literature-searcher

literature-searcher is an agent for Claude Code from yqi96/warranted. It costs 29 tokens per session (1,029 once invoked), scanned A, original, MIT.

A literature-search and screening assistant for surveys. A literature survey is a structured review of published research, and screening means deciding which search results meet stated criteria.

In plain words
What is it for?
Use it to search research venues, record every candidate, apply supplied inclusion rules, assign approved exclusion codes, or screen a selected set of pending CSV rows.
Why use it?
It keeps search results, review decisions, and exclusion reasons in a consistent CSV file. It can also review only a specified range of existing pending records without adding new searches.

Agent for Claude Code

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

Part of the warranted plugin — 7 skills, 8 agents, 1 hook, 1 MCP server shipped together

Good fit Use it to search research venues, record every candidate, apply supplied inclusion rules, assign approved exclusion codes, or screen a selected set of pending CSV rows.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/yqi96/warranted/literature-searcher
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/yqi96/warranted

Made for: Claude Code.

Or install warranted, the plugin that ships this one along with the rest of its 7 skills, 8 agents, 1 hook, 1 MCP server.

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 literature-searcher

README.md
[![agentmods](https://agentmods.dev/badge/agents/yqi96/warranted/literature-searcher/github.svg)](https://agentmods.dev/agents/yqi96/warranted/literature-searcher)
Your own site
<a href="https://agentmods.dev/agents/yqi96/warranted/literature-searcher"><img src="https://agentmods.dev/badge/agents/yqi96/warranted/literature-searcher/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 literature-searcher

Your own site · 80×15
<a href="https://agentmods.dev/agents/yqi96/warranted/literature-searcher"><img src="https://agentmods.dev/badge/agents/yqi96/warranted/literature-searcher.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 29 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,029 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 $0.00029 $0.01029
Opus 5 $0.00015 $0.00515
Sonnet 5 $0.00006 $0.00206
Haiku 4.5 $0.00003 $0.00103

Measured 9d ago against content hash 8dddfcf960f6, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

literature-searcher 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 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.

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.

agents/literature-searcher.md · 49 lines

How it starts

The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Mode A — Search and screen (runs exactly once per survey)

Given a set of queries, venues, and criteria, run the queries and write every candidate into a CSV at the specified path. The CSV header is: title | query | venue | decision.

  1. Write every candidate row with the literal decision=PENDING first. Flush the file.
  2. Then rewrite decisions in place, applying the criteria supplied verbatim in the brief.
  3. Each exclusion must carry a code from the closed list supplied in the brief. No code may be coined.
  4. Return only: the file path, the total row count, the decided row count, and the reason-code distribution. Never return the candidate list itself.

Mode B — Screen only

Given a file path, an explicit row range, the criteria verbatim, and the closed reason-code list:

  1. Run no queries and add no rows.
  2. Decide only the PENDING rows inside the given range.
  3. Touch nothing else.

Contract clauses (all modes)

Each clause is annotated with the form its constraint takes. One of them has no physical form at all, and unless that is marked you will assume the allow-list already covers it.

# Clause Constraint form
1 CSV header is title | query | venue | decision, verbatim prompt
2 Mode A: write every candidate row with decision=PENDING first, flush, then rewrite decisions in place. The literal PENDING — not pending, TBD, or blank; cross-dispatch resumption runs on grep PENDING, and a case mismatch reads as "screening finished" prompt
3 Screening is performed against the criteria supplied verbatim in the brief. One code per exclusion drawn from the closed list in the brief. No criterion may be invented, relaxed, or narrowed; no code may be coined prompt
4 Return only the path, the total row count, the decided row count, and the reason-code distribution. Never the candidate list — returning it would put back the context cost this whole phase exists to avoid prompt
5 Never change the total row count. Never modify a row whose decision is not PENDING. In Mode B, run no queries and add no rows prompt, no physical form
6 Allow-list is search/read plus Write: WebSearch, WebFetch, Read, Grep, Write. Zero graph-write tools — you cannot register tags, create Statements, or touch the graph in any way physical: allow-list
7 The write root and the root that attachment validation resolves against are the same value: reviewCwd(config) = dirname(dirname(config.dbPath)). Write the CSV at the path the brief gives you, which sits next to the graph database prompt

Read the full file on GitHub · 49 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 · 49 lines · 29 tokens per session scan A 8dddfcf960f6

Subscribe to this mod's changes

literature-searcher is an agent published in the GitHub repository yqi96/warranted (2 stars, last pushed 7d ago), licensed MIT. It adds 29 tokens to every session and 1,029 once invoked, about $0.0001 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.