spec-dual

A command that reviews a code change with two independent review engines and repeats fixes until both report no medium-or-higher findings.

In plain words
What is it for?
Use it to review specification paths, combine findings from Claude and Codex, and run a fix loop. A separate openai-codex plugin is needed for the Codex review engine.
Why use it?
It reduces the chance that one reviewer misses a problem by treating significant findings from either engine as blockers.

Command

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/shdennlin/agent-plugins/spec-dual
Clone the repo
git clone --depth 1 https://github.com/shdennlin/agent-plugins
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 1,398 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 $0.01398
Opus 5 $0.00010 $0.00699
Sonnet 5 $0.00004 $0.00280
Haiku 4.5 $0.00002 $0.00140

Measured 2d ago against content hash 554a754240d9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

spec-dual 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.

plugins/reviewer/commands/spec-dual.md · 135 lines

How it starts

The opening of the file, as written. The whole thing — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Dual-Engine Spec Review Command

Review a change with two independent engines (Claude + Codex) in parallel and drive fixes off the union of their findings. A finding only one engine sees is signal, not noise. The blocker rule is strict: MEDIUM or above in EITHER engine is a blocker, and a PASS from one engine alone is not enough.

This command dispatches a Workflow that owns the fan-out, the cross-engine union, the pure-code blocker count, and the fix loop — so there is no per-turn Stop hook to burn and no completion string to fake.

Dependencies

  • reviewer:spec-fixer — ships with this plugin (applies fixes).
  • codex:codex-rescue — from the external openai-codex plugin (the Codex engine). Without it, the Codex engine degrades to empty findings and the review becomes Claude-only; tell the user to install openai-codex for true dual-engine review.

Instructions

Step 1: Parse Arguments

From $ARGUMENTS, extract:

  1. help: if --help or -h is present, show the Help Output below and stop. Do NOT dispatch.
  2. paths: collect all positional arguments (not flags or flag values). Join multiple paths into one space-separated string as change.
  3. max_rounds: integer after -n or --max-rounds, default 3.
  4. no_explore: true if --no-explore is present.

Help Output

If --help or -h is present, display this and stop:

Usage: /reviewer:spec-dual [path...] [options]

Dual-engine (Claude + Codex) spec review. Both engines review the same change in
parallel; fixes loop until BOTH engines report zero MEDIUM+ findings, then LOW is cleared.

Positional arguments:
  path...                       One or more change folders or spec files to review

Options:
  -h, --help                    Show this help message
  -n, --max-rounds <N>          Maximum review→fix rounds (default: 3, hard cap)
  --no-explore                  Skip the shared codebase-context scan

Requires the openai-codex plugin for the Codex engine. Without it, review is Claude-only.

Examples:
  /reviewer:spec-dual openspec/changes/my-change/
  /reviewer:spec-dual proposal.md spec.md tasks.md -n 5
  /reviewer:spec-dual openspec/changes/my-change/ --no-explore

Read the full file on GitHub · 135 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. 2d ago First seen · 135 lines · 20 tokens per session scan A 554a754240d9

Subscribe to this mod's changes

spec-dual is a command published in the GitHub repository shdennlin/agent-plugins (2 stars, last pushed 8d ago), licensed MIT. It adds 20 tokens to every session and 1,398 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 2026-08-31.