AI & ML

swyxio Skill

Use when working with AI Engineer Europe 2026 developer-facing endpoints, public schedule JSON, speakers JSON, llms.txt files, MCP access, and the local aieng CLI.

swyxio Bundle

Design or revise a multi-candidate AI media-generation workspace with reference attachment, model/provider capability routing, generation history, or explicit draft-to-publish flows. Use when those workflow concerns are central to the requested product. Do not use for individual canvas tools, ordinary image editing, one-off provider integrations, inference-runtime work, or simple image/video generation without a media workspace.

ivy00johns Bundle

Explicitly-invoked read-only code review along two independent axes — Standards (does it follow the repo's conventions + a built-in code-smell baseline) and Spec (does it faithfully implement the originating issue/contract) — run as separate sub-agents and reported side-by-side, never merged into one score. Run on request for a thorough standalone review of a set of files or a diff; not auto-triggered and not an automatic build phase. During an orchestrated build, build-time diff review is handled by the external /code-review CLI, not this skill.

ivy00johns Bundle

Orchestrator-dispatched only. Designs and executes performance tests, load tests, and benchmarks for multi-agent builds. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Orchestrator-dispatched only. Manages database schema migrations, seed data, and schema evolution for multi-agent builds. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Create expert-quality mermaid diagrams — flowcharts, sequence, state, ER, Gantt, mindmap, block-beta and more — including dense 15-30+ node architecture maps that stay readable. Use whenever the user wants to visualize a system, process, workflow, architecture, data model, timeline, or relationship, or when another skill needs an embedded mermaid diagram. Trigger on "diagram this", "draw this", "show me how X works", "map the architecture", "chart the flow", "visualize this", or any request for a technical visual — even without the word "mermaid".

ivy00johns Bundle

The front door to the whole toolkit — one reliable entry point that reads what you want, picks the RIGHT starting skill (orchestrator, plan-builder, a loop, a role or workflow skill, or skill-explorer), and launches it for you — so you stop reaching for `orchestrator` on everything just to be sure a skill fires. Use when you don't know which skill to start with, want to kick off work but aren't sure of the entry point, or just want to summon the toolkit. Confirms before anything expensive (agent swarms, autonomous loops, full builds) and fires cheap routes immediately. Trigger on: "/madness", "madness", "where do I start", "which skill do I use", "route me", "kick this off", "get me into the right workflow", "I don't know what skill", "use the toolkit", "what's the front door", "what should I run for this". It routes and hands off — it does not do the work itself and it is not a meta-orchestrator.

ivy00johns Bundle

Coordinate multi-agent Claude Code builds end-to-end: read the plan/mission, design integration contracts, dispatch role-agents in parallel, gate on QA, ship. Under ultracode (standing opt-in) or an explicit "workflow" ask, it drives the implement + verify phases with the Workflow tool — fanning out role-agents against the contracts and adversarially verifying instead of hand-spawning agents one message at a time. Use when the user mentions agent teams, parallel/swarm builds, multi-agent work, a MISSION.md file, a multi-phase mission, or splitting work across Claude sessions. Triggers on "agent team", "parallel build", "team build", "multi-agent", "swarm build", "build X with agents", "coordinate the build", "run the mission", "workflow", "dynamic workflows", "ultracode build", "orchestrate with workflows". Does NOT preempt brainstorming, planning, design-brief, or feature-dev — it picks up after those produce artifacts.

ivy00johns Bundle

Orchestrator-dispatched only. Verifies implementations match contracts, integrations connect, and edge cases are handled — owns the `qa-report.json` build gate. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Wire any project or coding agent to FreeLLMAPI — a local proxy aggregating ~30 free LLM provider tiers behind one endpoint — so you can prototype without paying for API calls. Use to switch a project off paid OpenAI/Anthropic/etc. onto free models, point an app or CLI agent at a local LLM proxy, or stand up FreeLLMAPI itself. Speaks OpenAI, Anthropic Messages, native Gemini, and Ollama wire formats, covering Claude Code, Codex CLI, Gemini CLI, Cline, Aider, Continue, Zed, and JetBrains AI. Detects the project's LLM client, ensures the proxy is running, rewires base_url + api_key behind an env toggle, and verifies with a live test call. Trigger on "use freellmapi", "switch this to free models", "point this at freellmapi", "stop paying for openai here", "free llm proxy", "prototype without api costs", "run claude code on free models", "point claude code/codex/gemini cli at freellmapi", "free models for my coding agent".

ivy00johns Bundle

Generate machine-readable integration contracts (API, data layer, shared types, events) before any implementation begins in a multi-agent build. This is orchestrator Phase 4 — contracts written here BEFORE any implementation agent is spawned. Use when authoring API contracts, OpenAPI specs, AsyncAPI specs, Pydantic models, TypeScript interfaces, JSON Schema definitions, data layer interfaces, shared type schemas, integration boundaries between agents, or domain business rules. Trigger on "write the API contract", "define the shared types", "spec out the endpoints", "create the OpenAPI", "author the contract", or when the orchestrator needs contracts authored for a plan. Bundles six templates (OpenAPI, AsyncAPI, Pydantic, TypeScript, JSON Schema, data-layer YAML) — pick the one matching the project's stack.

ivy00johns Bundle

Orchestrator-dispatched only. Sets up logging, monitoring, metrics, and alerting for multi-agent builds. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Manage context window usage, compaction strategy, and session handoffs for long-running multi-agent builds. Writes and validates structured handoff files so continuation agents can pick up cleanly. Trigger on: "compact this", "handoff this conversation", "we're running out of context", "save state", "transfer to a new session", "context is too full", "summarize this session for the next one", "continue in a fresh session".

ivy00johns Bundle

Produce a durable, agent-ready work-item brief that captures intent precisely enough for an autonomous agent to grab and finish unattended. The brief excludes brittle references (file paths, line numbers, "the file we just edited") and instead names behavioral acceptance criteria, types/signatures, and explicit out-of-scope. Pairs with Beads or any ticket tracker. Use after grill-me or plan-builder, before dispatching. Trigger on "make a work-item brief", "write the agent brief", "agent-ready ticket", "make this dispatchable", "package this for an agent", "brief this".

ivy00johns Bundle

Orchestrator-dispatched only. Generates project documentation, API docs, READMEs, and changelogs for multi-agent builds. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Plan and apply changes to an existing skill in one workflow. Reads a skill-review report (bulk or deep-dive) or inline findings, drafts an edit list with the agent's recommended answer attached, walks the edits one at a time, applies them, and re-runs lint and frontmatter checks. Use after skill-review when you're ready to ship the changes. Trigger on "apply the review", "update this skill", "fix the skill", "ship the recommendations", "edit this skill", "apply the plan", "implement the changes".

ivy00johns Bundle

Generate new SKILL.md files conforming to the ecosystem's frontmatter spec and structure conventions. Use when creating a new agent role, meta skill, workflow skill, or contract skill — anything that needs a SKILL.md scaffold. Trigger on "create a skill", "new agent", "write a SKILL.md", "scaffold a skill", "add a role to the skill ecosystem".

ivy00johns Bundle

Bootstrap and operate LLM-maintained personal knowledge bases (wikis) for any project or domain — a persistent, compounding artifact, not RAG. Four modes: setup, ingest, query, lint. Trigger on "set up a wiki for X", "build me a knowledge base", "create an llm wiki", "organize my notes", "set up my second brain", "maintain a wiki for this project", and (inside an existing wiki) "add this article to the wiki", "what does the wiki say about X", "clean up the wiki", "process this source".

ivy00johns Bundle

The outer orchestration loop over the Agent Teams shared task list: the lead loops the board until EVERY task is completed AND each passes its TaskCompleted gate, feeding idle workers via the TeammateIdle hook so the team never stalls. One pass = read the board, assign the highest-priority unblocked task to each idle teammate, gate each completed task (pass counts it; fail rolls it back), re-verify the whole board. Experimental — gated behind CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1; falls back to the orchestrator's subagent/sequential path when off. Use to drain a task board or keep the team working until done under a build. Trigger on: "loop over the task list", "drain the task board", "keep the team working until done", "agent teams loop", "orchestrator task loop", "keep idle teammates fed". A configuration of loop-controller.

ivy00johns Bundle

Generate paste-ready prompts for Claude Design (the artifact / design-canvas tool at claude.ai, NOT Claude Code) that are committed enough to skip its Q&A loop and start drawing on message one. Use when the user wants hi-fi mockups, an interactive prototype, multiple directions on a canvas, or a safe / bold / experimental comparison inside Claude. Trigger on: "build mockups in Claude Design", "design canvas prompt", "hi-fi design mockup", "interactive prototype in claude", "claude design prompt", "I want directions A/B/C in claude", "stop letting claude design ask me 15 questions", "claude design keeps asking", "compare safe / bold / experimental". Sibling skill `ui-brief` targets Claude Code production builds — this one targets the design canvas. Works for any product type.

sapientinc Bundle

Fin effectiveness model

optimuslabs-io Skill

One-time SkillsGoat setup. Default is venv + scan. Use --goat only when you intend to link the collection into agent skill dirs. Use when installing SkillsGoat, refreshing fixture links, or adding SkillsGoat to a shared repo.

ivy00johns Bundle

Help the user discover, recall, understand, and pick the right skill from the available toolkit. Names the skill; does NOT invoke it. Use when the user is trying to find a skill ("I forgot the name of the one that does X", "what was that skill called"), asking what skills exist ("what skills do I have", "list all my skills", "show me the catalog"), asking what a specific skill does ("what does X do", "explain the X skill"), asking how skills relate ("how do these connect", "what does orchestrator spawn"), or asking for routing help ("which skill for this task", "what should I use to Y"). Also trigger when the user reaches for orchestrator on something that isn't a multi-agent build, or asks any meta-question about the skill ecosystem itself.

ivy00johns Bundle

Orchestrator-dispatched only. Builds API servers, business logic, and data layers for multi-agent builds. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Orchestrator-dispatched only. Audits codebases for security vulnerabilities, reviews auth implementations, and verifies OWASP compliance for multi-agent builds. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Bundle

Wire the Claude Code agent harness to pxpipe — a local, loopback-only proxy that renders re-sent bulk (system prompt, tool docs, older history) into dense PNGs so long sessions cost roughly half to a third as much in input tokens, with a live savings dashboard and a one-line off switch. Use when the user wants to cut Claude Code's own token bill, enable the token-saver proxy for a long or multi-agent run, point the harness at pxpipe, or check whether compression is actually paying. Opt-in and reversible: ANTHROPIC_BASE_URL behind an env toggle, model scope deferred to model-adaptation's allowlist, verification that compression fired AND the prompt cache stayed warm. Trigger on "use pxpipe", "token saver proxy", "compress my context", "cut session input tokens", "optical context compression", "my Claude Code sessions are expensive". Sibling use-freellmapi rewires an app's LLM client; this one rewires the agent harness itself.

ivy00johns Bundle

Wrap any task in a verifiable stop condition plus a mandatory guardrail stack so an autonomous loop converges instead of thrashing or burning the budget — the foundation harness every loop skill composes on. Use whenever you want Claude to keep working until something is provably true (tests pass, coverage hits a target, a contract's criteria hold, a queue is empty), to schedule a recurring check, or to pick the right loop primitive (/goal vs /loop vs Stop-hook vs a bash Ralph loop vs a dynamic workflow). Trigger on: "loop until", "keep going until", "run until green", "work until done", "autonomous loop", "agentic loop", "ralph loop", "/goal", "iterate until", "loop safely", "iteration cap", "loop budget", "runaway agent", "overnight build". Read it first when authoring any new loop skill.

ivy00johns Bundle

Adapt prompts, skills, and agent scaffolding when the underlying model changes — currently the Claude 5 family (Fable 5/Mythos 5) vs Opus 4.x, plus the cross-vendor DeepSeek landscape. Stronger models need LESS scaffolding: this skill says what to PRUNE, what backfires (narrating reasoning trips a reasoning_extraction refusal), and what to add for long autonomous runs. Canonical home of the model & effort tiering policy (declared-provider ladder — Anthropic or DeepSeek; FreeLLMAPI carve-out) and the capability-handoff technique — extract an operating manual from a stronger model to run on a cheaper one. Use when migrating a skill to a new model, agents get refused, or picking model/effort per role. Trigger on "migrate to Fable", "Fable 5", "Mythos 5", "deepseek", "model migration", "reasoning_extraction", "tune effort", "model tiering", "long-running agent hygiene", "extract operating manual", "pxpipe", "image proxy".

ivy00johns Bundle

Bootstrap the per-repo config other Skill-Madness skills consume: domain doc layout (CONTEXT.md, ADRs), contract format (OpenAPI / Pydantic / TS / JSON Schema), and work-item tracker (Beads / GitHub / GitLab / local). Writes docs/agents/ plus an "## Agent skills" block in CLAUDE.md or AGENTS.md (never both, never overwriting). Downstream skills fail loud if this config is missing. Run once per repo. Trigger on "setup project skills", "configure project skills", "bootstrap this repo", "/setup-project-skills".

ivy00johns Bundle

Build a working mental model of an unfamiliar codebase by looping: fan out read-only subagents to map subsystems, synthesize, find the gaps in your understanding, and repeat until a written architecture summary answers every seed question — each answer citing where in the code it comes from. The deep-research fan-out shape applied to the LOCAL codebase (Read/Grep, no web), with a subagent-parallelism cap and a loop-until-dry stop on understanding gaps. Use when onboarding to a new repo, reverse-engineering someone else's code, or answering "I don't understand how X works here" before you touch it. Trigger on "help me understand this codebase", "onboard me to this repo", "how does X work in this code", "build a mental model of this project", "explore this codebase", "map this codebase", "I don't understand this code", "explore until I understand". A configuration of loop-controller.

ivy00johns Bundle

Orchestrator-dispatched only. Authors the cross-package dependency manifest before parallel implementation agents are dispatched, preventing transitive version drift. Composed by orchestrator during multi-agent builds. Not user-invocable.

ivy00johns Skill

Apply full-agent when you need a role skill with every documented frontmatter field populated, for testing field-stripping behavior across all 11 tools.

ivy00johns Skill

Apply minimal-agent when you need a baseline role skill with only required fields for testing convert and lint tools.

sapientinc Bundle

Start, stop, resume, inspect status, open the independent read-only foreground TUI, detect active runs, and safely repair Praxist run lifecycle boundaries from a supported agent interface. Use when the user asks to launch a Praxist task, stop or kill a run, continue/resume an interrupted run, restart from the latest safe generation, query current run progress, open or exit a live monitor, detect or list currently running Praxist tasks in the environment, inspect generation status, view incubator/frontier/leaderboard performance, check hardware load, handle interrupted PI panel or Gems reset boundaries, inspect whether a task directory is runnable, or control Praxist lifecycle commands with `praxist start`, `praxist stop`, `praxist status`, `praxist --monitor`, `praxist resume`, or `praxist resolve`.

sapientinc Bundle

Orchestrate first-use Praxist onboarding, current task initialization, validation, and detached run launch from a supported agent interface. Use when a user wants a one-command or one-conversation path from an existing runnable research project to a started Praxist run, asks to onboard plus initialize plus start, asks for repo-to-task-to-run setup, or wants the agent to prepare and launch Praxist with minimal interaction. This skill composes onboarding, full task initialization or task-harness repair, runtime checks, control start, and confirmation gates without changing Praxist core.

sapientinc Bundle

Diagnose Praxist run health, artifact integrity, research-loop completeness, generation-scoped DIG/QD, PI/Gems/frontier/incubator consistency, peer memory freshness, diversity HHI, hardware utilization, LLM/runtime friction, task harness health, sustained low-performance causes, strongest variants/Pareto front, strong-variant lineage, and human-readable run reports. Use when the user asks an agent to investigate whether a current or historical Praxist run is healthy, why progress or performance is weak, whether artifacts or promotions are missing, whether guard or resource issues are blocking peers, to produce a detailed agent behavior analysis report, to generate an A/B/C run report, or to improve or optimize a task after diagnosis using task-directory-only parameter and prompt changes. Default diagnostics are analysis-only; explicit improvement mode may stop the selected run and edit task-level configuration/prompts, but must not modify Praxist core logic.