Review a change for modular-monolith dependency violations, domain leakage, plugin trust, adapter coupling, premature distribution, and future API/UI compatibility.
maelrx/Zugzwang
SkillsMP has collected 20 skills from maelrx/Zugzwang. Open a skill to review its source and details.
- Latest recorded source activity
- SkillsMP catalog refreshed
- skills collected
- 20
- GitHub stars
- 5
- GitHub forks
- 0
Skills in this repository
Showing 20 of 20 collected skills.
Audit a run, strategy, tool, dataset, prompt, or report for hidden operational assistance H, knowledge assistance K, engine leakage, retries, and attribution errors.
Scaffold the Zugzwang uv monorepo after the blocking human gates are ratified. Use for M0 workspace creation, package boundaries, tooling, lockfile, and the first fake-only vertical slice.
Implement, validate, import, redact, or evolve the self-contained run bundle and its checksummed evidence/analytical artifacts.
Perform an independent, defect-first review of a Zugzwang change covering correctness, architecture, scientific integrity, persistence, security, tests, migrations, and documentation.
Create or review an operational SQLite migration and any related event/bundle upcaster with interruption, compatibility, and evidence-preservation tests.
Design or preregister a Zugzwang experiment with causal hypotheses, controls, protocol coordinates, datasets, metrics, power, budgets, promotion gates, and claim limits.
Implement or modify the standard-chess environment, state/action codecs, legal transitions, renderers, tasks, opponents, or PGN export behind the chess port.
Implement an explicit R0-R7 decision strategy, candidate/critic flow, retry policy, or model-only orchestration without hiding calls or assistance.
Implement or change a core domain value object, port, event, manifest model, or public protocol contract while preserving dependency direction and compatibility.
Implement a deterministic evaluator or versioned metric with complete provenance, cache identity, slices, uncertainty inputs, and no live-decision leakage.
Author or implement static expert KnowledgePacket support, including provenance, controls, token matching, leakage checks, and K-class assignment before dynamic RAG exists.
Implement typed image/state observations, deterministic chess rendering, multimodal provider lowering, redundancy or conflict conditions, and modality provenance.
Add or update a model provider/backend adapter with explicit capabilities, raw evidence, normalized requests/responses, errors, usage, and no hidden agency.
Implement run, episode, step, attempt, cancellation, checkpoint, budget, or resume behavior in the durable local runtime.
Prepare a concise but complete evidence packet for a pending Mestre Mael decision without selecting or implementing the decision on the operator’s behalf.
Audit and prepare a Zugzwang foundation, alpha, or scientific release across gates, tests, migrations, bundles, licensing, SBOM, documentation, compatibility, and claims.
Design and execute fault-injection, interruption, idempotency, migration, or replay tests for the runtime, persistence, provider and bundle boundaries.
Research and integrate a new chess/LLM paper, benchmark, dataset, model, or relevant engineering source using primary evidence and explicit epistemic status.
Create, amend, supersede, or ratify an Architecture Decision Record with alternatives, direct/indirect impacts, reversibility, evidence, owner, and human gate linkage.