What problem does it solve?
Build and compilation errors often trigger a slow ping-pong where the user pastes one error, the agent guesses a fix, and the build stays broken for many rounds. This Skill takes ownership of the whole process: it runs the build itself, collects all errors at once, finds root causes via dependency analysis, and loops until the build is green.
Core Features & Use Cases
- Full error collection: Runs the build command (cargo, bun, expo) and parses every error into a structured list instead of fixing one at a time.
- Root-cause analysis: Groups errors by file and crate, distinguishes root causes from cascading failures, and fixes root causes first.
- Safe iterative loop: Creates git revert points before each fix, reverts failed hypotheses, tracks state across rounds, and escalates to the user when stuck.
- Use Case: After an Expo or dependency upgrade breaks the build with 12 cascading TypeScript and Rust errors, the agent runs the build, identifies 2 root causes, fixes them in order, and reaches 0 errors in a few rounds without user involvement.
Quick Start
Ask the agent to run error-diagnose-fix on the failing build, for example by saying the build is broken and to fix all compilation errors.
