afk

Drains the ready-for-agent issue queue by registering projects with the redskilled daemon.

20|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reddb-io/red-skills --skill afk-reddb-io
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: afk
Source: https://github.com/reddb-io/red-skills/tree/main/plugins/dev/skills/engineering/afk
Command: npx skills add https://github.com/reddb-io/red-skills --skill afk-reddb-io

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Manually working through a labeled issue backlog one ticket at a time is slow and error-prone. This Skill automates the loop: it registers your project with the always-on redskilled daemon, arms a drain over the ready-for-agent queue, and lets daemon-admitted Workers claim, implement, validate, land, and close each issue in isolated git worktrees. ## Core Features & Use Cases - Autonomous queue draining: Register the project at a chosen runner (claude, codex, opencode) and target width, then observe as Workers grind through the backlog with no hand-built processes. - Scoped selectors: Drain only Tickets linked to a Spec, an explicit issue list, tag territories, or issues authored by a specific GitHub user. - Safety-guarded execution: Binding git safety rules, per-file commit discipline, canonical gate validation, and DONE/BLOCKED sentinels keep autonomous work auditable and recoverable. - Use Case: After triaging a Spec into Tickets labeled ready-for-agent, run /afk --spec 42 to have Workers claim each Ticket, implement it via TDD in its own worktree, pass the declared gate, and land it through PR merge while you watch status. ## Quick Start Ask the agent to run /afk to register this project with the redskilled daemon and drain every open issue labeled ready-for-agent.

Frequently Asked Questions about afk

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

FAQPage Schema
How do I run autonomous agents on my GitHub issue backlog?▼

Run /afk to register the project with the redskilled daemon and arm a drain over issues labeled ready-for-agent. The daemon admits Workers that claim each issue, implement it in an isolated worktree, validate against the declared gate, and land it via PR merge.

How do I drain only specific issues or one spec with afk?▼

Use selector flags: /afk --spec 42 drains Tickets linked to that Spec, /afk --issues 356,359 drains an explicit list, --tags filters by tag labels with AND semantics, and --user filters by issue author. Selectors travel in the drain registration.

Which MCP tools does afk actually use today?▼

Only four rs_dev MCP verbs reach the daemon: drain, status, project_status, and project_stop. Every other tool in the surface refuses by name under issue #4113, and a refusal means the verb is unimplemented, not an outage.

Why does afk report the MCP as unreachable after installing the plugin?▼

Hosts register MCP servers at plugin load, so a mid-session install writes the declaration but starts no process. Run /reload-plugins or start a new session first; only if that is ruled out should you repair the daemon via /redskilled.

What git operations are forbidden inside afk workers?▼

Workers may only use git add, commit, status, diff, log, show, and mv. Forbidden everywhere are git reset, rebase, clean, restore, stash, branch -D, force pushes, --hard, --no-verify, branch switching, and any history rewriting; violations abort the run.

When should I use /go instead of /afk?▼

/afk is the default lane for all tracked backlog work; /go is the ad-hoc-only exception for untracked demand dispatches. An issue-form dispatch belongs to the registered drain, while worker_dispatch with a demand text handles one-off work.