completely:auto

An autonomous runner that repeatedly takes tasks from a Beads queue and gives each task a fresh coding-agent session. It continues until the queue is empty or the run reaches its limit.

In plain words
What is it for?
Use it to process queued repository tasks in the foreground, with each task handled separately and completed work saved in commits before the task is closed.
Why use it?
It automates the full cycle from understanding a task through testing, review, verification, committing changes, and closing the task. It also supports resuming work after an interrupted run.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/23ag1/completely/auto
Any agent
npx skills add 23ag1/completely --skill auto
Clone the repo
git clone --depth 1 https://github.com/23ag1/completely

Made for: Claude Code, Codex.

Per session 88 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 603 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00088 $0.00603
Opus 5 $0.00044 $0.00302
Sonnet 5 $0.00018 $0.00121
Haiku 4.5 $0.00009 $0.00060

Measured 2d ago against content hash 0f6c901631d0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

completely:auto scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugin/skills/auto/SKILL.md · 32 lines

What it actually says

Run cmpl auto --max N from the repo root to grind the Beads queue with the full task engine (core/task-engine.md): each task gets a fresh claude -p running the COMPLETE recipe (understand → plan-check → parallel subagents → TDD → checks → reviewers → verifier → evaluator → debug-on-fail → commit → close), then the next, until bd ready is empty.

CRITICAL — how to run it so it actually progresses:

  • Run it in the FOREGROUND. Do NOT background it. The nested claude -p workers only make progress while the launching session is active. If you background it (& / run_in_background) and yield your turn, the workers PAUSE. Block on the single cmpl auto call until it exits.
  • It is bounded by the foreground/session limits. A long run can be cut off (session limit). Because the engine commits BEFORE closing a bead, completed tasks are durable (committed work
    • closed beads) and never leave a closed bead with uncommitted code.
  • Resuming after an interruption. The task that was IN FLIGHT when the run died is left in_progress (claimed), NOT open — and bd ready excludes in_progress, so a bare re-run would silently skip it. cmpl auto therefore auto-reaps orphaned claims at startup: any in_progress bead whose holding run is gone/stale (heartbeat) is reopened with an audit comment, then re-dispatched. So just re-run cmpl auto and it genuinely continues. To inspect/clean manually: cmpl orphans (list) / cmpl orphans --reap (reopen). An interrupted worker may also leave uncommitted WIP in the tree — the run-report flags it; salvage or revert it before the next run.
  • Always pass --max N to bound the run; --dry-run previews the queue.

Use when the spec is frozen, tasks are atomic, write-zones disjoint, verification automated. Otherwise use /completely:control (one observed task at a time).

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 32 lines · 88 tokens per session scan A 0f6c901631d0

Subscribe to this mod's changes

completely:auto is a skill published in the GitHub repository 23ag1/completely (5 stars, last pushed 2mo ago), licensed MIT. It adds 88 tokens to every session and 603 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories