feature-intake
从文本与截图中提取并规范化功能需求为结构化项。
相关性达标的全量 Skills(含未精选长尾)· 共 6259 个
从文本与截图中提取并规范化功能需求为结构化项。
自动化代码审查并基于得分与阈值执行质量门控。
构建依赖感知的并发 agent 调度队列与分批执行策略。
通过 Babysitter SDK 执行并编排结构化流程与循环任务。
扫描云、仓库和本地并构建可查询的系统 atlas。
为 agent 提供查询 Atlas 知识图谱的标准化接口。
协助把新的 CLI 型代码代理适配器接入项目,包含审计、骨架、测试与文档。
对 .a5c/runs/<runId>/ 的运行进行迭代编排与执行管理,协调子代理工作流。
分析外部仓库中提交的 babysitter 运行记录并生成回顾与改进建议。
按 SDK 模式生成新的 babysitter 流程定义并提供三阶段实现指导。
检测 babysitter 仓库 staging 分支上失败的 Actions 流程并尝试修复或触发修复流程。
发现并维护使用 babysitter SDK 的公共仓库目录,生成去重的用户清单。
拉取分配给 a5c-agent 的 GitHub issue 并自动处理或分派工作。
在公共 GitHub 仓库中搜索并收集 SKILL.md,归类提取常见工作流模式。
为任意金融市场提供策略性情报与投资/宏观分析框架与执行建议。
用 OpenArt 生成预渲染画面,构建随滚动驱动的连续飞行式沉浸落地页(无切换)。
指导在 PinMe 部署的 Cloudflare Worker 中使用自动注入的 env.R2 进行对象存储操作。
PinMe 零配置部署:上传 IPFS 或部署全栈项目(Worker+D1 等)。
为 PinMe Cloudflare Worker 提供 UniwebPay 支付集成代码与示例。
生成静态分享页并通过 pinme 上传,返回最终分享 URL 的流程指南。
在 PinMe Worker 中接入 OpenRouter 代理的 LLM 与搜索 API 的使用指南。
示范在 PinMe Worker 中调用平台邮件发送接口的实现指南。
指导在 PinMe Worker (TypeScript) 中接入身份验证代理 API 的用法。
串联 Maieutic 与 Elenchus 完成意图确认+变异挑战的完整审查流程。
将代码变更转为需决策的意图合约并生成/验证单元测试契约。
用风险等级变异挑战测试套件,评估测试覆盖与回归风险。
将学术或办公 Markdown 转换为结构化可编辑的 DOCX,保留层级、公式与图表样式。
The primary entry point for the taskforge engineering workflow - create and import tasks, query and manage the backlog, route work, and unblock humans. Use whenever the user wants to add or import work ("add a task", "import issue
Implement a taskforge task against its active Specification and verify it with an independent fresh-context review subagent. Use when a task's readiness is "run", or when the user asks to run, implement, execute, build, or ship a taskforge task that has a specification. Covers the implement-review-retry loop, recorded-and-auditable reviewer isolation, root-cause escalation back to taskforge-refine or taskforge-explore, and scope discipline (out-of-scope discoveries become follow-up tasks). Tasks without an active specification route to taskforge-refine instead.
Independently review a taskforge task's implementation against its Specification and emit only the verdict — the review half of the split run workflow (docs/ADR-0001 Phase 6). Use when a task's readiness is "review" (an implementation awaits review). Runs in fresh context against the recorded reviewer prompt; produces one review artifact (approved, or rejected with a root_cause). The KERNEL routes on the verdict (approved -> done, rejected -> refine/explore/retry) — the review never emits a done or escalate signal.
The universal entry point of the taskforge workflow - assess whether a task is executable and produce its Specification, or route it. Use whenever the user says to refine, spec, scope, groom, or prepare a taskforge task, asks "what's next for TASK-x", or wants to move a new task forward - and whenever a task's readiness is "refine" (including after escalations or invalidation cascades). Four modes - adopt a well-written task nearly verbatim, elaborate an underspecified one, generate blocking clarification prerequisites, or escalate to taskforge-explore for an architectural decision.
Implement a taskforge task against its active Specification and emit only the implementation — the build half of the split run workflow (docs/ADR-0001 Phase 6). Use when a task's readiness is "run" (ready to build). Produces one implementation artifact; it does NOT review its own work — an independent taskforge-review follows. Escalates upstream defects (spec unimplementable, approach wrong, spec premise provably false) instead of pushing through. Tasks without an active specification route to taskforge-refine.
Produce an engineering Decision for a taskforge task - chosen approach, real alternatives with rejection reasons, trade-offs, risks - and propose decomposition or related work for human approval. Use when a task's readiness is "explore" (an escalation is pending from refine, run, or a child task, or a research topic was started via the explore command), or when the user explicitly asks to explore, architect, decide an approach for, or break down a taskforge task. Not for writing specifications (taskforge-refine) or implementing (taskforge-run).
Build and review high-quality technical docs as well as agent instruction files in your repository.
Clipboard-ready handoff prompt for another agent to investigate or continue a task.
GitHub archive: issue/PR search, sync freshness, duplicate clusters, gh-shim PR status, and Gitcrawl repo work.
Pre-commit/ship code review: Codex default; optional Claude or Pi.
Fetches the current top stories from Hacker News. Returns title, URL, score, and author for each. Use when the user asks what's on Hacker News, what's trending in tech, or for news headlines.
将俄语 AI 文本改写为更自然、更有人味的语言。
将俄语 AI 文本改写为更自然、更有人味的语言。