Capture conversation insights and propose validated SKILL.md updates.
npx skills add https://github.com/dvy1987/agent-loom --skill learn-from-chat
Please help me install this Agent Skill. Skill: learn-from-chat Source: https://github.com/dvy1987/agent-loom/tree/main/.agents/skills/learn-from-chat Command: npx skills add https://github.com/dvy1987/agent-loom --skill learn-from-chat
This Skill ensures insights that emerge during agent-user conversations are captured, validated, and turned into safe, targeted updates to skills or processes so documentation and behaviors stay accurate and current.
Capture a discovered issue by asking the agent to summarize the insight, provide evidence, classify it, and propose exact SKILL.md changes for approval.
High-intent search queries and answers about installing and using this skill.
To capture learnings from chat conversations, you extract a one-line insight with concrete evidence, classify it, and propose targeted updates to SKILL.md files. This ensures agent documentation stays accurate and current based on interactive debugging sessions.
The best way to automate skill maintenance is to surface repeated failure modes from debugging sessions, propose guardrail additions to affected skills, and log changes after explicit user approval. This process requires scoped diffs and validation checks before applying updates.
Chat learning classification for process orchestration works by categorizing extracted insights into types like GOTCHA, TECHNIQUE, FAILURE_MODE, METRIC, or CONTRADICTION. The system then locates affected SKILL.md files to propose exact diff-style changes for documentation updates.
No, proposing guardrail changes to SKILL.md files requires explicit user approval and validate-skills execution. The system presents exact diff-style change proposals, bumps metadata versions, and appends a dated citation to docs/research-learnings only after validation passes.
You need to log research learnings from agent-user chats when interactive debugging, postmortems, or orchestrator observations reveal evidence that a skill, workflow, or guardrail should change. A dated entry is appended to a shared research log after user approval.
Limitations of using conversation insights for skill improvement include the requirement for explicit user approval before any changes apply. The system only generates scoped diffs to SKILL.md files and cannot automatically merge updates without running validate-skills execution and logging citations.
pipemind-comExtract actionable lessons from task conversations and propose CLAUDE.md updates.
EmasoftDetect and categorize Git changes into structured CHANGELOG.md entries.
lailtonjuniorIdentify root causes of software defects through structured evidence collection and reproducible steps.
zbsayginSearch, classify, and map citations across economics literature.
