What problem does it solve? Draining a backlog of triaged production errors requires an operator to manually claim each root cause, read the triage analysis, write a fix, and open a PR. This Skill automates one full iteration of that loop against the Chronicle fix queue, running the fix work in your local agent session instead of a metered operational LLM key. ## Core Features & Use Cases - Lease-based queue claiming: Claims the highest-severity open root_cause from the Chronicle fix queue with a 2-hour lease, heartbeat support, and clean handling of empty queues (204) and lease conflicts (409). - Guided fix workflow: Reads the triage agent's root-cause analysis from the linked GitHub issue, branches off origin/main as agent/fix-<n>, verifies with cargo check/test/fmt, and opens a PR with Closes #<n> linkage. - Triage mismatch handling: Releases the claim with a reason when the triage agent matched the wrong root cause, and documents discrepancies in the PR body when suspected files were wrong. - Use Case: An operator cd's into their local methodic checkout and runs the skill in a loop (e.g. claude /loop fix-error-queue) to drain the error queue, with a human reviewing each PR before merge — no autonomous merging at any stage. ## Quick Start From inside your local methodic checkout with CHRONICLE_SERVER_URL and CHRONICLE_API_KEY exported, ask the agent to fix the next error in the queue.
