菜单
Use when completed work needs to be recorded as a lore artifact covering what happened, what surprised, what broke, and what drifted. Observation only, not interpretation. Triggers include "let's do a retro", "/retro", "review what happened", "capture what happened", and "write up the session".
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when checking whether field-guide reference pages still match current code and tests. Reads pages plus fg-evidence anchors, reports or fixes semantic drift, and can use sub-agents for parallel audit batches. Triggers include "resolve drift", "semantic drift", "check accuracy", "verify the reference against code", and "audit the wiki".
Use after ingesting or editing field-guide pages to attach living code/test anchors. Adds or refreshes fg-evidence metadata without trying to prove prose accuracy. Triggers include "update evidence", "wire evidence", "add evidence anchors", and "connect the wiki to code".
Use when extracting knowledge from .lore/ artifacts into the reference wiki. Converts work artifacts (retros, specs, plans, research, notes) into structured wiki pages grouped by knowledge type. Reads Markdown or HTML sources and writes Markdown by default. Triggers include "ingest this", "extract knowledge from", "add this to the wiki", "update the reference from", and "populate the field guide".
Use when setting up field-guide on a project for the first time, or to refresh the scheduled lint job after it expires. Triggers include "init field-guide", "set up field-guide", "refresh field-guide lint schedule", and "bootstrap the wiki".
This skill is used when the project wiki needs a health check. Triggers include 'lint the wiki', 'check wiki health', 'run field-guide lint', 'are there stale wiki pages', 'find orphaned pages', 'check for contradictions in the wiki'.
Use when asking questions about the project wiki, querying accumulated knowledge, searching project decisions or lessons. Triggers include "query the wiki", "what does the wiki say about", "search project knowledge", and natural language questions directed at the field-guide wiki.
| name | retro |
| description | Use when completed work needs to be recorded as a lore artifact covering what happened, what surprised, what broke, and what drifted. Observation only, not interpretation. Triggers include "let's do a retro", "/retro", "review what happened", "capture what happened", and "write up the session". |
Record what happened. The retro is the witness, not the analyst.
Future sessions will find this via lore-researcher. Make it searchable: tag the problem types, technologies, and patterns the session touched. That's what surfaces a retro when someone plans similar work later.
The body describes what happened. It does not interpret. Forbidden:
lesson, insight, we learned, takeaway don't appear in retro output.What happened? What did the plan say versus what actually happened? Where did things get tripped up? What surprised? What dead ends were tried? What context is worth preserving for the next session in this area?
Write it as prose. Use structure only when it helps the reader follow.
Read relevant work artifacts first (spec, plan, notes) to reconstruct context before writing. Save to .lore/work/retros/[descriptive-name].md. Load ${CLAUDE_PLUGIN_ROOT}/shared/frontmatter-schema.md for the frontmatter fields.
Title describes the work, not a takeaway. "Auth-flow rollout" is a title. "Why we should always test migrations" is not. Tags and modules are how lore-researcher finds this later — keep them grep-discoverable.
Write the body in Markdown per the "Body Format" section of ${CLAUDE_PLUGIN_ROOT}/shared/frontmatter-schema.md. Reach for embedded inline HTML only when a visual carries meaning prose can't, such as a timeline diagram of how the session unfolded.