Ex-imanity
- 8 skills
- 0 followers
- 14 hours ago last updated
- ▌ Case Lite · ex-imanity bundle小需求测试用例生成 skill。用户提供飞书 Docx、Wiki 或云空间原生 Markdown 文件链接 → 浏览并手动选择章节 → 生成场景结构 → 生成完整用例 → 可选 agent 自检补漏 → 写回搬山。不含模块拆分和自动选章,适用于单一功能点的小需求。当用户说"小需求用例"、"简单需求生成用例"、"case-lite",或明确只有 1-2 篇文档且无需模块拆分时触发。
- ▌ Case Reorganize · ex-imanity bundle将搬山中已有的测试用例整理为链路 case。支持合并冗余用例、去除边界 case、将串联操作合并为一个场景。输出 full.md 后写回搬山(替换原 case 或追加到目标 case)。当用户说"整理搬山用例"、"用例重构"、"链路case"、"合并冗余用例"时触发。
- ▌ Gapm MCP Recovery · ex-imanity bundleUse when GAPM MCP tools are missing from Codex, gapm_agent_tools is unavailable, OAuth reports invalid_client or authentication_required, App Server shows serverInfo null or empty tools, or a GAPM log investigation appears to require restarting Codex or opening a new conversation.
- ▌ Cross Agent Review · ex-imanity bundleUse when one local AI agent (Codex, ClaudeCode, or Grok) should have a DIFFERENT local agent perform a review of a plan or code without changing the primary artifact, and return an evidence-cited verdict, while the primary keeps continuity and the fix responsibility. Triggers on cross-agent review, "让另一个 agent review", Codex/ClaudeCode/Grok 互审, plan-and-code review handoff between local CLI agents. Do NOT trigger for single-agent self-review or normal code review.
- ▌ Dify Dsl Generator · ex-imanity bundleGenerate, refactor, or review Dify workflow DSL/YML for chatflow, workflow, agent, or webhook-triggered apps. Use when translating business requirements, backend APIs, rule-based workflows, or existing Dify YAML into importable DSL; when choosing node types and graph structure; or when migrating complex repo workflows such as TCG_A/TCG_B/TCG_C into cleaner Dify designs.
- ▌ Context Resilient Task · ex-imanity bundleContext-resilient task management via a filesystem Minimum Recovery Set (MRS in .task-state/). Reconstructs task state from on-disk artifacts so work survives /clear, session interruption, agent switches, and context-window loss. Use this skill whenever the user mentions multi-phase tasks, multi-session work, cross-session recovery, task state restoration, MRS, .task-state, lost context, hallucinated todos, forgotten work, 任务状态恢复, 跨会话任务, 多会话开发, 上下文丢失, /clear 后继续, 任务恢复, or asks the agent to remember a task across sessions. Also trigger proactively when starting any task likely to span more than one session, even if the user doesn't explicitly request recovery — the upfront MRS structure prevents context-loss surprises later.
- ▌ Internal API Cookie Auth · ex-imanity bundleHandle Cookie (CAS) authentication for internal HTTPS API calls without asking users to manually paste session Cookies. Use this skill whenever a script, service, or API debugging task lacks a Cookie, reports a session expiry, redirects to cas.baijia.com or test-cas.baijia.com, returns HTTP 401, returns a CAS-style JSON code 700, or may have failed because of authentication. Also use it when the user pastes a browser "copy as cURL" command for an internal host (gaotu100.com, baijia.com): it teaches how to wrap the request in any form (command, bash, or Python) with a fresh Cookie and minimal headers instead of the pasted Cookie, and to confirm before replaying side-effecting POST/PUT/PATCH/DELETE calls. Use it before hardcoding a Cookie. Only probe an unknown host after the user authorizes a real read-only test; treat HTTP 403 as a possible authorization failure, not evidence that a Cookie refresh will help.
- ▌ Case Design Strategy Skill · ex-imanity bundleTest design strategy and coverage-review guidance for requirement-based cases. Use when the user asks for 用例设计策略, 覆盖度补充, 用例评审, 边界/异常/权限/状态/埋点覆盖, telemetry/cross-end risk analysis, or when no end-to-end case generator is responsible. Also use during `case-lite` self-review when the active workflow explicitly enters coverage review / boundary-exception review. Do not trigger for `case-lite` document ingestion, chapter selection, full generation, or writeback steps.