queue-workflow

A queue-based workflow for work that needs multiple serious passes of review and improvement. It is designed to catch both stopping too early and adding unnecessary duplicate work.

In plain words
What is it for?
Use it for medium or large code, documentation, skill, runbook, or architecture changes. Each pass inspects the context, performs work, validates the result, simplifies it, and reports clearly.
Why use it?
It makes follow-up work deliberate instead of treating the first successful check as completion. It also helps detect prompt echoing, leaked history, missing tests or documentation, naming drift, and weak fallback behavior.

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/longbkit/clisbot/queue-workflow
Any agent
npx skills add longbkit/clisbot --skill queue-workflow
Clone the repo
git clone --depth 1 https://github.com/longbkit/clisbot

Made for: Claude Code, Codex.

Per session 62 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,608 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 $0.00062 $0.01608
Opus 5 $0.00031 $0.00804
Sonnet 5 $0.00012 $0.00322
Haiku 4.5 $0.00006 $0.00161

Measured 3d ago against content hash 535d96386d8e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

queue-workflow 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 3d 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.

skills/queue-workflow/SKILL.md · 134 lines

How it starts

The opening of the file, as written. The whole thing — 134 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Queue Workflow

Use this skill when a task needs more than one serious pass, or when the user explicitly asks for queue workflow.

Problem

AI can fail in two opposite ways:

  • stop early, act lazy, or treat a green first pass as done
  • keep "improving" by adding duplicate sections, files, functions, fallbacks, names, or explanations

Queue workflow protects against both. It also catches prompt echoing, history leaking, topic drift away from exact continue, shallow queued micro-tasks, and review overload for the human.

The goal is not to split work into tiny tasks. Each pass, including the current pass and every later queue-runner invocation, must inspect context, do the work, validate, simplify, and report clearly.

When To Use

Use queue-first when:

  • the user explicitly asks for queue workflow, even for small work
  • the task is medium or large
  • code, docs, skill, runbook, or architecture changes may sprawl
  • regressions, missing tests/docs, naming drift, or duplicate structure are plausible
  • the user needs a final version, not draft-history explanation inside the artifact

Skip it only when the user did not ask for queue workflow and the task is clearly tiny, low-risk, and fully verifiable in one short pass.

Read references/content-architecture.md as the structural foundation for both the current pass and queued passes.

Core Workflow

Principles:

  • Durable queues are an external clisbot queue-runner flow; the current agent must not drain pending queue items locally.
  • Queuing follow-up work never lowers the quality bar for the current pass.
  • Run the current pass deeply and thoroughly before reporting: inspect state, edit, validate, simplify, and name what is done now.
  • When the queue runner invokes a queued prompt later, that invocation must also run as a full-quality session, not a small visible micro-task.
  • Preserve the current thread with exact continue before switching to a new lens.
  • At queue creation time, do not depend on knowing whether a future pass will finish cleanly; add the control prompts up front after important milestones.
  • Keep final artifacts clean: do not echo user wording or leak draft-history notes unless the user needs a changelog.

Read the full file on GitHub · 134 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 3d ago First seen · 134 lines · 62 tokens per session scan A 535d96386d8e

Subscribe to this mod's changes

queue-workflow is a skill published in the GitHub repository longbkit/clisbot (114 stars, last pushed 3d ago), licensed MIT. It adds 62 tokens to every session and 1,608 once invoked, about $0.0003 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 2026-08-30.