yofine/mexus/run
Claude Code and agent skill by yofine/mexus
- community
Checking that the SKILL.md file is still in the repository.
Use when executing the next tasks from a Markdown-backed Agent Team kanban with Claude Code or Codex background Agents.
Install
npx vskill@latest install yofine/mexus/runSKILL.md
/mexus-skill:run
Use this skill when the user invokes /mexus-skill:run.
This command executes work from the active Agent Team kanban. It does not create a Mission and it does not start the board. By default it targets the latest active Mission created by /mexus-skill:team.
Steps
- Locate the active mission under
agent-team/missions/, or use--name <mission-name>if provided. If no active lifecycle marker exists, fall back to the newest mission. - Read:
agent-team/mission-workflow.mdagent-team/agents.mdagent-team/missions/<mission-name>/mission.mdagent-team/missions/<mission-name>/agents.mdagent-team/missions/<mission-name>/kanban.mdagent-team/missions/<mission-name>/roundtable.mdagent-team/missions/<mission-name>/squad-lead.md
- Act as Squad Lead for task selection and dispatch. Do not execute or fill another Agent's task result.
- Select the next actionable tasks from
To Claim. - Use the Agent tool to start background Agents in parallel for the selected task batch. Respect
--limit <count>when present; otherwise dispatch a small safe batch. - Require every background Agent to claim, update, and complete its own task block in
kanban.mdbefore reporting back. - Review completed tasks only when the task publisher is Squad Lead, or when the mission workflow says the current lead owns the acceptance standard.
- Perform mission-level acceptance after publisher reviews. If the overall result is below expectation, publish new focused tasks to the responsible Agents.
Background Agent Prompt Requirements
Every background Agent prompt must include:
- Mission name.
- Agent name and responsibility from mission
agents.md. - Agent names must follow
agent-team-mission-workflow: short stable names from the Ars Goetia / Lesser Key of Solomon name set. - Exact current task block from
kanban.md. - A clear statement that this is the first/current assigned task, not all future work.
- Requirement to read the mission files before editing.
- Requirement to claim the task by moving the full task block to
In Progress. - Requirement to write
Result,Files,Verification, andUpdatedin its own task block. - Requirement to move the task to
Doneonly after verification. - Requirement to leave
Reviewfor the publishing Agent named inFrom. - Instruction to prioritize kanban-assigned work.
- If the task is outside its responsibility and the correct owner is clear, reassign it in
kanban.mdwith a brief reason inUpdated. - If ownership is unclear, publish a clarification task assigned to the Squad Lead.
- No Mexus server, pane dispatch, Mission Inbox, or external A2A dependency.
Rules
- Kanban is the source of truth.
Squad Leadis a fixed coordination role name and is not an execution Agent name.- Kanban task
Frommust be Squad Lead or an executing Agent, neverUser. - Every work item follows publisher acceptance: publisher creates task, executor completes and self-tests, publisher fills
Review. - Squad Lead owns mission-level acceptance and publishes new tasks for gaps instead of rewriting another Agent's task result.
- After meaningful completed work is accepted, update the repository-level
agent-team/agents.mdroster with the Agent's durable module ownership, known strengths, and Work history. Do not put mission-specific prompts in the repository roster. - By default, execute only the current active Mission. Archived Missions are historical unless the user explicitly passes
--name. - The executing background Agent updates the kanban task it owns. The Squad Lead does not fill execution results for that Agent.
- Do not execute tasks that are already
In Progressunless the user explicitly asks for recovery. - Do not mutate board process state.
- Do not invent agent names when a suitable mission agent already exists.
- Publish follow-up tasks instead of crossing scope boundaries.
- Use roundtable.md for decisions that affect multiple Agents, shared interfaces, mission scope, product direction, sequencing, acceptance criteria, or major risk trade-offs. Roundtable is not a default kanban blocker.
Publisher Review
When no assigned To Claim task remains:
- Check
Donefor tasks whereFromis the current Agent andReviewhas no accepted marker. - Review only those tasks.
- If accepted, fill
Review. - If rejected, publish a focused fix task under
To Claim.
Output
After dispatching or executing work, report:
- Mission name.
- Tasks selected.
- Background Agents dispatched.
- Any tasks reassigned or clarification tasks created.
- Verification status for completed tasks.
Security
- Security scanPassed, 91 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/mexus-agent-team/skills/run/SKILL.md
Popularity
- Repository stars
- 83
- Repository forks
- 4
- 7-day trend score
- 22
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/yofine/mexus/run)