Turn past agent transcripts into durable coding rules and skills
zeeg · x · 2026-07-26
A practical workflow for improving coding agents: feed them a chunk of past transcripts, ask them to find recurring failure patterns, then have them suggest durable fixes such as new prompt guidance, skills, or AST-grep rules. The author says this worked on Warden’s code-quality loop, though the output may still include hallucinations and requires manual judgment.
The core idea is to turn your own interaction logs into a feedback loop:
- Review historical transcripts around a task you care about.
- Ask the agent to summarize patterns where outputs went wrong.
- Convert those patterns into reusable guidance stored outside the main context.
The takeaway is that LLMs are strong at pattern matching and repetitive work, so transcript mining can make them much more effective over time.
More from coding & agent
- Claude + Kit MCP: Automating Email Marketing Workflows via Natural Language — evielync · 2026-07-26
- Builder replaces hundreds of installed agent skills with one registry-backed skill — abhimanyu_saharan · 2026-07-26
- Anthropic says its internal Slack agent now writes 65% of product team code — VraserX · 2026-07-26
- Aethos Memory lets Claude Code, Cursor, and Windsurf share one persistent brain — Longjumping-Koala396 · 2026-07-26
- Open-sourced MCP memory server for coding agents uses 3-pass RAG retrieval — Longjumping-Koala396 · 2026-07-26
- LLM IMO 2026 test shows harnesses help, but frontier models still stay ahead — pequalnp92 · 2026-07-26