AI & ML

andreasbloomquist Skill

Use when adding tests or evals for an LLM agent (LangGraph / langchain 1.x), or when "the tests pass but the agent regressed", "CI is paying for real model calls", "the event streamed but the panel was blank", or a tool-call trajectory must be asserted without real APIs. Pinned to langchain-core 1.6.3 / langchain 1.4.2 / langgraph 1.2.11 for the fake-model recipe.

andreasbloomquist Bundle

Use when writing or debugging code that streams a LangGraph agent to a browser over AG-UI — emitting structured payloads from tools, mapping LangGraph events to AG-UI frames, subscribing in the browser, or diagnosing "the event never arrived" — pinned to ag-ui-protocol 1.0.0 / ag-ui-langgraph 0.0.45 / @ag-ui/client 1.0.0. Model recall of this protocol is stale (pre-1.0 THINKING_* events, copilotkit_* names); read this instead.

andrewdryga Skill

Engineer a change in the Go modules (runner/ or mcp/) the house way — read the project AGENTS.md first, match the Go house style (slog, fmt.Errorf %w, stdlib table-driven tests, no new deps), honor the runner's security posture, and run the gate. Use for any code change in runner/ or mcp/ (the Go analog of the portal Elixir skills).

andrewdryga Skill

Root-cause a crash, exception, stacktrace, failing test, or wrong behavior in portal/ — reproduce, read the real error, trace to the actual cause, propose the minimal fix. Use when something raises/fails/misbehaves and you need the cause, not a guess (Ecto changeset/constraint errors, recurrent job failures, LiveView/MCP errors, runner-socket issues, flaky tests).

andrewdryga Skill

Audit context boundaries and coupling in portal/apps/emisar via mix xref + greps — find the web/MCP layer reaching past contexts, one context reaching into another's Query/Schema/Changeset, and dependency cycles. Use before splitting a context, after adding cross-context calls, or to check architecture health.

mabbit143 Bundle

Use when a Head of People Ops, BizOps lead, or Internal Communications owner needs to draft and sequence an internal-only change-management communication — a re-org announcement, a tool rollout, a policy change, a benefit change, a leadership transition, a layoff, an acquisition close, or an internal product launch — and the audience is employees (not customers). Triggers on "all-hands announcement", "town-hall script", "change comms", "internal newsletter", "rollout comms", "policy change announcement", "re-org announcement", "internal FAQ", "manager talking points", "Prosci ADKAR", "Kotter 8-step", "layoff comms", "RIF comms", "internal memo". Pairs Prosci ADKAR (Awareness / Desire / Knowledge / Ability / Reinforcement) and Kotter's 8-step change model with deterministic stdlib-only Python tools to produce a sequenced touchpoint calendar, a Kotter-compliant primary announcement, an audience-segmented FAQ, and manager cascade talking points. Industry-tuned via --profile {tech-startup, scaleup, enterprise, pu

andreasbloomquist Skill

Use when writing or reviewing a tool that an LLM agent will call — a langchain/langgraph @tool, an Anthropic tool definition, an MCP tool — especially one that wraps an external API, returns data a UI will render, or can fail. Also use when a tool "works in tests but the model calls it wrong", "returns too much and blows the context", or "crashes the whole run on a 500".

andreasbloomquist Bundle

Use when writing or reviewing LangGraph / langchain 1.x agent code — create_agent, AgentMiddleware hooks, checkpointers, interrupt/Command human-in-the-loop, astream_events, ChatAnthropic — pinned to langgraph 1.2.11, langchain 1.4.2, langchain-core 1.6.3, langchain-anthropic 1.7.2. Also use when an agent "runs but the middleware didn't short-circuit", "jump_to is ignored", or an import from langgraph.prebuilt warns as deprecated.

ngrok Skill

Get authenticated and ready to use ngrok on whatever surface the task needs - agent CLI, an SDK, or infrastructure-as-code. Detects what is already installed, walks the one-time authtoken step, and routes to the right surface. Use before other ngrok tasks, or whenever an ngrok command fails with an authentication or configuration error. Use when the user says "set up ngrok", "install ngrok", "add my authtoken", "ngrok isn't authenticated", or "getting an auth error from ngrok".

andrewdryga Skill

Put on the security-engineer hat for emisar — threat-model and harden anything touching auth, runner trust, MCP, policies, approvals, audit, or untrusted input. Use when reviewing or building auth/session/MFA, the runner socket, the MCP API, policy evaluation, approval flows, audit logging, secret handling, or any code that ingests runner/LLM input. emisar IS a security product — this hat is mandatory there.

andrewdryga Skill

Investigate, contain, and permanently remediate production incidents through Emisar. Use when an operator asks an agent to investigate production, explain why a service is degraded, stop the bleed, stabilize an environment, or fix an infrastructure incident; gather evidence, test hypotheses, use the smallest authorized action, fall back to explicitly approved break-glass access only when Emisar has no viable path, turn the root cause into a source-controlled application or infrastructure-as-code fix, verify the operator's deployment, iterate until recovery, and write the incident report. An investigation-only request authorizes observation, never a production-changing or break-glass action.

robertoatila Skill

Coordinates teams of cooperating Codex agents with script-managed persistent state.

robertoatila Skill

Inspects, searches, exports, and reconstructs coding-agent session logs across platforms.

robertoatila Skill

Batches and dispatches parallel agent jobs with dependency resolution.

robertoatila Skill

Defines and executes directed acyclic graphs of interdependent agent tasks.

ngrok Bundle

Diagnose and fix common ngrok errors and connection problems - offline endpoints (ERR_NGROK_3200), agent authentication and connection failures, upstream/tunnel connection refused, and Traffic Policy errors. Reads the error code and explains the specific cause and fix. Use when an ngrok URL returns an error, a tunnel won't come up, or the agent reports a connection problem. Use when the user says "ngrok says endpoint offline", "ERR_NGROK_3200", "my tunnel isn't working", "why is my ngrok URL down", or pastes an ngrok error code.

pdparchitect Skill

Every AGENTS.md must have a sibling CLAUDE.md symlink pointing to it. Use when creating, moving, renaming or deleting an AGENTS.md or CLAUDE.md file in any directory, or when auditing agent instruction files.

robertoatila Skill

Quantizes LLMs to 8-bit or 4-bit NormalFloat for memory-efficient GPU inference.

robertoatila Skill

Constrains LLM generation to valid JSON, XML, or Pydantic schemas using FSMs.

robertoatila Skill

Builds declarative LM programs, modular RAG systems, and automated prompt optimizers.

robertoatila Skill

Accelerates transformer attention computation with IO-aware memory access and tiling.

robertoatila Skill

Half-Quadratic Quantization for fast 2/3/4-bit LLM compression without calibration data.

robertoatila Skill

Orchestrates autonomous multi-agent teams with role-based collaboration and hierarchical execution.

robertoatila Skill

Ingests, indexes, and queries heterogeneous documents for advanced RAG architectures.

pizza-bot-app Skill

Produce a formatted health report of connected MCP servers and tools.

glitterkill Bundle

Use when working in an SDL-MCP-enabled repository, including repository exploration, task context, code inspection, runtime execution, edits, or SDL-MCP tool calls.

luocfprime Bundle

Use Labtasker v2 to queue and run independent ML inference, evaluation, or experiment Tasks; migrate pipelines; design routes and Workers; inspect Task demand and Worker activity; and recover Tasks. Do not use it as a GPU allocator, cluster scheduler, workflow DAG, or artifact store.

luocfprime Skill

Develop and regression-test Labtasker's public Agent Skill with concise real-user requests, independent examiner and candidate agents, outcome checks, and bounded self-revision. Use when changing skills/labtasker, evaluating its usability, or adding feature workflow coverage; not for product unit/integration tests or ordinary documentation alone.

bennettxai Bundle

Newsletter Agent — skill

bennettxai Bundle

Brand Deal Agent — skill

robertoatila Skill

Browser Automation with agent-browser

robertoatila Skill

Diagnoses agent installation health, environment variables, and CLI tooling dependencies.

robertoatila Skill

Accelerates LLM GPU inference throughput with NVIDIA TensorRT-LLM optimized kernels.

robertoatila Skill

Constrains LLM outputs with grammars and regex for structured data generation.

robertoatila Skill

Open-source LLM observability, distributed tracing, and evaluation platform using OpenTelemetry.

macotasu Bundle

PR番号からNotionの「お知らせ」ページをWIPとして作成するスキル。 概要を自動生成し、PR内のスクショを抽出して添付、開発者コメント欄を空で残す。 Notion MCPが使える場合は直接ページを作成、使えない場合はMarkdownを書き出してユーザーに渡す。 手動で /dev-crew:release-note <PR番号> で呼び出し。