skill

Use when designing or implementing a premium animated landing page with AI-generated visuals or looping video, especially when the workflow involves Higgfield-style image and video generation, Google AI Studio-style code prompting, React or Tailwind landing-page implementation, dark-mode video blending, scroll-tied playback, aspect-ratio-matched media blocks, or motion-first section design.

Use when the user asks to fine-tune on books, create an SFT dataset from books, train a style-transfer or author-voice model, extract ePub text, segment long-form book content, or prepare literary data for LoRA or small-model training.

Research the codebase, turn locked decisions into a plain-English phase plan, prepare the current phase for execution, and validate it through a mandatory gate before any code is written. Covers discovery, synthesis, phase planning, contract/story/bead creation, and the full validation gate (structural verification across 8 dimensions, spike execution for HIGH-risk items, bead graph polishing with bv analytics, and explicit user approval). Use after anionzo:exploring completes. Reads CONTEXT.md, retrieves institutional learnings, runs discovery and synthesis, writes discovery.md, approach.md, phase-plan.md, and then writes current-phase contract/story artifacts plus beads for that phase only. No execution begins until the validation gate passes and the user explicitly approves.

Post-execution quality verification skill for the anionzo ecosystem. Invoke after the final phase swarm completes. Runs 5 parallel specialist review agents, 3-level artifact verification, human UAT, and finishing (PR, cleanup, epic close). Review issues become beads instead of per-finding markdown files; P1 still blocks merge while P2/P3 become non-blocking follow-up beads. Absorbs finishing responsibilities and hands off to compounding.

Orchestrates parallel worker agents for phase execution and defines the worker implementation loop. Use after the anionzo:validating skill approves the current phase for execution. Initializes the overseer/orchestrator context, spawns bounded worker subagents, monitors Agent Mail for completions/blockers/file conflicts, coordinates rescues and course corrections, and hands off either to planning for the next phase or to reviewing after the final phase. The orchestrator TENDS — it never implements beads directly. Workers self-route from the live bead graph, implement beads atomically, and report back through Agent Mail.

Bootstrap meta-skill for the anionzo agentic development ecosystem. Load first on any anionzo project. Lists all 9+2 skills with routing logic, session scout/bootstrap, small-change vs standard-feature vs high-risk mode selection, go mode (full-auto pipeline with 4 human gates), priority rules, and state resume. Invoke when starting a new session, choosing which skill to use, running the full pipeline end-to-end, or resuming after a handoff.

Use when starting a new session, when the task shape is unclear, when a request mixes multiple intents, or when you need to route to the right skill and working mode. This is the router — load it first, then hand off to the correct skill.

Use when creating a new anionzo skill, editing an existing anionzo skill, or verifying a skill works under pressure before deploying. Use when you need an agent skill to be bulletproof against rationalization. Do NOT use for project-specific AGENTS.md conventions or one-off solutions.

Comprehensive research skill for codebase exploration, repository onboarding, prompt strengthening, and anti-reinvention discovery. Use for: (1) quick-search: targeted lookup in a known repo; (2) repo-bootstrap: onboard to an unfamiliar repository; (3) deep-scout: research-first feature discovery — map repo stack, find local reuse, check upstream patterns via deepwiki, verify official docs via exa before planning or implementing; (4) prompt-upgrade: strengthen a rough prompt into an execution-ready instruction set; (5) codebase-intel: architecture snapshot using gkg MCP tools.

Refine fuzzy requests into concrete directions through ideation, spec-driven development, Socratic exploration, and decision locking. Three modes: quick (explore and lock direction only), spec (full brainstorm with FR/NFR/AC/Scenarios), and deep-explore (Socratic dialogue, domain classification, locked decisions, optional CONTEXT.md output).

Review code changes with a risk-first, multi-perspective approach using severity-based triage. Covers both giving reviews and receiving feedback. Includes a verification gate to prevent false completion claims.

Systematic debugging for blocked workers, test failures, build errors, runtime crashes, and integration issues. Invoked standalone ("debug this error") or by other skills (reviewing spawns debugger on UAT failure, executing invokes it on blocker). Reads history/learnings/critical-patterns.md to avoid re-solving known issues. Writes debug notes that compounding can later capture.

Write, update, and maintain documentation so it reflects the system's actual current state. Supports targeted section editing, full document creation, architecture diagrams, and prompt refinement for docs-refresh tasks. Use when writing README files, onboarding docs, runbooks, API usage notes, architecture diagrams, or any documentation that should stay in sync with code.

Extract reusable patterns, decisions, and failure learnings from completed work into documentation that compounds knowledge over time. Covers four modes: in-progress session handoff (handoff), durable learning extraction from a completed task (extract), post-merge deep compounding with three parallel subagents writing a dated learnings file (compound), and a manual consolidation pass over Codex artifacts and accumulated learnings (dream). Trigger phrases: extract learnings, document what we learned, capture this pattern, what did we do so far?, summarize where we are, compound, lessons learned, dream pass, consolidate learnings, anionzo:compounding skill. Key output: critical-patterns.md is read at the start of every planning and exploring Phase 0 — this is the flywheel that makes the ecosystem smarter over time.

Delivers code changes using one of three discipline modes — standard feature delivery (inspect-first, smallest-approach), test-driven development (red-green-refactor, failing-test iron law), or refactor-safe (behavior-preserving restructure with verified safety net). Select the mode that matches the nature of the change.