skills.rest
Vimurai
VimuraiCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

ai-task

Marks completed E-## tasks DONE, runs handover, and surfaces the next task in AI-OS.

Updated Feb 26, 2026
Search Tags:#task management#workflow automation#mcp#handover#ci gate#task lifecycle#ai-os
One-click install
npx skills add https://github.com/Vimurai/ai-os --skill ai-task-vimurai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-task
Source: https://github.com/Vimurai/ai-os/tree/main/.claude/skills/ai-task
Command: npx skills add https://github.com/Vimurai/ai-os --skill ai-task-vimurai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Closing out engineering tasks in AI-OS requires several coordinated steps — updating task status, verifying CI results, running handover, and signaling the Architect when the queue is empty. This Skill automates that entire lifecycle so no step is forgotten after an E-## implementation. ## Core Features & Use Cases - Task Status Management: Marks completed E-## tasks as DONE through task-synchronizer-mcp with a required summary, refusing to close tasks with failing tests or unpassed gates. - CI Gate Enforcement: Checks the local CI status of HEAD before marking anything DONE, blocking completion when the commit is untested, failing, or only dirty-tested. - Handover & Queue Handoff: Runs the orchestrator handover to stamp implementation deltas for Architect review, then either continues with the next open task or emits a handoff signal back to the Architect when the queue is exhausted. - Use Case: After finishing an E-## implementation, invoke this Skill to verify CI passed, mark the task DONE, run handover, and automatically hand control back to the Architect if no tasks remain. ## Quick Start Mark the E-## task I just finished as DONE, run the handover, and tell me which task to work on next.

Frequently Asked Questions about ai-task

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I mark an AI-OS task as DONE after implementation?▼

Invoke the ai-task skill after completing an E-## implementation. It calls task-synchronizer-mcp's update_task_status with status DONE and a one-line summary, but only if tests pass and required gates like the CI gate have been cleared.

Why does update_task_status refuse to mark my task DONE?▼

The DONE gate rejects completion when HEAD has no green local CI run, returning a [CI_GATE] error. Run ai ci run to test the committed HEAD in a clean worktree, or set AI_OS_CI_GATE=0 to explicitly disable the gate.

What happens when all E-## tasks are finished in AI-OS?▼

The skill hands control back to the Architect by running the ai-handoff skill or emitting ai handoff architect with a summary message. This signal keeps the autonomous ping-pong loop running instead of waiting for a human keypress.

Can I edit TASKS.md directly instead of using the task synchronizer?▼

No, the skill explicitly forbids modifying .ai/TASKS.md directly. All status changes must go through task-synchronizer-mcp so the SQLite state and markdown file stay synchronized, with verify_markdown_sync available to check consistency.

Does a passing working-tree test run count as CI verification?▼

No, only a clean ai ci run against the committed HEAD counts as verification. Working-tree passes can hide environment-specific state, so DIRTY or NONE CI statuses block reporting a task as DONE-and-verified.

Related Skills

Explore More Software Engineering Skills→
abnegateabnegate
2

implement

Implement features through test-driven development with review cycles.

Community
Advanced
virajpvirajp
1

plan

Generate reviewable implementation plans from product blueprints and existing code.

Community
Advanced
axone-protocolaxone-protocol
124

rust-quality-gates

Execute Rust project quality gates for format, lint, tests, coverage, schema, and docs.

Official
Intermediate
openaiopenai
5.3k

connect-chatcut-desktop

Installs and connects ChatCut Desktop to restore missing chatcut_desktop MCP tools.

Official
Intermediate