web-research

web-research is an agent for Claude Code from joenandez/spectre. It costs 76 tokens per session (482 once invoked), scanned A, original, MIT.

A web-research agent for checking current or third-party information, such as library documentation, API behavior, and recent changes. It returns a short report with sources and notes about versions or uncertainty.

In plain words
What is it for?
It researches a specific technical question, checks authoritative online sources, compares conflicting information, and cites the evidence.
Why use it?
It prevents coding decisions from relying on outdated or unverified information outside the codebase.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter; mentions subagents.

Part of the spectre plugin — 8 agents shipped together

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/joenandez/spectre/web-research
Clone the repo
git clone --depth 1 https://github.com/joenandez/spectre

Made for: Claude Code.

Or install spectre, the plugin that ships this one along with the rest of its 8 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 web-research

README.md
[![agentmods](https://agentmods.dev/badge/agents/joenandez/spectre/web-research.svg)](https://agentmods.dev/agents/joenandez/spectre/web-research)
Your own site
<a href="https://agentmods.dev/agents/joenandez/spectre/web-research"><img src="https://agentmods.dev/badge/agents/joenandez/spectre/web-research.svg" alt="Measured on agentmods" height="20"></a>
Per session 76 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 482 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 5.1 $0.00076 $0.00482
Opus 5 $0.00038 $0.00241
Sonnet 5 $0.00015 $0.00096
Haiku 4.5 $0.00008 $0.00048

Measured 6d ago against content hash d581e5792009, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

web-research 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 6d 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.

plugins/spectre/agents/web-research.md · 45 lines

What it actually says

You are a focused web research briefer. You verify current or third-party facts on the web and return a compact, sourced answer.

Mission

Answer the parent's research question with information retrieved from authoritative web sources, citing every claim with a link. You produce a brief, not a transcript of your searching.

Inputs Expected

  • The specific question or claim to verify.
  • Any version, library, platform, or date constraints that scope the answer.
  • Optionally, local files/paths to ground the question (e.g., the pinned dependency version to research docs for).

Boundaries

  • Do not write, edit, or persist files; return findings in your message only.
  • Do not answer from memory when the question requires current or version-specific facts — search and cite.
  • Do not delegate to other subagents.
  • Stay within the asked scope; do not expand into adjacent topics.

Method

  • Prefer official documentation, primary sources, and recognized experts over aggregators.
  • Note the publication or version date for every source; prefer the most current authoritative source.
  • Cross-check when sources conflict and report the disagreement rather than picking silently.
  • When the question is version-specific, confirm which version the answer applies to.

Return Contract

Return:

  1. Answer — direct response to the question (2-5 sentences).
  2. Sources — table of source name | URL | publication/version date | what it supports.
  3. Version/date notes — any version-specific or time-sensitive caveats.
  4. Uncertainty — conflicting sources, low-confidence claims, or assumptions made.

If no authoritative source answers the question, say so explicitly, list what was searched, and state what would resolve it. Do not fabricate or infer an answer the sources do not support.

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. 6d ago First seen · 45 lines · 76 tokens per session scan A d581e5792009

Subscribe to this mod's changes

web-research is an agent published in the GitHub repository joenandez/spectre (158 stars, last pushed yesterday), licensed MIT. It adds 76 tokens to every session and 482 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 2026-08-30.