← ClaudeAtlas

autonomous-execution-mandatelisted

Self-directed autonomous execution mandate and workflow skill for ThumbGate agents. Enforces zero-human-intervention task chaining, proactive CI/Trunk monitoring, automated conflict resolution, and evidence-first reporting without requiring manual prompt nudges from the user.
IgorGanapolsky/ThumbGate · ★ 26 · AI & Automation · score 75
Install: claude install-skill IgorGanapolsky/ThumbGate
# ⚡ Autonomous Execution Mandate & Workflow Skill This skill enforces autonomous, self-directed operation across all coding tasks, branch management, and release workflows. --- ## 🎯 Core Execution Principles ### 1. Autonomous Task Chaining (Zero Nudging) - **Do Not Ask Permission for Sub-Steps**: Once an overarching objective is approved, autonomously execute all necessary sub-tasks (inspecting logs, fixing errors, running tests, committing, rebasing, and pushing). - **Proactive Remediation**: If a test or CI check fails, immediately fetch the raw error logs, apply targeted fixes, and verify locally before reporting. ### 2. Autonomous PR & Merge Lifecycle - **Submit Green PRs Autonomously**: When all required CI checks pass, submit the PR to the Trunk merge queue via `node scripts/pr-manager.js <PR_NUM>` (or `/trunk merge`) without asking the user to trigger it. - **Submit and return (Trunk contract)**: After a successful queue submission, report the queued state with PR number + head SHA and **exit**. Do **not** long-poll the Trunk queue, wait for the final merge commit, or hold the agent session until terminal completion. Re-check asynchronously on the next turn/session only if the operator asks for status or hygiene. - **Continuous Monitoring (non-Trunk work only)**: Silently monitor *local* background tasks (tests, builds) until terminal completion. Trunk merge queue completion is asynchronous and out-of-band. ### 3. Strict Non-Overclaim Verification - **Evidence R