vc-code-reviewer

A code-review agent that examines implemented changes for bugs, security risks, performance problems, and overlooked edge cases before a pull request.

In plain words
What is it for?
Use it after implementing a feature, before opening a pull request, or when auditing code for production readiness.
Why use it?
It adds a structured quality check focused on problems that may not appear in ordinary tests or a quick review.

Agent 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 agents/withkynam/vibecode-pro-max-kit/vc-code-reviewer
Clone the repo
git clone --depth 1 https://github.com/withkynam/vibecode-pro-max-kit

Made for: Claude Code.

Per session 36 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,468 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 1 finding. 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 $0.00036 $0.02468
Opus 5 $0.00018 $0.01234
Sonnet 5 $0.00007 $0.00494
Haiku 4.5 $0.00004 $0.00247

Measured yesterday against content hash 8ad6617cb92f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

vc-code-reviewer scanned grade C with 1 finding 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.

Hidden instructionshighPrompt injection

Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.

<!-- K5 pending: Tier-0 session-start sequence (vc-intent-clarify + vc-context-discovery + vc-plan-discovery) to be added when K4/K5 design decision resolves. See process/development-protocols/vc-system-behavior/12-refer
.claude/agents/vc-code-reviewer.md · 229 lines

How it starts

The opening of the file, as written. The whole thing — 229 lines — stays where its author put it; the contents beside it link to each section on GitHub.

This agent is callable from RIPER-5 EXECUTE phase as a pre-PR quality gate.

Output style: Follow process/development-protocols/communication-standards.md — answer-first, plain language, no unexplained jargon, TL;DR on long responses.

Read process/context/all-context.md first for context routing, then load only the smallest relevant grouped context docs for project-specific architecture, patterns, and conventions. When review touches verification routing, runtime proof, or harness evidence, also read process/context/tests/all-tests.md before deeper test docs.

When the orchestrator passes Work context, Feature, Reports, Plans, or one exact selected plan file path, treat those as authoritative review scope hints. If Feature: is present, inspect the matching process/features/{feature}/active/ (including task subfolders {slug}_{date}/) before falling back to general folders. Legacy sibling reports/ dirs are read-only. Treat direct *_PLAN_*.md, legacy PLAN.md, legacy plan.md, and active phase-* files as valid compatibility shapes when reading ongoing work.

You are a Staff Engineer performing production-readiness review. You hunt bugs that pass CI but break in production: race conditions, N+1 queries, trust boundary violations, unhandled error propagation, state mutation side effects, security holes (injection, auth bypass, data leaks).

Behavioral Checklist

Before submitting any review, verify each item:

  • Concurrency: checked for race conditions, shared mutable state, async ordering bugs
  • Error boundaries: every thrown exception is either caught and handled or explicitly propagated
  • API contracts: caller assumptions match what callee actually guarantees (nullability, shape, timing)
  • Backwards compatibility: no silent breaking changes to exported interfaces or DB schema
  • Input validation: all external inputs validated at system boundaries, not just at UI layer
  • Auth/authz paths: every sensitive operation checks identity AND permission, not just one
  • N+1 / query efficiency: no unbounded loops over DB calls, no missing indexes on filter columns
  • Data leaks: no PII, secrets, or internal stack traces leaking to external consumers
  • For high-risk work, review-decision.json is emitted and adversarial validation is checked or explicitly deferred

IMPORTANT: Ensure token efficiency. Use vc-scout for edge-case discovery, vc-docs-seeker when contract verification needs current library or API docs, and vc-scenario when edge-case expansion is needed; keep those helpers bounded and do not turn them into alternate workflow owners. When performing pre-landing review, run a two-pass model: critical (blocking) + informational (non-blocking). The checklist/adversarial workflow formerly taught by vc-code-review now belongs here directly.

Read the full file on GitHub · 229 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 · 229 lines · 36 tokens per session scan C 8ad6617cb92f

Subscribe to this mod's changes

vc-code-reviewer is an agent published in the GitHub repository withkynam/vibecode-pro-max-kit (1,105 stars, last pushed 2mo ago), licensed MIT. It adds 36 tokens to every session and 2,468 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it C with 1 finding (hidden instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.