Lincyaw

Structured procedure to locate the root cause of a confirmed bug. Uses keyword search → focused reading → call-graph tracing. Outputs a specific file + line + hypothesis.

Step-by-step procedure to reproduce a bug. Use this FIRST before trying to fix anything. Produces a minimal script that demonstrates the failure.

AgentM SDK development guide — manifest-as-agent-unit philosophy, SDK programmatic invocation, dynamic workflow orchestration, atom contract, Operations abstraction, event system, service communication, CLI conventions, scenario authoring, logging, structured output, and config resolution. Use whenever writing, editing, or reviewing code under src/agentm/ (atoms, core, gateway), contrib/scenarios/ (manifests), or contrib/extensions/ (workspace-member atoms). Also trigger when creating new atoms, modifying MANIFEST declarations, registering tools or events, touching FileOperations / BashOperations / ResourceWriter, writing CLI subcommands, configuring model profiles, spawning child sessions, building multi-agent orchestrators, writing workflow scripts, or when a code change looks like it might bypass the SDK's existing abstractions. If you catch yourself about to write raw os.stat / open() / subprocess.run in an atom, importing openai/anthropic directly instead of going through the provider layer, or shelling

Guide for adding new benchmarks to the agentm-eval framework and running experiments through the unified `agentm eval` CLI. Covers the adapter interface, experiment lifecycle, result schema, ClickHouse linking, and output conventions. Use whenever creating a new eval adapter, modifying an existing benchmark integration, writing code under contrib/evals/src/agentm_eval/, discussing experiment management, or when the user mentions "add a benchmark", "new eval", "接入评估", "新benchmark", "实验管理", "eval adapter". Also trigger when reviewing eval-related code or debugging experiment output structure.

How to write workflow scripts that orchestrate multiple agents in AgentM. Covers agent(), parallel(), pipeline(), phase(), return values, atom_config data passing, and the journal-based resume / invalidation / partial re-run mechanism (workflow_lineage, workflow_invalidate). Trigger when writing a workflow script, choosing an orchestration pattern, recovering from a wrong node result, or when the user says 编排 / workflow / 多 agent / 并行 / 扇出 / 调度 / 写个 workflow / 失效 / 部分重跑 / 重算. Also trigger when you catch yourself building prompts inside a workflow script — pass data via atom_config instead and let the agent build its own prompt.

Analyze agentm session traces using `agentm trace` atomic commands. Use when inspecting trajectories, token economics, tool calls, or aggregating stats across a trace tree (parent + child sessions). Compose atomic commands via shell pipes — never parse OTLP JSONL directly.

Develop, test, and validate Terminal User Interface applications using tmux as your "headless browser" for terminals. Use when asked to run a TUI application, capture its output, verify its behavior, or debug rendering issues. This skill provides a structured workflow for TUI development, including session management, interaction patterns, state capture, and regression detection.

Develop, test, and validate Terminal User Interface applications using tmux as your "headless browser" for terminals.

Reference for writing and evaluating acceptance criteria. Use when an AC is flagged as untestable, when the spec-review agent needs to judge testability, or when deriving ACs from a vague requirement. Trigger on: acceptance criteria, AC, testability, 验收标准, how to write AC, is this AC testable, untestable criterion.

How to review code against a spec with structured per-AC verdicts. Use when reviewing a PR, verifying code fulfills acceptance criteria, or selecting between multiple parallel implementations. Trigger on: code review, review standards, PR review, 代码审查, 代码评审, review agent, per-AC verdict, structured review.

Overview of the engineering pipeline and how each stage connects. Use when an agent is confused about what stage it's in, what it should do next, or how its output feeds into the next stage. Trigger on: engineering workflow, pipeline, what stage am I in, what comes next, workflow overview, 工程流程, 开发流程, pipeline stage, stage transition.

How to write a design spec that downstream agents can implement and test against. Covers four composable methods: Use Case for interaction flows, Given-When-Then for behavioral scenarios, Design by Contract for function-level guarantees, and plain AC for simple tasks. Use when writing a spec, reviewing spec completeness, or when a dev/test-gen agent reports the spec is too vague. Trigger on: spec, 设计文档, 需求文档, 接口设计, write spec, design doc, acceptance criteria, use case, given when then, design by contract, 契约, 用例.

How to derive tests from a spec. Covers extracting test cases from Use Case extensions, GWT scenarios, and Design by Contract invariants. Core insight: the test oracle is an invariant — tests find their value in checking exception branches and properties that must always hold, not in re-stating the happy path. Trigger on: test generation, write tests, test design, integration test, 写测试, 测试设计, test from spec, test oracle, invariant, property test, 不变量.

Monitor the workbuddy agent pipeline, detect stuck issues, zombie processes, and missed redispatches. Works for both the 3-unit bundle (supervisor + coordinator + worker) and single-process `serve`. Use when the user says 'monitor the pipeline', 'check workbuddy status', '监工', '看看跑完了没', or asks to watch issue progress.

Explain and operate workbuddy: what it is, which command to reach for, and where to find deeper references. **Trigger whenever the user mentions 'workbuddy' in any context** — including operational requests (register/onboard a repo, add/bind a worker, 注册仓库, 加仓库, 接入新仓库, onboard repo, set up worker), explanatory questions (how to use workbuddy, workbuddy guide, teach me workbuddy, 怎么用workbuddy, 使用指南), deployment/running/operating/debugging workbuddy, and questions about how issues/tasks are scheduled or how dependencies between issues work (depends_on, blocked, dependency gate, 依赖, 调度, 排队, '为什么没派发'), and questions about parallel-development workflows (rollouts:N fan-out, synthesize/synthesis reduce step, cherry-pick, '并行开发', '多个方案', 'rollout'). Invoke this skill before running workbuddy CLI commands so the documented onboarding flow (coordinator register + worker repos add) and the dependency/scheduling model aren't re-derived from scratch.

Workbuddy deployment topology guide — explains the 3-unit bundle (supervisor + coordinator + worker), the rolling-restart property, and the cutover from the legacy single-process `serve` install. Use when the user says 'how should I deploy workbuddy', 'install workbuddy as a service', 'switch to bundle', 'rolling restart', '部署 workbuddy', '升级到 bundle'.

Use when extracting reusable workflow patterns, debugging lessons, or project-specific operating knowledge from completed work in this repository.

Use when capturing or checking a repository workflow checkpoint before a risky change, after a milestone, or during a design-to-implementation handoff.

Use when defining or revising repository architecture, design documents, concept boundaries, or cross-document relationships before implementation.

Use when defining or checking acceptance criteria, release readiness, or completion evidence for a repository change or feature.

Use when executing a confirmed repository plan or task and writing code that must stay within documented scope and project conventions.

Use when checking or updating the repository's `.claude/index.yaml` so design, plan, and task relationships stay internally consistent.

Use when a confirmed design or multi-step requirement needs to be broken into repository-local plans and tasks before implementation.

Use when reviewing repository changes for behavioral correctness, design consistency, quality risks, or verification gaps before considering the work complete.

Guide for using the `arl` CLI to inspect, debug, and manage the ARL (Agentic RL) runtime: experiments, warm pools, sessions, files, snapshots, replay, private containers, logs, and metrics. Use this skill whenever the user asks about checking pool status, viewing experiment sessions, creating or debugging sessions, transferring files, restoring snapshots, replaying trajectories, streaming logs, exporting trajectories, waiting for pool readiness, running commands in private sidecar containers, or any operational task involving the ARL system. Also trigger when the user mentions `arl` CLI commands, asks "how do I see what's running", "show me the pools", "check experiment X", "debug this session", "看看 pool 状态", "查看实验", "导出 trajectory", or wants to combine multiple arl operations into a workflow.

Conceptual map for maintaining or using the `arl-env` Python SDK (`from arl import ...`). Use this skill when the user asks about ARL Python SDK capabilities, how SDK features map to gateway/runtime code, where to implement a SDK change, or how Python users should interact with ARL sessions, managed experiments, warm pools, files, replay, trajectories, logs, auth, or interactive shells. Prefer this as a feature/file guide, not a detailed API tutorial.

Reference for using the iroh P2P QUIC library (iroh 1.0+) by n0-computer in Rust and Python. Use this skill whenever writing or modifying code that touches iroh — creating Endpoints, connecting peers, opening QUIC streams, implementing protocol handlers, TCP tunneling, or using the Python iroh bindings. Also use when debugging iroh connection issues, NAT traversal problems, or relay connectivity. Triggers on: iroh, QUIC, EndpointId, iroh transport, iroh tunnel, iroh relay, iroh connect, direct-connect, P2P QUIC, iroh Python, dumbpipe, iroh protocol handler, ALPN.

Kubernetes operations, testing, and validation. Use when working with Kubernetes clusters for deploying resources, verifying deployments, testing operators/CRDs, debugging pods, monitoring workloads, or performing end-to-end testing and validation of K8s applications.

Professional code review for anti-patterns, extensibility, architectural soundness, and code quality. Use when users request code review, ask to check for anti-patterns, need architectural validation, want to ensure extensibility, or request analysis of code quality. Apply to tasks involving code auditing, refactoring recommendations, design pattern validation, or ensuring every line of code has purpose and meaning.

Modern documentation management with MkDocs Material theme and GitHub Pages publishing. Use when users need to create, maintain, or publish professional documentation websites. Supports initialization of new documentation projects, live preview with auto-reload, production builds, and automated deployment to GitHub Pages. Features modern Material Design theme with dark mode, full-text search, mobile responsiveness, code highlighting, and rich markdown extensions including Mermaid diagrams, admonitions, and tabs.

Advanced Python development with architecture expertise and automated code quality enforcement. Use when developing Python applications, libraries, or scripts that require high code quality, type safety, best practices, and professional architecture patterns. Includes automated linting with Ruff, type checking with MyPy, security scanning with Bandit, and comprehensive testing. Apply for tasks involving Python code creation, refactoring, quality checks, project initialization, or implementing design patterns.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Use this skill when reviewing an existing PPT or SlideCraft deck, diagnosing layout problems, inspecting preview PNGs, or iterating on slides based on current deck state rather than blindly editing code.

Use this skill whenever creating, modifying, or generating PowerPoint presentations (.pptx) with paperops.slides. This includes native PowerPoint slide composition, themes, layouts, animations, and deck implementation once the talk structure and visual language are ready.

Use this skill whenever the user wants to turn one or more papers, notes, or source documents into an academic talk plan before building slides. Trigger for conference talks, seminars, research-line talks, job talks, speaker notes, slide scripts, talk outlines, or requests to organize source material into a PPT story. This skill plans the narrative, timing, audience framing, and slide-by-slide content; use slidecraft only after the talk plan is stable.

Use this skill whenever the user wants to turn one or more papers, notes, or source documents into an academic talk plan before building slides. Trigger for conference talks, seminars, research-line talks, job talks, speaker notes, slide scripts, talk outlines, or requests to organize source material into a PPT story. This skill plans the narrative, timing, audience framing, and slide-by-slide content; use slidecraft only after the talk plan is stable.

Use this skill for PPTX-native workflows: reading or extracting content from existing .pptx files, editing templates or unpacked Office XML, combining/splitting decks, or doing final artifact QA on a generated presentation. Do not use it as the default implementation path when a deck is being authored in repo-native slide code such as `paperops.slides`; in those cases route visual-system work to `visual-language`, slide implementation to `slidecraft`, and rendered-deck diagnosis to `slide-review`, then use this skill only for PPTX-specific IO or final QA.

Use this skill when a PPT needs a coherent design language, symbol vocabulary, or diagram style. It defines the presentation's visual system before SlideCraft code is written or revised.

Use this skill whenever creating matplotlib/seaborn figures for academic papers, research reports, or publication-quality plotting tasks using paperops.plotting. Trigger when the user mentions plots, figures, charts, graphs, or visualizations in a research/paper context.

Run repo-aware verification for paperops changes, including code checks, skill-doc consistency checks, and slide/plot smoke tests when relevant.

Use when processing conversations to extract reusable knowledge (entities, facts, processes)

Use after completing any task to extract reusable patterns

Use when you've identified a recurring pattern worth abstracting into a skill

Use before starting any task to find relevant skills

Use when an existing skill needs updating based on new evidence

Advanced Python development guidance with focus on architecture, design patterns, code quality, type safety, and modern best practices. Use when designing Python systems, refactoring codebases, implementing complex features, reviewing architecture decisions, applying design patterns, or ensuring code follows best practices for maintainability, testability, and performance.

Use when running or extending the fmhub formal-verification benchmarks — driving LLMs (codex / DeepSeek V4 Flash & Pro / GLM) to add ACSL/Dafny/Verus specs that a verifier proves, and lifting weaker/cheaper models toward codex quality via loop engineering. Covers the harness toolkit, the proven workflow levers, the Flash+Pro cascade, model/endpoint gotchas, and how to add a solver. Trigger on tasks about FMBench, spec generation, multi-turn verifier loops, hybrid model cascades, or improving a weak model's pass rate on these benchmarks.

Use when integrating or benchmarking an LLM API (especially reasoning models like GLM-5.2, DeepSeek V4, or any new provider/endpoint), or when a model "looks weak" and you need to rule out integration artifacts before concluding. Covers anti-hang call layers, reasoning-model handling (max_tokens, empty content), endpoint comparison, prefix caching, cheat-proof evaluation (faithfulness), and variance-aware measurement. Generalizable beyond fmhub.

Use when shifting from manual prompting of coding agents to designing automated loops. Apply the 4-condition test, 5 building blocks, and MVP construction order for repetitive machine-verifiable tasks like CI triage, dependency bumps, or lint fixes. Trigger on automation, loop design, or agent orchestration requests.

E2E testing with Playwright. Use when writing, debugging, or maintaining end-to-end tests. Covers locators (getByRole, getByLabel, getByTestId), assertions, Page Object Model, authentication setup, network mocking, debugging (trace viewer, UI mode), and test configuration. Triggers on E2E test files (.spec.ts), Playwright config, test debugging, flaky test fixes.

UI/UX design intelligence. 50 styles, 21 palettes, 50 font pairings, 20 charts, 9 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, mobile app, .html, .tsx, .vue, .svelte. Elements: button, modal, navbar, sidebar, card, table, form, chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, flat design. Topics: color palette, accessibility, animation, layout, typography, font pairing, spacing, hover, shadow, gradient. Integrations: shadcn/ui MCP for component search and examples.

Use when adding, deduplicating, or maintaining FaultCards in this repo. Triggers on requests like "add a fault card", "import this postmortem", "review this card", "merge duplicate cards", or when the user pastes a bug/incident description.

UI/UX design intelligence. 50 styles, 21 palettes, 50 font pairings, 20 charts, 9 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, mobile app, .html, .tsx, .vue, .svelte. Elements: button, modal, navbar, sidebar, card, table, form, chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, flat design. Topics: color palette, accessibility, animation, layout, typography, font pairing, spacing, hover, shadow, gradient. Integrations: shadcn/ui MCP for component search and examples.

Draw architecture diagrams on Feishu whiteboards via API. Use when user asks to create system architecture diagrams, flowcharts, module relationship diagrams, or visualize code structure on Feishu/Lark whiteboards. Supports shapes, connectors, mind maps, and iterative visual refinement by downloading and viewing results.