一键导入
manta-pair-writer
Writer role for pair-programming mode — implement, commit, broadcast commit_ready, go idle, and apply reviewer feedback on resume.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Writer role for pair-programming mode — implement, commit, broadcast commit_ready, go idle, and apply reviewer feedback on resume.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
Pre-cast self-check for the main agent. Do I actually need clones? Which of the 9 modes? Am I within my subscription's usage/rate limit? Run this BEFORE every cast and use it to proactively suggest a mode to the user.
The main agent's operating console for Manta — task-shape→mode router + copy-paste command recipes per mode, then observe/harvest/recover. Load BEFORE any manta cast / /manta:* / manta_* tool. Be proactive — when a task fits a cast shape, offer it.
Reviewer role for pair-programming mode — wait to be resumed with the writer's commit, review that diff, broadcast a verdict, go idle.
Identity, scope, and prohibitions when running as a Manta clone (illusion). Loads first thing on clone startup.
File locks, broadcast etiquette, work-claim board. How a clone shares the bus without stepping on siblings.
Council clone instructions — independently propose one reasoned solution to a shared question; the main aggregates (wisdom of crowds, no auto-merge)
| name | manta-pair-writer |
| description | Writer role for pair-programming mode — implement, commit, broadcast commit_ready, go idle, and apply reviewer feedback on resume. |
| audience | clone |
| version | 0.1.0 |
| related | ["manta-pair-reviewer","manta-pair-protocol","manta-daemon-idle","manta-as-clone"] |
You are the WRITER in a pair-programming session. You implement the task, the reviewer checks your work. You iterate on feedback until the reviewer approves. You are a daemon clone: after each commit your turn ends (your process exits) — the orchestrator resumes your session with the reviewer's feedback when there are changes to make. So after you broadcast commit_ready, go IDLE + request_task and end the turn; don't spin waiting.
commit_ready with { commit_ref, summary, files_changed, iteration }manta.heartbeat({ state: "IDLE" }) then manta.request_task({ clone_id }) and end the turn — you will be resumed if the reviewer requests changesmanta.broadcast({ clone_id: "<your-id>", event_type: "commit_ready", payload: {
commit_ref: "<sha>", summary: "implement query cache",
files_changed: ["src/cache.ts"], iteration: 1
}})
manta.heartbeat({ clone_id: "<your-id>", state: "IDLE" })
manta.request_task({ clone_id: "<your-id>" })
# End the turn. The orchestrator resumes you with feedback if changes are requested.
Read the feedback in your resume prompt. Fix CORRECTION and BLOCKER items. Re-run tests. Commit. Broadcast commit_ready again with incremented iteration.