self-improvement

To automatically log errors and learnings into structured markdown, this skill uses lightweight hook scripts that detect command failures and user corrections during agent sessions, writing them to standardized markdown templates in the .learnings directory.

The best way to capture user corrections and command failures is using hook-based detection scripts that automatically evaluate interactive agent sessions and log operational learnings into structured markdown files like LEARNINGS.md and ERRORS.md.

To promote recurring development learnings into project memory, this skill provides promotion workflows and scripts that elevate high-value entries from markdown logs into CLAUDE.md, AGENTS.md, or extracted skills for broader reuse.

Yes, you can use hook scripts to detect tool errors in CI-assisted runs and developer workspaces, as the skill is designed to capture post-task discoveries, recurring gotchas, and external tool errors across multiple integration scenarios.

Available markdown templates for logging feature requests and operational errors include standardized formats for LEARNINGS, ERRORS, and FEATURE_REQUESTS, keeping entries uniform and searchable across developer and agent sessions.

No, this skill requires no external dependencies to capture operational learnings in agent sessions, relying solely on lightweight internal hook scripts and markdown templates for error detection and structured logging.