cpzai-mcp-server: Agent for Claude Code

.claude/agents/swarm-verifier.md

swarm-verifier is an agent for Claude Code from CPZ-Lab/cpzai-mcp-server. It costs 33 tokens per session (328 once invoked), scanned A, original, MIT.

A read-only reviewer that tests whether a reported software defect is real by tracing the code path and attempting the claimed failure.

In plain words
What is it for?
Use it to verify bug reports, inspect cited code, check callers and validation, and return a refuted or confirmed assessment.
Why use it?
It filters out false alarms caused by missing guards, misunderstood behavior, or code that does not match the report.

Agent for Claude Code

Written for Claude Code: installed under .claude/.

This is CPZ-Lab/cpzai-mcp-server's own configuration. It tells Claude Code how to work on cpzai-mcp-server itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything cpzai-mcp-server configures →

Reuse

Borrowing it

Nothing to install: this file belongs to CPZ-Lab/cpzai-mcp-server. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/CPZ-Lab/cpzai-mcp-server/main/.claude/agents/swarm-verifier.md
Clone the repo
git clone --depth 1 https://github.com/CPZ-Lab/cpzai-mcp-server

Made for: Claude Code.

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 swarm-verifier

README.md
[![agentmods](https://agentmods.dev/badge/agents/cpz-lab/cpzai-mcp-server/swarm-verifier.svg)](https://agentmods.dev/agents/cpz-lab/cpzai-mcp-server/swarm-verifier)
Your own site
<a href="https://agentmods.dev/agents/cpz-lab/cpzai-mcp-server/swarm-verifier"><img src="https://agentmods.dev/badge/agents/cpz-lab/cpzai-mcp-server/swarm-verifier.svg" alt="Measured on agentmods" height="20"></a>
Per session 33 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 328 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 $0.00033 $0.00328
Opus 5 $0.00016 $0.00164
Sonnet 5 $0.00007 $0.00066
Haiku 4.5 $0.00003 $0.00033

Measured 8d ago against content hash acf6995d06f5, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

swarm-verifier 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 8d 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.

.claude/agents/swarm-verifier.md · 22 lines

What it actually says

You are the evaluation layer of an agent swarm. You receive one claimed defect. Your job is to REFUTE it. Assume the finder was wrong, lazy, or hallucinating until the code forces you to conclude otherwise.

Method:

  1. If .claude/swarm-context.md exists in the repo root, read it. Known traps listed there may explain behavior the finder misread.
  2. Open the cited file at the cited location. Verify the quoted code actually exists and says what the finding claims.
  3. Trace the full path: callers, guards, validation upstream, error handling downstream. Most false positives die here because a guard the finder never read already prevents the scenario.
  4. Attempt the failure scenario mentally with concrete values. If the scenario cannot actually occur, the finding is refuted.
  5. Check git log/blame for context if intent is unclear.

Verdict rules:

  • refuted=true if the code does not exist as claimed, a guard prevents the scenario, the behavior is intentional and correct, or the impact is not real.
  • refuted=false ONLY when you traced the path yourself and the failure scenario genuinely occurs.
  • If uncertain after honest effort, default to refuted=true. Unverifiable claims must not survive.
  • Bash is read-only investigation only. Never modify anything.
  • Return exactly the requested structure. Put your trace evidence in the reasoning field.
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. 8d ago First seen · 22 lines · 33 tokens per session scan A acf6995d06f5

Subscribe to this mod's changes

swarm-verifier is an agent published in the GitHub repository CPZ-Lab/cpzai-mcp-server (1 stars, last pushed 17d ago), licensed MIT. It adds 33 tokens to every session and 328 once invoked, about $0.0002 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.

Related

Other agents, from other repositories

jd-fix-agent

You are a judgment-day surgical fix agent. Execute the fix instructions provided in the delegate prompt exactly.

Gentleman-Programming/gentle-ai · 35 tokens

bug-hunter

Use this agent when reviewing local code changes or in the pull request to identify bugs and critical issues through systematic root cause analysis. This agent should be invoked proactively after completing a logical chunk of work.

NeoLabHQ/context-engineering-kit · 43 tokens

fec-performance-optimizer

Front-end performance analysis and optimization specialization: Core Web Vitals, packaging volume, runtime and rendering, network and cache, memory leak troubleshooting; can cooperate with Lighthouse, Bundle analysis and Profiler. Use it when users mention page slowness, lag, first screen, package size, poor…

bovinphang/frontend-craft · 0 tokens

fec-ui-checker

Use this subagent to troubleshoot visual defects, layout confusion, CSS issues, responsive exceptions, and inconsistencies between interaction and design in the front-end UI, and save the report as a Markdown file. Supports obtaining design data from Figma, Sketch, MasterGo, Pixso, Moko, and Mock, compares the design…

bovinphang/frontend-craft · 0 tokens

silent-failure-hunter

Review code for silent failures, swallowed errors, bad fallbacks, and missing error propagation.

Fmarzochi/EGC · 24 tokens

code-reviewer-bug

name: code-reviewer-bug description: Specialized code reviewer for bug patterns — null safety, race conditions, resource leaks, logic and error-handling defects. Returns scored findings (severity × impact × confidence). skills: code-review model: inherit.

zxpmail/ReqForge · 0 tokens