agentmods.dev

prd-review

prd-review is a command for Claude Code from assafkip/prd-os. It costs 15 tokens per session (650 once invoked), scanned A, original, MIT.

A command that reviews an active product requirements document (PRD), which describes what a product or feature should do, against a fixed checklist.

In plain words
What is it for?
Checking problem clarity, scope, task breakdown, risks, and dependencies before implementation.
Why use it?
It turns broad review output into consistent findings, with one concern per JSONL record and a defined severity level.

Command for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: mentions Codex.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the prd-os plugin — 1 skill, 14 commands shipped together

not rated 10repo 2mo ago A scan 15 tokens original MIT

Good fit Checking problem clarity, scope, task breakdown, risks, and dependencies before implementation.

Compare 6 commands from other repositories ↓
View source ↗ assafkip/prd-os
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add assafkip/prd-os
Claude Code
/plugin install prd-os

Made for: Claude Code.

Or install prd-os, the plugin that ships this one along with the rest of its 1 skill, 14 commands.

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 prd-review

README.md
[![agentmods](https://agentmods.dev/badge/commands/assafkip/prd-os/prd-review.svg)](https://agentmods.dev/commands/assafkip/prd-os/prd-review)
Your own site
<a href="https://agentmods.dev/commands/assafkip/prd-os/prd-review"><img src="https://agentmods.dev/badge/commands/assafkip/prd-os/prd-review.svg" alt="Measured on agentmods" height="20"></a>
Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 650 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 $0.00015 $0.00650
Opus 5 $0.00008 $0.00325
Sonnet 5 $0.00003 $0.00130
Haiku 4.5 $0.00002 $0.00065

Measured 6d ago against content hash bc66f4e14d78, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

prd-review 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 6d 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/prd-os/commands/prd-review.md · 55 lines

What it actually says

Review the active PRD against templates/review-rubric.md. The rubric defines five dimensions (problem clarity, scope discipline, atomic decomposition, risk surface, dependencies) and the severity scale (blocker | major | minor | nit).

Do not hand-edit the findings JSONL. Do not pass raw Codex output to the writer verbatim. The writer accepts ONLY {severity, body} objects; extra keys are rejected. This is the deterministic normalization layer — its whole purpose is to refuse drifted Codex output shape.

Steps:

  1. Resolve the active PRD:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/prd_runner.py" status

Capture the prd_id and spec_path from the JSON output. If no PRD is active, stop and tell the author to run /prd-start first.

  1. Read the PRD body and templates/review-rubric.md. Run Codex against them. Codex may return prose, markdown, or malformed JSON. That is expected.

  2. Translate Codex's findings into the writer's input shape: a JSON array where every element is EXACTLY {"severity": "blocker|major|minor|nit", "body": "one concrete concern"}. One concern per element. No extra keys, no nesting. If Codex flagged the same concern twice, deduplicate.

  3. Record the review.

    • If the array has at least one item, append via the writer. The writer assigns sequential ids, stamps created_at, sets disposition: pending, validates every field, AND stamps the PRD frontmatter with codex_reviewed_at as a side effect. That stamp is the approval gate's proof that Codex actually ran:
    echo '<JSON_ARRAY>' | \
  python3 "${CLAUDE_PLUGIN_ROOT}/scripts/findings_writer.py" \
  add "<prd-id>" --source codex-review
    • If Codex found nothing (clean pass), do NOT fabricate findings. Stamp the PRD with record-review so the approval gate can still fire:
    python3 "${CLAUDE_PLUGIN_ROOT}/scripts/findings_writer.py" \
  record-review "<prd-id>" --source codex-review

    Source must be codex-review for a standard pass or codex-adversarial for an adversarial one. Without the stamp, /prd-approve will refuse to advance the PRD — that is the intended behavior. Do not try to work around it.

  4. Advance the PRD to in-review so the findings gate will block approval:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/prd_runner.py" advance in-review
  1. Tell the author to triage with /prd-triage. Show the count of new findings and the severity breakdown (or "0 findings, clean pass recorded" if the review was clean).
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. 6d ago First seen · 55 lines · 15 tokens per session scan A bc66f4e14d78

Subscribe to this mod's changes

prd-review is a command published in the GitHub repository assafkip/prd-os (10 stars, last pushed 2mo ago), licensed MIT. It adds 15 tokens to every session and 650 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.

Claude CodePlanningai-workflowfounder-toolsprdproduct-management
Related

Other commands, from other repositories

screens

Structures product UX screen design — inventory, flow, states, and reusable briefs for external design tools — without generating pixels or UI code.

AInsteinsBR/renata · 27 tokens

assumption-test

Surfaces the riskiest business assumption behind the PRD and designs the cheapest test to falsify it before building.

AInsteinsBR/renata · 24 tokens

bug-report

Converts a raw bug report from production into a structured, severity-classified item and routes it to the right next step.

AInsteinsBR/renata · 25 tokens

hypothesis-check

Confronts each PRD hypothesis with the real measured number, forces an explicit verdict, and triggers a concrete action.

AInsteinsBR/renata · 24 tokens

incident

Coordinates the live response to an active production incident — timeline, status, and resolution — then hands off to /renata:retro for the post-mortem.

AInsteinsBR/renata · 33 tokens

metrics

Defines or refines product metrics across 4 layers, each with baseline, formula, source, and a kill criteria tripwire.

AInsteinsBR/renata · 26 tokens