audit-only mode — "/wiki-ingest --audit-only sulforaphane", "what did we miss in the sulforaphane ingest?". Orchestrates the pipeline — stage source, present the pre-check, run wiki-extractor, run wiki-auditor, append the gap report. Does not commit to git; the human reviews first.

Health-check the supplements wiki structurally and conceptually. Use when the user says "lint the wiki", asks whether the vault is clean, asks for maintenance findings, or wants low-risk structural fixes. Runs deterministic lint first, then performs targeted judgment-dependent review. Does not commit to git.

Run a structured periodic review (weekly, monthly, or quarterly). Orchestrates the full review workflow — pull data, validate structure, work through checklist, include bear-case/disconfirming evidence, and write the review note. Use when the user says "run a weekly review", "time for the monthly review", "let's do the quarterly", or when briefing.py flags an overdue review.

Run a weekly thesis-relevant news scan. Reads all active thesis/theme docs, runs targeted Tavily searches per thesis, checks for hypothesis triggers and disconfirming evidence, and writes a dated scan artifact to scans/. Use when the user says "run the weekly scan", "scan for news", "check thesis triggers", "what happened this week", or when the research methodology weekly cadence is due.

Research a specific item from docs/research-backlog.md using trusted sources defined in docs/research-methodology.md. Fetches evidence, summarizes findings with proper claim typing, updates the relevant thesis/theme doc, and marks the item resolved. Use when the user says "research B001", "work through the backlog", "resolve item B005", "look into [claim]", or similar.

Use when extracting facts from saved real-estate listing HTML in raw/, cleaning source-specific listing data, filling missing fields in data/heritage_sales.csv through reviewable candidates, proposing supplemental model features, or preserving evidence/conflict flags for property sale-price analysis.