docs-tune-ai-chatlisted

# docs-tune-ai-chat — Tune AI chat system prompt from real feedback ## Workflow 1. **Verify connection and plan** — resolve which workspace you are operating on and read its current configuration; confirm it is on PRO or PRO+. On Free plan, stop and print an upgrade prompt. Confirm with the user that they want to modify the system prompt before proceeding. 2. **Collect the failure signal** — gather the chat conversations readers rated badly and the questions the assistant could not answer, over the last 30 days. For the rated-badly ones, keep the user question, the AI answer, and any free-text reason given; for the unanswered ones, the interactions where the assistant explicitly said it didn't know or retrieval returned nothing useful. 3. **Cluster by topic** — group the combined signal into 3–8 topic clusters. For each cluster, record a label, item count, up to three sample questions, and a one-sentence description of the inferred failure mode. 4. **Generate a prompt update** — read the current `system_prompt`. Produce a minimally invasive replacement that keeps all existing brand voice, persona, and refusal rules intact, and adds explicit guidance for the top 3–5 clusters. Cap the result at 1,500 tokens. 5. **Show the diff** — render a before/after diff with annotations mapping each changed chunk back to the cluster that motivates it. 6. **Apply on confirmation** — call `set_chat_system_prompt` only after the user explicitly confirms. Accept `yes`, `no`, or `edit`; on `ed