Recently added

The autonomous coding loop with governance built in: GitHub issues in -> reviewed PRs out, with a fail-closed producer!=reviewer!=merger guard.

Never reconstruct where you left off. Local-first continuity for you and your coding agents — resume, decisions, and what's still in the air. CLI, MCP, and Agent Skills for Cursor, Claude Code, Copilot, and Codex.

Coral-pink octopus desktop pet that lives in your Mac. Auto-spawned by spec-conformant agent clients as a stdio MCP server; reacts to agent task state with idle / working / error / ... animations. Reuses assets from the octopus-meme skill.

Turn committed, project-sized outcomes into evidence-backed rolling-wave plans after an explicit five-question intake.

Codex adapter for Jarvis Memory, a local Markdown operational memory runtime for agents.

Think before AI thinks. Includes 8 skills.

Evidence-led architecture boundary triage before implementation.

Production-grade engineering skills for AI coding agents covering the full software development lifecycle from spec to ship.

Evolve and score executable algorithms with Monte Carlo graph search.

A skill suite for content creation, domain learning, product architecture baselines, first-cut architecture, and review-gated 3D asset production.

Evidence-based market-entry planning and human-AI team design.

The Product Design plugin is built for turning early ideas into prototypes teams can review. It starts design and build work by confirming the brief, then helps teams explore product directions, audit user flows, prototype from a live URL, and make static screenshots interactive.

Connect Codex to policy-approved team memory in Luthn Cloud.

Cross-team communication, devcontainer orchestration, and tool proxying for agent teams. Provides crosstalk, dispatch, connector, and evie-bot bridge tools via MCP.

crules-flutter——面向 Flutter 工程的独立协作规则 plugin（fork 自 crules v74-fork-base，独立演进）

Designs agent graphs — and refuses to build one when the work isn't actually wide. Runs a fake-edge test before fanning out, enforces fresh-context verifiers and a hard agent cap, and never launches a fleet without your go-ahead.

Six process skills that enforce spec-before-plan, plan-before-code, TDD, root-cause debugging, and evidence-before-claim. Stack-agnostic: per-project commands come from a .claude/subpowers.md contract.

Captures Codex App and CLI sessions into Trinity for allowlisted repositories.

Captures Claude Code sessions into Trinity for allowlisted repositories.

Zadania w firmowym TMS prosto z rozmowy z Claude'em — zakładanie, podzadania, materiały, statusy

Engineering workflow plugin: spec-to-docs (project-init) and feature lifecycle orchestration (lifecycle) with compact/gate-enforcement hooks.

Production-grade fintech engineering, UI/UX and audit skills for Claude Code. Makes Claude rigorous about financial correctness, security, concurrency, idempotency, auditability and production readiness.

Reproducible systematic-search and screening workflow with explicit eligibility criteria, stable record deduplication, and an auditable decision trail.

Reproducible public-database retrieval and exploratory analysis for tables and datasets without unrelated document-conversion tooling.

Research-image inspection with explicit source identity, transformation history, measurement definitions, exclusions, and interpretation boundaries.

Testable hypothesis development and reproducible experimental or observational study design without unrelated literature or manuscript tooling.

Traceable qualitative-text analysis using an explicit codebook, coded excerpts, reflexive notes, disagreements, and evidence-bounded thematic synthesis.

Repeatable monitoring of new scholarly publications with explicit queries, stable deduplication, checkpoints, and source-linked update digests.

Literature synthesis, evidence-bounded academic writing, editable research diagrams, and manuscript review without installing study-design procedures.

Compatibility entry point for the complete research cycle; installs the focused research-design and literature-publication packs as declared dependencies.

Reproducible conversion of supported research documents into inspectable Markdown without installing database retrieval or dataset analysis tools.

Compatibility entry point for mixed document-and-data projects; installs focused document-evidence and structured-data-analysis packs as dependencies.

Statistical analysis, power and sample-size planning, uncertainty and units, and publication-grade scientific visualization for quantitative research.

Domain checks for life-science protocols, biological controls, bias, biosafety boundaries, reporting context, and image acquisition metadata.

A non-published reference pack that demonstrates the Evidence Lab package structure, routing evaluations, deterministic scripts, and review metadata.

Agent-first research foundations, fast researcher onboarding, reproducible pack selection, source verification, citation handling, and scientific critical thinking.

Morrison-Lab AI agent skills, rules, commands, and lifecycle hooks for Codex.

A read-only Takealot shopping assistant for product research and balanced buying guidance.

Growmax team Claude Code skills and workflows. Auto-discovers skills/, commands/, agents/, and hooks/. Ships: github-repo-analyzer (skill); terse-mode (skill — low-token terse reply style, in-house alternative to caveman-style compressors; trigger with 'terse mode'/'be terse'/'caveman mode', levels lite|full|off); learn-app + product-manager (per-repo product notebook — /learn-app builds and maintains docs/product/ from a code census + optional read-only walk, with an async question ledger the human answers in-file; the product-manager agent answers questions and writes specs from the notebook with per-claim provenance, never inventing product truth); feature-review (end-of-development multi-agent review — TDD compliance, contextual-creation UX, UI-standards compliance (compose-never-fork + tokens, per .claude/UI-STANDARDS.md), 10k-row scalability + security sweep, advisory architecture ideas — one verified scorecard with a per-finding HIGH/MEDIUM/LOW confidence report, a run ledger, and a periodic-calibration process for improving precision over time); ux-audit + ux-migrate (census UI drift against the repo's UI standard into an approvable backlog, then pay it down autonomously by composing shared components — mirrors e2e-map + e2e-batch; a write-time hook nudges the mechanical rules for free); ui-builder (pre-build shared-component resolver — /ui-builder maps a UI request to the exact shared component per .claude/UI-STANDARDS.md Table 1 with a reuse/extend/new verdict BEFORE code is written; ships the ui-builder-resolver agent and a prompt-time reminder hook, gated to repos that ship the rulebook); app-cartograph (map a live app into durable context — navigation + business flows + knowledge graph — via a code-seeded frontier and a write-safe human-in-the-loop walk; runs upstream of E2E); e2e-flow (gated multi-agent E2E test-generation for a single flow, web + api); e2e-map + e2e-batch (whole-app flow census → one approval → autonomous suite generation); and the bug-fix pipeline: confirm-bug (turn a raw bug report into a human-confirmed RED reproduction — read-only diagnosis with a cross-surface divergence table and ranked root causes, a human ruling on correct behavior, then a deterministic env-gated failing spec + repro/BUG-id/ contract that is frozen as the grader; reports UNREPRODUCIBLE instead of faking a red), fix-bug (fix the root cause against that frozen grader in a fresh session, then have an independent read-only validator grade it — 'fixed' requires the runner green AND the named assertion actually passing AND the repro byte-identical to its confirmed commit; on PASS the repro is promoted out of its env gate into a permanent regression test), bugfix (the whole lap as ONE command — a thin chain of confirm-bug into fix-bug, same three interrupts; safe because context isolation lives in the subagents and grader integrity in the pushed tag, not in session boundaries), and validate-fix (the same grader standalone, for a fix you didn't watch or as a CI check — default FAIL, and an edited repro fails regardless of how green the run is). The pipeline holds a 3-interrupt budget per bug (ruling+strategy · confirm-red · ship): a risk tier tunes rigor instead of adding gates, repros are matrices (boundary + cross-tenant rows, not one case), and a mechanical mutation check proves the repro would fail without the fix — assurance that costs no human attention. The graders are themselves adversarially evaluated: regression/bugfix-pipeline/ ships a zero-dependency fixture app and four eval cases — reproducible, unreproducible, a counterfeit fix branching on the repro's env gate, and a repro tampered after confirmation — plus a protect-repro PreToolUse hook (self-tested) and a pushed confirmation tag (repro-BUG-id) as the validator's tamper-proof baseline. gmax (the agentic-development operating layer): /gmax-setup installs 6 personas + 7 skills + standards + workflow.config.md into any project non-destructively — one gated spine (understand → design → plan → build waves → senior review) with three human gates; detects the project's EXISTING knowledge base and points at it, never creates one; /gmax-update refreshes only the canonical files via .gmax-version diffing, never touching project-filled config, standards, or KB; harness shims for Claude Code + OpenCode, zero-shim for Hermes; regression-covered at regression/gmax-setup/.

LinkStart v1 Preview integration: one link-start workflow reconnects an interactive HTML or localhost app to the same Claude Code session or Codex thread that produced it, using the exact bundled Runtime, private session context, and one-call arm/respond monitor flow. Opt-in.

Auto-inject ArcadeDB graph context per project; slash commands for decisions, queries, indexing, and status. Pre-bundled hooks, zero runtime npm deps.

Ad-hoc, read-only-by-default queries against the SQL Server database of any .NET application via sqlcmd on Windows. Finds the connection string itself, including legacy .NET Framework, EF6/EDMX, LocalDB, and encrypted-config cases.

Turn a raw, half-formed task description into a high-quality prompt by hunting the user's blind spots first: a codebase/domain reconnaissance pass plus a structured interview, then a calibrated, ready-to-paste prompt.

Extract what the user actually wants instead of what they think they should want, through a one-question-at-a-time interview until ~95% confidence about the underlying intent. Runs before any plan, spec, or code exists.

Refine raw ideas into sharp, actionable concepts through structured divergent and convergent thinking — expand options before converging, stress-test assumptions before committing to a plan.

Turn any folder of files (code, docs, papers, images, videos) into a persistent, navigable knowledge graph with community detection, god nodes, and query/path/explain tools. Outputs interactive HTML, GraphRAG-ready JSON, and a plain-language report.

Find a better implementation of a hot, pure function with OpenEvolve, driven by the Claude Code CLI (no API key). Includes the judge-design rules (frozen reference, seeded fuzzing, speedup ratios) that keep evolution honest. Requires OpenEvolve installed locally.

Subject every non-trivial decision to a fresh-context adversarial review before it stands. For when correctness matters more than speed: unfamiliar code, production, security-sensitive logic, irreversible operations.

Work with Azure DevOps work items and pull requests through the az CLI — read, create, link, WIQL queries, PRs — with the org/project derived from the git remote and a list of hard-won pitfalls. Requires the az CLI with the azure-devops extension.

Organizational Agent Architecture (OAA). Structure multi-agent systems the way companies are structured — agents fill roles, roles require skills, skills require tools, bounded authority composes down the chain. Bundles two skills (org-agent-architecture, oaa-ontology-design) and two MCP servers (harness for compile/validate/run, ontology for domain memory).