4,908 件 ・ 関連度順
概要と使いどころ
Use when working in a repository that uses bd or Beads for durable project task tracking, issue dependencies, blocker management, multi-session handoff, or shared work memory. Trigger when the user asks to find ready work, claim or close tasks, create follow-up work, inspect blockers, recover project context, or choose between local planning and persistent project tracking.
日本語の概要は準備中です。原文の説明を表示しています。
Review every Beads issue waiting on a person, one at a time, using the interactive questionnaire UI rather than plain-text prompts. Use when asked to go through blockers, clear the human queue, answer beads waiting on the user, triage `bd human list`, or unblock agent work that needs a decision.
日本語の概要は準備中です。原文の説明を表示しています。
融合单元的范围判定与收益前置评估：判定「哪些计算该被融进同一个融合单元、边界画在哪、 融了值不值」。结构侧按融合单元构造规则切边界——函数边界即候选组；FA（含 BSA 等变种）为 禁融锚点；以 norm / rope / FA 为分界分区，全 Vec 区域优先试融；区域内有 Cube 时以 Cube 为 首算子向后包裹 Vec（Σvec 小于 cube 或遇下一 Cube 封口）；一轮之后纯 Vector 单元之间再迭代； MLP / MoE 以矩阵乘族（MatMul / BatchMatMul / GroupedMatmul）为界；若该段计算已由单个 融合算子承载，按规则 6 先确认载体是否已生效（不重复建单元）。 收益侧按计算单元利用率判型（memory_bound、*_vec|mac|mte2|mte3_ratio、cube_utilization）， 叠加带宽下限 / 区域占比 / 天花板法 / 地板先行 / Amdahl 传导校验给出 go/no-go。 当用户问「这几个算子能不能融成一个」「融合边界画在哪」「融多大范围」「融了值不值得」， 或 profiling-analyze 给出融合机会候选需要定范围与收益时使用；即使用户只说「这条链要不要 合成一个 kernel」「这里能不能省一次读写」也应触发。 **已经做完、实测收益却打平或为负时同样触发**——此时的任务不是接受"融合无收益"， 而是按事后复核做**等功忙计数对照**归因，并在缺陷落在融合核自身时**升级 operator-dev**。 边界：只出「融合单元草案 + 收益判定 + 否决理由」。不做采纳与选点（dit-perf-opt 依收益选 top 并派活）、不做采集与瓶颈定位（profiling-collect / profiling-analyze）、不写 kernel、不写 pattern、 不做使能与并行选型、不定义验收标准（引 accuracy-gate / perf-gate）。 由 profiling-analyze（融合机会候选）与 operator-dev（实现前的收益评估）调用。
日本語の概要は準備中です。原文の説明を表示しています。
算子级开发与性能优化：Triton / Ascend C / Catlass / PyPTO / TileLang 算子 编写、精度对齐与性能调优。优先路由到外部 cannbot-skills 技能库（不重复其内容）， 本 skill 只保留场景 → skill 映射与 MindIE-SD 特有补充。 当用户需要新增/优化算子、定位算子性能或精度问题时使用此 skill。 即使用户只提到"写个 triton kernel"或"这个算子怎么加速"而未说算子，也应触发； pattern 融合/编译后端见 pattern-dev，算子基准选型/接入测试见 benchmark-dev。 由 dev-workflow 或 pattern-dev 的 replacement kernel 场景触发。
日本語の概要は準備中です。原文の説明を表示しています。
对 profiling 数据（kernel_details.csv / trace_view.json / step_trace_time.csv）做统一分析： 5 层递进管道（analyze_trace.py）+ baseline/优化后 kernel diff（compare_traces.py），输出 瓶颈诊断、算子执行序、可融合候选与 P0-P2 方向建议。输入统一来自 profiling-collect。 只要用户有 profiling 产出并问"为什么慢/瓶颈在哪/优化前后差多少/往哪个方向"，都用本技能；采集走 profiling-collect，方案选档走 dit-perf-opt，实现/并行/基准走 pattern-dev/dit-parallel-opt/benchmark-dev——本技能只做诊断与方向。 由 model-auto-optimization 的 S1（融合分析）/S3（并行）阶段调用，亦由 dev-workflow 的分析阶段指引加载。 融合机会候选只做**识别**；候选的**边界与收益判定**（能不能融成一个单元、融多大、值不值）交 `fusion-scope-analyze`（本技能的辅助技能）——本技能不判融合边界。
日本語の概要は準備中です。原文の説明を表示しています。
**精度验收标准**：凡是"改动不应改变结果"的场合（等价替换、算子/子模块融合、并行切分、 编译与图下发、拷贝消减），都按本标准判"合格 / 不合格"——等价分层（L1 逐位 / L2 数值门 / L3 有损门）+ 三级验收序（① 同配置重跑逐位 → ② 跨配置数值门 + 产物 md5 不变 → ③ 质量门）+ 判据不达标时的排障入口。当用户说"这个算子能不能换个写法/换个 kernel/等价实现/ 无损替换""替换后结果会不会变""怎么证明逐位一致""结果不对""花屏""尾部塌了" "CPU 跑对 NPU 跑不对"，或发现某个 hotspot 占了大头（例如某类算子在阶段里占 90% 以上）想动手时， 都应触发；即使用户只说"这样改有没有把结果改坏""能不能判它是无损的"也应触发。 覆盖：等价分层与判据、满足本标准的实现约定（per-shape 对拍写进实现，把索引/相位/顺序类重写 错误在首次调用抓住）、per-shape 条件性等价（同一改动在不同形状下结论可能不同）、 判据不达标 → `references/silent-failure-localization.md` 排障入口、否决案例的形态学 （换写法未换 kernel / 差异极小仍非逐位 / 等价但 OOM）、以及收益口径的归属（性能数字一律交 `perf-gate`）。 **近义分流**：选量化档 / 特性选档（要不要开量化、开哪一档）→ `dit-perf-opt`； 并行选型（USP / CP / TP 怎么切）→ `dit-parallel-opt`；本技能只判"结果是否被改变"， 不负责选档与选型。
日本語の概要は準備中です。原文の説明を表示しています。
Use when building APIs on Hono (Cloudflare Workers, Bun, Deno, Node), debugging route ordering, wiring middleware, validating with @hono/zod-validator, returning streaming responses, configuring CORS with credentials, handling errors via app.onError, or composing typed RPC clients. Triggers: c.env binding types, c.header + c.redirect interaction, missing await next() bugs, Set-Cookie not attaching to redirect, route-precedence surprises, JWT middleware setup, hono/client typed RPC end-to-end. NOT for Express/Fastify/Koa idioms, tRPC/GraphQL paradigms, or Next.js Route Handlers.
日本語の概要は準備中です。原文の説明を表示しています。
Expert DevOps engineer for CI/CD, IaC, Kubernetes, and deployment automation. Activate on: CI/CD, GitHub Actions, Terraform, Docker, Kubernetes, Helm, ArgoCD, GitOps, deployment pipeline, infrastructure as code, container orchestration. NOT for: application code (use language skills), database schema (use data-pipeline-engineer), API design (use api-architect).
日本語の概要は準備中です。原文の説明を表示しています。
Cloudflare Pages and Workers deployment checklist for wrangler deploys, secret audits, D1 migrations, Worker-compatible SDK checks, and browser-auth verification. Use when shipping Cloudflare changes to production. NOT for general CI/CD architecture, non-Cloudflare deployments, or post-launch UX QA.
日本語の概要は準備中です。原文の説明を表示しています。
Expert in background job processing with Bull/BullMQ (Redis), Celery, and cloud queues. Implements retries, scheduling, priority queues, and worker management. Use for async task processing, email campaigns, report generation, batch operations. Activate on "background job", "async task", "queue", "worker", "BullMQ", "Celery". NOT for real-time WebSocket communication, synchronous API calls, or simple setTimeout operations.
日本語の概要は準備中です。原文の説明を表示しています。
Conversation patterns and interaction protocols for multi-agent systems. Covers request/response, pub/sub, blackboard, delegation chains, debate, critique, consensus, fan-out/fan-in, supervisor-worker, and peer negotiation. Deep analysis of AutoGen conversation patterns, CrewAI delegation, LangGraph state passing, and FIPA-ACL performatives. Teaches how to design what agents say to each other and in what order. Activate on: "agent conversation", "agent protocol", "multi-agent debate", "agent delegation", "supervisor worker pattern", "agent voting", "consensus protocol", "fan-out fan-in", "agent negotiation", "blackboard pattern", "agent dialogue", "conversation topology", "agent handoff". NOT for: wire format or serialization (use agent-interchange-formats), orchestration infrastructure (use agentic-infrastructure-2026), single agent behavior (use agentic-patterns).
日本語の概要は準備中です。原文の説明を表示しています。
Manage JIRA issue relationships via the jira-as wrapper: link, unlink, blockers, dependencies, clone, bulk-link. Triggers: 'link KEY blocks KEY', 'what blocks KEY', 'clone KEY', 'unlink KEY from KEY'. NOT for sprint/epic relationships (mk:jira-agile).
日本語の概要は準備中です。原文の説明を表示しています。
DraftForge test infrastructure for writing backend unit tests, Playwright E2E tests, and managing test data population. This skill should be used when writing new tests, adding test fixtures, creating populate functions for new features, adding Playwright E2E specs, using login/auth fixtures, creating test users/orgs/leagues/tournaments, or understanding the test data architecture. Covers backend Django tests via Docker, Playwright config (projects, workers, sharding), test auth endpoints that expose backend login to frontend, and feature-isolated test data patterns.
日本語の概要は準備中です。原文の説明を表示しています。
Python Invoke task automation for DTX website. This skill should be used when running repo commands, backend tests via Docker, database migrations, updating invoke tasks, Docker operations, or workflow automation. Supports dev/test/prod environments with run, exec, up, down commands. References docs/development/invoke-tasks.md for complete command reference.
日本語の概要は準備中です。原文の説明を表示しています。
Worker-Reviewer iterative pipeline for quality-critical code with review cycles
日本語の概要は準備中です。原文の説明を表示しています。
Docker patterns for optimized containerization
日本語の概要は準備中です。原文の説明を表示しています。
Processes eDNA metabarcoding from raw paired-end reads to species tables, navigating ASV (DADA2, UNOISE3) vs OTU (swarm v2) decision (Callahan 2017 vs Schloss multi-copy-16S critique), marker/primer choice (Leray COI, MiFish 12S, 515F/806R 16S, ITS2) with primer-specific bias, OBITools3 v3 command-name break (obi stats plural; .tar.gz taxonomy), tag-jumping with dual-indexing (Schnell 2015; NovaSeq 10x MiSeq), decontam as screening-not-classifier (Davis 2018), read-counts-not-abundance critique (Lamb 2019), site-occupancy modeling (Ficetola 2015), Naive-Bayes calibration limits (Bokulich 2018), and eDNA decay (Strickler 2015). Use when going from raw eDNA FASTQ to species tables, picking marker + denoising pipeline, deciding whether read counts represent abundance, applying occupancy modeling, configuring OBITools3 v3, or interpreting decontam output. Not for clinical 16S microbiome (see microbiome/amplicon-processing).
日本語の概要は準備中です。原文の説明を表示しています。
Expert guide for Backend.AI distributed computing platform. Automatically activates when users ask about: - Backend.AI architecture, components (Manager, Agent, Storage Proxy, Webserver, App Proxy) - Features (session scheduling, Sokovan orchestrator, multi-tenancy, resource allocation) - APIs (REST, GraphQL), authentication, RBAC authorization - Container runtime (kernels, jail sandbox, hook library, virtual folders) - Accelerator support (CUDA, ROCm, TPU, NPU, Graphcore IPU) - Client SDKs (Python, Java, JavaScript, PHP) - Setup, requirements (Python 3.13+, Docker, PostgreSQL, Redis, etcd) - How WebUI connects to/interacts with Backend.AI backend - Plugin interfaces, development setup, infrastructure Use when user mentions "Backend.AI", "backend.ai", "Sokovan", component names, or asks about the backend platform this WebUI connects to. allowed-tools: WebFetch, Read
日本語の概要は準備中です。原文の説明を表示しています。
Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.
日本語の概要は準備中です。原文の説明を表示しています。
Execute a swarm plan wave by wave with filesystem isolation via git worktrees. For each wave, main context decides the scenario recipe + implementation contract for every task, then spawns one swarm-worker per task in parallel. Each worker executes its recipe and reports JSON status. Worktree merge-audit verifies write-set discipline before changes land on main. Aborts remaining waves on any audit or task failure.
日本語の概要は準備中です。原文の説明を表示しています。
Workflow Phase 6 — TDD coordinator. Decides the scenario recipe and the implementation contract in main context, writes them to a state file, seeds per-worker tasks (scenario, implement, verify-tick, design-ui-tick) into the TaskList, and yields with harness_state continue so the harness invokes each worker as its own tick. No subagent delegation; no nested Skill calls.
日本語の概要は準備中です。原文の説明を表示しています。
Oracle-bound spec-review check that every structured Rollout prerequisite binds to an enforcement-type acceptance criterion. A prerequisite whose `enforced-by` is missing, dangling, or points at a non-enforcement AC is a BLOCKER (hard-blocks implementation entry via the checker fan-out verdict); a prerequisite left in free prose is ADVISORY. Read-only. Runs in the spec-review fan-out before implementation.
日本語の概要は準備中です。原文の説明を表示しています。
Sets up an `## Agent skills` block in AGENTS.md/CLAUDE.md and `docs/agents/` so the engineering skills know this repo's issue tracker (GitHub or local markdown), triage label vocabulary, and domain doc layout. Run before first use of `to-issues`, `to-prd`, `triage`, `diagnose`, `tdd`, `improve-codebase-architecture`, or `zoom-out` — or if those skills appear to be missing context about the issue tracker, triage labels, or domain docs.
日本語の概要は準備中です。原文の説明を表示しています。
Iterative Python via live Jupyter kernel (hamelnb).
日本語の概要は準備中です。原文の説明を表示しています。