From idmp-plugin
IDMP AI skill. Use it to confirm AI availability and usable data, inspect chat sessions, generate recommended questions or panels, create analyses or panels from prompts, and decide when `record` should persist the session.
How this skill is triggered — by the user, by Claude, or both
Slash command
/idmp-plugin:idmp-aiThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Read [`../idmp-shared/SKILL.md`](../idmp-shared/SKILL.md) first.**
Read ../idmp-shared/SKILL.md first.
record:true is acceptable.datasource available list=false as a hard stop.idmp-cli datasource available list is the hard availability gate.recommend can return an empty array.idmp-cli ai anydata list measures AI-visible data, not chat history.idmp-cli ai chat sessions is the safest proof that recorded chat is visible.ai recommend create and ai recommend create-post still need questionType values from the backend QuestionType enum. Use GENERAL_QUESTION, GEN_PANEL_QUESTION, or GEN_ANALYSIS_QUESTION; do not send intuitive aliases such as PANEL or ANALYSIS.ai recommend create calls /api/v1/ai/panels/recommend, while ai recommend create-post calls /api/v1/ai/prompts/recommend.[]; classify that as empty output instead of inventing advice.Can't parse the value: PANEL to AiGenPromptType, classify it as a backend contract mismatch instead of datasource or auth unavailability.sessionId:null.datasource available list and ai anydata list are healthy; classify context deadline exceeded as backend AI/API latency and hand off to structured analysis workflows instead of mutating the request semantics blindly.idmp-cli datasource available list and stop immediately if availability is false.idmp-cli ai anydata list to prove the AI backend can see usable data instead of assuming data visibility from availability alone.idmp-cli ai chat sessions before any recorded chat or follow-up AI write so the session state is grounded in a real reread.idmp-cli ai recommend create-post --ack-risk --data with the final questionType locked to GENERAL_QUESTION, GEN_PANEL_QUESTION, or GEN_ANALYSIS_QUESTION, then classify the result as populated, empty, structured failure, or backend contract mismatch.idmp-cli ai create create --ack-risk --data or idmp-cli ai create create-post --ack-risk --data only when a draft object is requested, then prove the response actually contains a generated object before claiming success. If analysis draft creation times out, stop and hand off to the structured analysis workflow instead of claiming AI unavailability.datasource available list and ai anydata list agree on the backend state.idmp-cli datasource available list to prove the backend is reachable.idmp-cli ai anydata list to prove the AI service can see usable data.idmp-cli ai chat sessions to inspect visible chat history before new writes.idmp-cli ai recommend create-post --ack-risk --data '{"elementId":123,"questionType":"GENERAL_QUESTION","async":true}' for generic prompt suggestions.idmp-cli ai recommend create-post --ack-risk --data '{"elementId":123,"questionType":"GEN_PANEL_QUESTION","async":true}' or ...GEN_ANALYSIS_QUESTION... for panel or analysis-oriented recommendations.idmp-cli ai create create --ack-risk --data '{"elementId":123,"prompt":"Create a 1-minute average-current analysis","record":true,"deepThinking":false,"deviceDocument":false}' for analysis drafts.idmp-cli ai create create-post --ack-risk --data '{"elementId":123,"prompt":"Create a maximum-current trend panel","record":true,"deepThinking":false,"deviceDocument":false}' for panel drafts.Run idmp-cli datasource available list and idmp-cli ai anydata list first. Success here only means the environment can support later AI work.
Use idmp-cli ai anydata list after availability passes. If the list is empty, treat that as weak AI context rather than a fabricated success.
Read idmp-cli ai chat sessions before recorded chat creation. Only use session detail reads if the list result gives you a trusted session ID.
Use idmp-cli ai recommend create-post --ack-risk --data with a locked element scope and an explicit backend enum such as GENERAL_QUESTION, GEN_PANEL_QUESTION, or GEN_ANALYSIS_QUESTION. Empty arrays, structured backend errors, and enum-contract errors are all valid, different outcomes.
If idmp-cli ai create create --ack-risk --data or idmp-cli ai create create-post --ack-risk --data fails, classify the failure and stop. Distinguish backend timeouts from true availability failures, and never report a generated object unless the response proves it.
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub taosdata/agent-skills --plugin idmp-plugin