reverse-spec

reverse-spec is a command for Claude Code from Zts0hg/codexspec. It costs 23 tokens per session (9,778 once invoked), scanned A, original, MIT.

A command for reverse specification: examining an existing project and producing documentation that describes its behavior and design.

In plain words
What is it for?
It is for reverse-engineering a project, creating specification documents, and generating reconciliation reports.
Why use it?
It helps turn undocumented code into a written reference and compare the result with an expected baseline.

Command for Claude Code

Written for Claude Code: allowed-tools in frontmatter.

Good fit It is for reverse-engineering a project, creating specification documents, and generating reconciliation reports.

Compare 6 commands from other repositories ↓
Install with agentmods
npx agentmods add commands/zts0hg/codexspec/reverse-spec
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.

Clone the repo
git clone --depth 1 https://github.com/Zts0hg/codexspec

Made for: Claude Code.

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 reverse-spec

README.md
[![agentmods](https://agentmods.dev/badge/commands/zts0hg/codexspec/reverse-spec/github.svg)](https://agentmods.dev/commands/zts0hg/codexspec/reverse-spec)
Your own site
<a href="https://agentmods.dev/commands/zts0hg/codexspec/reverse-spec"><img src="https://agentmods.dev/badge/commands/zts0hg/codexspec/reverse-spec/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for reverse-spec

Your own site · 80×15
<a href="https://agentmods.dev/commands/zts0hg/codexspec/reverse-spec"><img src="https://agentmods.dev/badge/commands/zts0hg/codexspec/reverse-spec.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 23 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 9,778 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00023 $0.09778
Opus 5.5 $0.00009 $0.03911
Sonnet 5.5 $0.00005 $0.01956
Haiku 4.5 $0.00002 $0.00978

Measured 11d ago against content hash ff932b243c73, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

reverse-spec 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 11d 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.

.claude/commands/codexspec/reverse-spec.md · 688 lines

How it starts

The opening of the file, as written. The whole thing — 688 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Reverse Specification

Language Preference

Read .codexspec/config.yml. Two independent language controls apply (each falls back to language.output, then English):

  • Interaction language (language.interaction): language for all conversation with the user — questions, explanations, status messages, and codexspec CLI terminal output.
  • Document language (language.document): language for generated artifact files (the reverse-derived spec/design and the reconcile report).

Converse in the interaction language and author artifacts in the document language. Apply the project's translation standard to both: translate by meaning (not word-for-word), keep English for terms with no good native equivalent, and write as if originally in that language. Exception: in reconcile.md, location and evidence quote the code and the baseline verbatim — path, line, and the quoted spans on both sides — and MUST NOT be translated. A translated quote can no longer be checked against its source, which is exactly what the both-side evidence rule exists to make possible. Verbatim does not mean secret-bearing: apply the global sensitive-value redaction rule under Instruction and Evidence Trust before persisting or briefing any observation.

Expression Standard

IMPORTANT: Everything this command produces — documents, reviews, reports, commit messages, diagnostics, and replies — is written for a reader who cannot see your working context. Apply the rules below to every artifact and message you output:

  • Write for the reader at hand-off. Every reference must resolve without access to this session: no session-only identifiers or section numbers, no narration of what changed during the conversation, no arguments with absent reviewers. State current reality and cite committed, reachable sources.
  • Preserve every proposition. Before summarizing or trimming, list the facts a passage carries: actors, conditions, ordering, modalities (must, never), negative guarantees, and consequences. Remove only reasoning transcripts, repetition, and decoration. Shorter is not clearer if any fact is lost.
  • State what the surface requires. Diagnostics name what failed, which rule was violated, and the correction. Problem reports carry the defect, its location, its impact, and the evidence. Decisions record the alternatives they beat. Rejections give the reason in one line. Shipped work is described in the present tense; plans and open questions are labeled as such.
  • Define terms before relying on them. Prefer the concrete rule, field, or behavior over a coined label; give a project-specific term a plain-language definition at first use, then use it consistently.
  • Be honest, not agreeable. Verify claims before accepting them; fix or rebut on technical grounds. One substantiated blocker is worth more than a list of nitpicks. When a decision is needed, present only viable options, recommend one, and state the real difference between them.
  • Declare what is binding. Say explicitly which instructions are hard requirements and where judgment is required. Keep one explanation in one place and link to it, but keep at the point of use the contract a reader needs there.

Read the full file on GitHub · 688 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. 11d ago Changed · +1 tokens per session ff932b243c73
  2. 18d ago Changed · +15 lines e1d112595f3e
  3. 24d ago Changed · +72 lines bfa8f5e6ef1a
  4. 1mo ago Changed · +11 lines 12199c19e75d
  5. 1mo ago First seen · 590 lines · 22 tokens per session scan A 7dd7d7ef7510

Subscribe to this mod's changes

reverse-spec is a command published in the GitHub repository Zts0hg/codexspec (6 stars, last pushed 10d ago), licensed MIT. It adds 23 tokens to every session and 9,778 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-09-03.