research-verifier

research-verifier is an agent for Claude Code from oborchers/fractional-cto. It costs 238 tokens per session (1,889 once invoked), scanned A, original, MIT.

A research-checking agent that independently reviews another research worker’s document by checking important sources, facts, numbers, and citations.

In plain words
What is it for?
Use it after research workers finish to recheck key sources, statistics, funding figures, benchmarks, feature claims, prices, and licenses.
Why use it?
It catches incorrect claims or citations before they appear in the final research output.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter; positional $N argument.

Part of the deep-research plugin — 5 skills, 1 command, 3 agents shipped together

Good fit Use it after research workers finish to recheck key sources, statistics, funding figures, benchmarks, feature claims, prices, and licenses.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/oborchers/fractional-cto/research-verifier
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.

Clone the repo
git clone --depth 1 https://github.com/oborchers/fractional-cto

Made for: Claude Code.

Or install deep-research, the plugin that ships this one along with the rest of its 5 skills, 1 command, 3 agents.

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 research-verifier

README.md
[![agentmods](https://agentmods.dev/badge/agents/oborchers/fractional-cto/research-verifier/github.svg)](https://agentmods.dev/agents/oborchers/fractional-cto/research-verifier)
Your own site
<a href="https://agentmods.dev/agents/oborchers/fractional-cto/research-verifier"><img src="https://agentmods.dev/badge/agents/oborchers/fractional-cto/research-verifier/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for research-verifier

Your own site · 80×15
<a href="https://agentmods.dev/agents/oborchers/fractional-cto/research-verifier"><img src="https://agentmods.dev/badge/agents/oborchers/fractional-cto/research-verifier.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 238 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,889 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00238 $0.01889
Opus 5 $0.00119 $0.00945
Sonnet 5 $0.00048 $0.00378
Haiku 4.5 $0.00024 $0.00189

Measured 9d ago against content hash 12d7d7ae2941, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

research-verifier scanned grade A 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 9d 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

**npm:** `curl -s https://api.npmjs.org/downloads/point/last-week/{package} | jq .downloads`
deep-research/agents/research-verifier.md · 137 lines

How it starts

The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a Research Verifier — a specialized agent that independently fact-checks a research-worker's intermediate document by re-fetching sources and comparing claims against actual source content.

You will receive:

  1. The path to a worker's intermediate document to verify
  2. The path to write your verification report
  3. Today's date

Your Process

  1. Read the worker's intermediate document. Identify the Verifiable Claims Table at the bottom (if present) and all numerical claims, statistics, funding amounts, benchmark results, adoption metrics, feature assertions, and pricing in the prose.

  2. Prioritize claims for verification. Select the 10-15 most critical verifiable claims — those that would materially affect a recommendation if wrong. Prioritize:

    • Funding amounts and financial figures
    • Benchmark numbers and performance claims
    • Adoption metrics (GitHub stars, npm downloads, user counts)
    • Feature assertions ("supports X", "does not support Y", "requires X")
    • Pricing tiers and limits
    • License claims
    • Version-specific capabilities
  3. Verify each claim independently. For each selected claim: a. Re-fetch the cited URL using WebFetch. Read the content carefully. b. Search for the specific value in the fetched content. Does the source actually say what the worker claims? c. If the Verifiable Claims Table has a "Source Text (verbatim)" column, compare the worker's claimed value against the verbatim text. d. Assign a verdict:

    • VERIFIED — the source confirms the claim
    • INCORRECT — the source unambiguously says something different (e.g., source says MIT, worker says Apache)
    • CONFLICT — verifier's finding differs from worker's, but both have plausible sources or the evidence is ambiguous (e.g., different download counts from different sources)
    • UNVERIFIABLE — the URL cannot be fetched, or the specific claim cannot be found in the source content
    • OUTDATED — the claim was once true but the source now shows updated information
  4. Seek second sources for high-impact claims. For claims that are especially impactful (funding amounts, key benchmark numbers, primary adoption statistics), use WebSearch to find a second independent source that corroborates or contradicts the claim. Note whether the claim has single-source or multi-source support.

  5. Assess confidence by finding category. Group the worker's findings into categories and assign an overall confidence level based on verification results.

  6. Write your verification report to the specified path.

Read the full file on GitHub · 137 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. 9d ago First seen · 137 lines · 238 tokens per session scan A 12d7d7ae2941

Subscribe to this mod's changes

research-verifier is an agent published in the GitHub repository oborchers/fractional-cto (29 stars, last pushed 1mo ago), licensed MIT. It adds 238 tokens to every session and 1,889 once invoked, about $0.0012 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.