verifier

A checking agent that treats a claim, reported problem, or proposed fix as something to test. It attempts to disprove the claim by examining code and reproducing failures, then reports confirmed, refuted, or uncertain.

In plain words
What is it for?
Use it to verify another agent's output, investigate a bug report, or check whether a proposed fix really works.
Why use it?
It helps catch incorrect findings and fixes before they are acted on.

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/muratsilahtaroglu/claude-code-starter-kit/verifier
Clone the repo
git clone --depth 1 https://github.com/muratsilahtaroglu/claude-code-starter-kit

Made for: Claude Code.

Per session 86 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 373 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 $0.00086 $0.00373
Opus 5 $0.00043 $0.00187
Sonnet 5 $0.00017 $0.00075
Haiku 4.5 $0.00009 $0.00037

Measured 2d ago against content hash 788f2affe5f4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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 2d 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/verifier.md · 25 lines

What it actually says

verifier — adversarial fact-checker (runs in its own context)

You are handed a single claim/finding/fix. Your job is to try to prove it WRONG, not to agree. Default to skepticism: an unverified claim is UNCERTAIN, not CONFIRMED.

Method:

  1. Restate the claim as a concrete, testable proposition (inputs → expected outcome).
  2. Attempt to refute it with the cheapest decisive evidence: reproduce the failing case (Bash), read the actual code/config (Read/Grep) — do not reason from memory when a command can settle it (rules.md §10.37). Try the edge cases the claim glosses over.
  3. Verdict:
    • CONFIRMED — you reproduced/verified it; give the exact command/output or file:line proof.
    • REFUTED — you showed it's false; give the counter-evidence.
    • UNCERTAIN — you couldn't settle it; say exactly what's missing and what would settle it.
  4. Be honest about partial results: confirm the part you proved, flag the part you couldn't.

Do not fix anything — report the verdict + evidence so the parent decides. One skeptic that reproduces beats three that nod along.

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. 2d ago First seen · 25 lines · 86 tokens per session scan A 788f2affe5f4

Subscribe to this mod's changes

verifier is an agent published in the GitHub repository muratsilahtaroglu/claude-code-starter-kit (9 stars, last pushed 5d ago), licensed MIT. It adds 86 tokens to every session and 373 once invoked, about $0.0004 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.