What problem does it solve? Google Jules runs coding sessions asynchronously, leaving completed work, unanswered questions, and unmerged patches scattered outside your GitHub workflow. This Skill closes the loop by polling the Jules API, scoring each session, promoting approved work to pull requests, and handing PRs off for merge management. ## Core Features & Use Cases - Session Triage & Scoring: Polls the Jules API for COMPLETED and AWAITING sessions, analyzes activities and changeSets, and assigns a weighted confidence score to decide promotion. - Configurable Autonomy Modes: Supports full, approve-high-risk, and approve-all modes with per-agent rules for recurring agents like Bolt and Palette. - Interactive Feedback Handling: Detects AWAITING_USER_FEEDBACK sessions, auto-answers simple confirmations, and drafts codebase-grounded answers to substantive questions via the sendMessage API. - PR Promotion & Handoff: Creates PRs from gitPatch changeSets when Jules did not auto-create one, then hands off to github-babysitter pr-care with a structured payload. - Use Case: Schedule a daily 9am run that scans all Jules sessions across multiple GitHub orgs, promotes high-scoring Bolt performance fixes to PRs, answers a blocking question on a Palette session, and posts a digest digest summarizing promoted, pending, and rejected sessions. ## Quick Start Ask the agent to triage your Jules sessions now, or request to schedule Jules triage daily at 9am for a specific repository.