auto-reason

auto-reason is a command for Codex from marcusquinn/aidevops. It costs 20 tokens per session (1,248 once invoked), scanned A, original, MIT.

A command that asks several models or reviewers to improve and judge an answer, then chooses whether to replace the current version. It can work from a one-line request, an existing answer, or a saved program.

In plain words
What is it for?
Use it to compare architecture choices, arguments, or other answers where quality is partly a matter of judgment. It can generate revisions, ask judges to score them, combine the results, and stop when improvement is not justified.
Why use it?
It provides a structured way to refine subjective answers and check proposed revisions without relying on one judgment. Blind judging helps reduce the influence of which version was written first.

Command for Codex

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 commands/marcusquinn/aidevops/auto-reason
Clone the repo
git clone --depth 1 https://github.com/marcusquinn/aidevops

Made for: Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for auto-reason

README.md
[![agentmods](https://agentmods.dev/badge/commands/marcusquinn/aidevops/auto-reason.svg)](https://agentmods.dev/commands/marcusquinn/aidevops/auto-reason)
Your own site
<a href="https://agentmods.dev/commands/marcusquinn/aidevops/auto-reason"><img src="https://agentmods.dev/badge/commands/marcusquinn/aidevops/auto-reason.svg" alt="Measured on agentmods" height="20"></a>
Per session 20 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,248 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 $0.00020 $0.01248
Opus 5 $0.00010 $0.00624
Sonnet 5 $0.00004 $0.00250
Haiku 4.5 $0.00002 $0.00125

Measured 3d ago against content hash 9f367f45bf7f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

auto-reason 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.

.agents/scripts/commands/auto-reason.md · 179 lines

How it starts

The opening of the file, as written. The whole thing — 179 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Run provider-agnostic subjective self-refinement: create or accept an incumbent answer, generate an adversarial revision and synthesis, then use fresh blind judges to decide whether to change or stop.

Arguments: $ARGUMENTS

Invocation Patterns

Pattern Example Behaviour
One-liner /auto-reason "decide the best architecture for X" Build a temporary program and run now
--program <path> /auto-reason --program todo/research/reason-product-strategy.md Run from a saved program
--incumbent <path> /auto-reason --incumbent draft.md "improve this argument" Use existing answer as A
--judges <list> /auto-reason --judges simple,standard,openai/gpt-5.5 "..." Override judge tiers/models
Bare /auto-reason Interactive setup

Step 1: Resolve Invocation

if $ARGUMENTS contains "--program ":    → Program Mode
elif $ARGUMENTS contains "--incumbent ": → Incumbent Mode
elif $ARGUMENTS is non-empty:             → One-Liner Mode
else:                                     → Interactive Setup

Step 2: Interactive Setup

Ask sequentially; show inferred default as option 1; Enter accepts default.

Q1 — What decision, answer, or artifact should be refined?

Capture a clear task prompt. If the user supplies a file path, read it as the incumbent A only after verifying the path exists.

Q2 — What rubric should judges use?

Default rubric:

1. Correctness: answers the actual question and avoids factual errors.
2. Usefulness: gives actionable, decision-grade output.
3. Restraint: avoids scope creep, unnecessary expansion, and decorative rewrites.
4. Clarity: concise structure, explicit trade-offs, no vague hedging.

Domain-specific additions:

Domain Extra rubric
Architecture maintainability, reversibility, integration cost
Strategy evidence quality, risk coverage, opportunity cost
Prose audience fit, voice consistency, logical flow
Policy enforceability, edge cases, security posture
Review synthesis finding validity, severity calibration, non-duplication

Read the full file on GitHub · 179 lines

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 · 179 lines · 20 tokens per session scan A 9f367f45bf7f

Subscribe to this mod's changes

auto-reason is a command published in the GitHub repository marcusquinn/aidevops (391 stars, last pushed today), licensed MIT. It adds 20 tokens to every session and 1,248 once invoked, about $0.0001 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.