issue-review

issue-review is a command for coding agents from assafkip/prd-os. It costs 21 tokens per session (1,240 once invoked), scanned A, original, MIT.

A command for running the required native and adversarial reviews for an active software issue.

In plain words
What is it for?
Use it to review in-scope code, claim available review slots, and run the issue review process in its required order.
Why use it?
It checks that the issue is verified and limits review work to the files approved for that issue.

Command

Installs and runs on its own, but its text points at files inside its plugin — anything it tells you to read at a ${CLAUDE_PLUGIN_ROOT} path is only there once the plugin is installed. Installing the plugin gets both.

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

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/assafkip/prd-os/issue-review
Clone the repo
git clone --depth 1 https://github.com/assafkip/prd-os

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 issue-review

README.md
[![agentmods](https://agentmods.dev/badge/commands/assafkip/prd-os/issue-review.svg)](https://agentmods.dev/commands/assafkip/prd-os/issue-review)
Your own site
<a href="https://agentmods.dev/commands/assafkip/prd-os/issue-review"><img src="https://agentmods.dev/badge/commands/assafkip/prd-os/issue-review.svg" alt="Measured on agentmods" height="20"></a>
Per session 21 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,240 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 5.1 $0.00021 $0.01240
Opus 5 $0.00010 $0.00620
Sonnet 5 $0.00004 $0.00248
Haiku 4.5 $0.00002 $0.00124

Measured 5d ago against content hash 595030de1f86, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

issue-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 5d 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/issue-review.md · 82 lines

How it starts

The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Run the required reviews for the active DSSE issue. Execute in order:

  1. Run python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_runner.py" status. Confirm an issue is loaded AND receipts.verified is set. If verified is null, stop. Tell the founder to run /issue-verify first. Reviewing unverified code wastes Codex runtime.

  2. Pull the snapshotted scope. The runner stamped allowed_files into state at load/approve so mid-issue spec edits cannot expand the review surface:

    ALLOWED=$(python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_runner.py" allowed-files)
    ISSUE_ID=$(python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_runner.py" status | python3 -c "import sys,json; print(json.load(sys.stdin).get('issue_id',''))")
    

    $ALLOWED is a JSON array (e.g. ["example-project/.q-system/agent-pipeline/schemas/**","example-project/.q-system/scripts/**"]). If empty, stop and tell the founder the spec has no scope.

  3. Try to claim a "standard" review slot before invoking Codex:

    python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_runner.py" record-review standard
    
    • Exit 0: slot claimed, proceed to step 4.
    • Exit 2: cap reached (default 2 standard rounds). The runner names the cap and the override env. Stop and ask the founder if they want to opt in via ISSUE_ALLOW_REVIEW_REPEAT=1 for a single retry, OR triage and close with the existing review.
  4. Invoke codex:review (Codex native review) against origin/main. Build the focus text with the scope filter inline so Codex stays inside the contracted surface:

    Scope filter: $ALLOWED
    Limit findings to changes inside these paths. Code outside these paths is out of contract for this issue. If a finding requires touching files outside this list, mark it out-of-scope explicitly and do not propose a patch.
    

    Use --base origin/main --background unless the diff is trivially small (1-2 files). When the command returns, wait for completion via codex:result / codex:status. Paste the full verdict block to the founder verbatim.

    Immediately pipe the standard-review findings to disk (compaction hardening: if the conversation compacts after this point, the findings survive because they are on disk, not in narrative memory). Translate Codex's free-form verdict into [{severity, body, affected_path}]. The writer assigns sequential ids, stamps created_at, marks out_of_scope=true for paths outside $ALLOWED, and sets disposition=pending:

    echo '<JSON_ARRAY>' | python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_findings.py" \
      add "$ISSUE_ID" --source codex-review --allowed-files-json "$ALLOWED"
    

    If the standard review returned approve with no findings, skip the writer call for this source. Do this before invoking the adversarial pass in step 6.

  5. Try to claim an "adversarial" review slot:

    python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_runner.py" record-review adversarial
    
    • Exit 0: slot claimed, proceed to step 6.
    • Exit 2: cap reached (default 1 adversarial round per slice). Stop. Ask the founder if they want to opt in via ISSUE_ALLOW_REVIEW_REPEAT=1 OR proceed straight to closeout.
  6. Invoke codex:adversarial-review against origin/main with focus text built from $ALLOWED plus the contract directive:

    Scope filter: $ALLOWED
    Contract slice: this issue ships ONLY the change visible in the diff above. Limit findings to defects inside these paths. Do not raise edge cases that would require new follow-up issues. Do not flag pre-existing patterns in unchanged code. Do not propose architectural rewrites. If you have no defect inside the contract slice, return approve.
    

    Same --base origin/main --background pattern. Wait for completion. Paste verdict verbatim.

    Immediately pipe the adversarial-review findings to disk (same compaction reason as step 4):

    echo '<JSON_ARRAY>' | python3 "${CLAUDE_PLUGIN_ROOT}/scripts/issue_findings.py" \
      add "$ISSUE_ID" --source codex-adversarial --allowed-files-json "$ALLOWED"
    

Read the full file on GitHub · 82 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. 5d ago First seen · 82 lines · 21 tokens per session scan A 595030de1f86

Subscribe to this mod's changes

issue-review is a command published in the GitHub repository assafkip/prd-os (10 stars, last pushed 2mo ago), licensed MIT. It adds 21 tokens to every session and 1,240 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.