review

A command for deeply reviewing code changes against stated requirements. It checks relevant files, tests, quality gates, and the task plan, then reports findings on screen with file and line references.

In plain words
What is it for?
Use it to review uncommitted changes or the work for a named task, including checks for formatting, linting, types, and tests.
Why use it?
It helps identify requirement gaps, defects, and quality problems before changes are accepted.

Command for Claude Code

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/ecro/embedeval/review
Clone the repo
git clone --depth 1 https://github.com/Ecro/embedeval

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 891 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 $0.00000 $0.00891
Opus 5 $0.00000 $0.00445
Sonnet 5 $0.00000 $0.00178
Haiku 4.5 $0.00000 $0.00089

Measured yesterday against content hash 01366a0f43db, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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 yesterday.

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/review.md · 97 lines

How it starts

The opening of the file, as written. The whole thing — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/review — Code review & quality check

Deep review of task changes. Step 4 of the embedeval 5-stage workflow. Reports to screen only — no ANALYSIS document.

Communication

  • Direct, evidence-based. Cite file:line for every finding.
  • Lead with critical issues, then warnings, then nits.
  • Before analyzing, reframe as: "Does this meet the stated requirements without issues?" — counters confirmation bias toward the author's intent.
  • Don't fold on pushback without new evidence.

Usage

/review [task-slug]       # review files relevant to PLAN-<slug>
/review                   # review uncommitted + last commit

Context

  • Repo: /home/noel/embedeval
  • PLAN (if slug given): plans/PLAN-<slug>.md
  • Standards: CLAUDE.md → "Quality Gates" and "Learned Corrections"
  • Quality gates to verify pass before reviewing:
    • uv run ruff format --check src/
    • uv run ruff check src/
    • uv run mypy src/
    • uv run pytest tests/

Steps

  1. Determine scope

    • Slug provided → Read PLAN; list "Affected files" and "Success criteria". Review those files plus their tests.
    • No slug → git diff HEAD~1 -- ':!plans/' plus uncommitted working tree.
  2. Run quality gates first — if any fail, report and stop. Don't review on top of a broken build.

  3. Invoke specialized reviewers in parallel (via Task tool):

    • reviewer — general code quality and correctness (always)
    • walkthrough-reviewer — trace happy path, errors, edge cases through changed code (when logic changed)
    • side-effect-reviewer — blast radius of API/contract changes (when exports or public APIs changed)
    • test-quality-reviewer — do behavioral changes have matching tests?
    • concurrency-reviewer — threading/locking touched?
    • resource-lifecycle-reviewer — FD/connection/thread lifetimes touched?
  4. Consolidate — merge reviewer outputs, dedup, rank by severity.

Focus areas

  • Correctness: API contracts, edge cases, off-by-one, None handling
  • Security: input validation, secrets, injection
  • Architecture: separation, coupling, cohesion
  • Tests: new behavior has matching tests (CLAUDE.md "new feature = new tests")
  • Embedeval-specific — cross-check against CLAUDE.md "Learned Corrections":
    • scoped_contains scope argument explicit (not the default stripped)?
    • Run-scoped artifacts written under run_dir/, not output_dir/?
    • L1/L2 skipping correctly for non-compilable cases?
    • L4 mutation lambdas avoid hardcoding reference literals?
    • Check regex accepts API variants (k_msleep/k_sleep, printf/printk/LOG_*)?
    • Pydantic v2 uses @computed_field for JSON-serialized derived fields?

Read the full file on GitHub · 97 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. yesterday First seen · 97 lines · 0 tokens per session scan A 01366a0f43db

Subscribe to this mod's changes

review is a command published in the GitHub repository Ecro/embedeval (11 stars, last pushed 26d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 891 tokens. 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.