Overview

Repairledger is a Python CLI and agent skill for durable repair observations. A repair observation records an agent-discovered tool failure, validation false positive, unclear command contract, missing documentation, skill gap, workflow pain point, or concrete improvement opportunity.

The primary user is a coding agent working on another task. The agent can log an observation quickly from any source repository, continue the primary task, then export a maintainer-facing Markdown report from the user-global data path.

Key concepts

  • Repair observation: A structured record describing a failure, gap, or improvement.

  • Component: A modular Markdown file within a repair (e.g., summary, reproduction, suggested fix).

  • Version snapshot: A complete copy of a repair after every mutation.

  • Rendered artifact: A standalone Markdown document with YAML front matter.

  • Aggregate report: A global-data-path Markdown report covering all active observations.

  • User-global ledger: A single config at ~/.config/ledger/repairledger.toml and a single data path at ~/.local/share/ledger/repairledger/. Agents log observations from any repository without writing into the current source tree.