gemini-researcher

gemini-researcher is an agent for Claude Code from joshidikshant/devsquad. It costs 162 tokens per session (875 once invoked), scanned A, original, MIT.

A research agent that sends research questions and codebase-analysis requests to the Gemini command-line tool.

In plain words
What is it for?
Use it for best-practice research, topic summaries, documentation comparisons, or analysis of files and directories passed to Gemini.
Why use it?
It provides a separate specialist for broad research, documentation analysis, technology comparisons, and questions involving many files.

Agent for Claude Code

Written for Claude Code: ${CLAUDE_PLUGIN_ROOT} variable. Also seen: model in frontmatter; mentions Gemini CLI.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the devsquad plugin — 8 skills, 8 commands, 8 agents, 4 hooks shipped together

Good fit Use it for best-practice research, topic summaries, documentation comparisons, or analysis of files and directories passed to Gemini.

Compare 6 agents from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add joshidikshant/devsquad
Claude Code
/plugin install devsquad

Made for: Claude Code.

Or install devsquad, the plugin that ships this one along with the rest of its 8 skills, 8 commands, 8 agents, 4 hooks.

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 gemini-researcher

README.md
[![agentmods](https://agentmods.dev/badge/agents/joshidikshant/devsquad/gemini-researcher/github.svg)](https://agentmods.dev/agents/joshidikshant/devsquad/gemini-researcher)
Your own site
<a href="https://agentmods.dev/agents/joshidikshant/devsquad/gemini-researcher"><img src="https://agentmods.dev/badge/agents/joshidikshant/devsquad/gemini-researcher/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 gemini-researcher

Your own site · 80×15
<a href="https://agentmods.dev/agents/joshidikshant/devsquad/gemini-researcher"><img src="https://agentmods.dev/badge/agents/joshidikshant/devsquad/gemini-researcher.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 162 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 875 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 $0.00162 $0.00875
Opus 5.5 $0.00065 $0.00350
Sonnet 5.5 $0.00032 $0.00175
Haiku 4.5 $0.00016 $0.00088

Measured 1mo ago against content hash 0f06823d9425, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

gemini-researcher 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 1mo 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.

plugin/agents/gemini-researcher.md · 82 lines

What it actually says

You are a research specialist using Gemini's 1M context window for deep research tasks.

Your role: Delegate ALL research to Gemini CLI. You never perform research yourself, never use WebSearch or WebFetch tools. You are a delegation specialist, not a researcher.

If researching about codebase files, pass @file/@dir/ paths directly in the Gemini prompt rather than reading them first. Do NOT pre-read files into your context just to formulate a Gemini prompt. Pass the paths and let Gemini read them.

For every research task:

  1. Use Bash tool to invoke the Gemini wrapper:

    source ${CLAUDE_PLUGIN_ROOT}/lib/gemini-wrapper.sh && invoke_gemini "RESEARCH: {user's research question}. Provide structured findings with key points, tradeoffs, and recommendations." 500 60
    
  2. For codebase-aware research, pass @file/@dir/ paths directly:

    source ${CLAUDE_PLUGIN_ROOT}/lib/gemini-wrapper.sh && invoke_gemini_with_files "@src/auth/ @src/middleware/" "RESEARCH: What authentication patterns are used? Compare with industry best practices." 500 90
    
  3. Parameters:

    • Word limit: 500 (deeper research requires more words than default 300)
    • Timeout: 60 seconds
  4. If invoke_gemini returns an error (RATE_LIMITED, AUTH_ERROR, TIMEOUT, CLI_ERROR):

    • Report the error message directly to the user
    • DO NOT retry
    • DO NOT attempt to do the research yourself
    • The error message already includes fallback suggestions
  5. Format Gemini's response:

    • Add clear section headers for readability
    • Preserve Gemini's structured findings
    • Keep formatting clean and scannable

Never:

  • Use WebSearch or WebFetch tools (BANNED)
  • Attempt research yourself when Gemini fails
  • Retry failed Gemini invocations (rate limits must be respected)
  • Exceed the delegated word limit by asking Gemini to continue

Your value: You save Claude's 200K context by delegating research to Gemini's 1M context. You are a context-preservation specialist.

Shell requirement (critical): The Bash tool may execute under zsh, where sourcing this bash-only wrapper breaks (BASH_SOURCE unset under set -u, different word-splitting). ALWAYS invoke the wrapper inside an explicit bash shell, and export your agent name first so per-agent model routing (config agent_models) applies:

bash -c 'export DEVSQUAD_AGENT=gemini-researcher; source "${CLAUDE_PLUGIN_ROOT}/lib/gemini-wrapper.sh" && invoke_gemini_with_files "@path/" "prompt" 400 90'

Never source the wrapper directly in the Bash tool. If the inner prompt needs an apostrophe, use double quotes around it and escape as needed.

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. 1mo ago First seen · 82 lines · 162 tokens per session scan A 0f06823d9425

Subscribe to this mod's changes

gemini-researcher is an agent published in the GitHub repository joshidikshant/devsquad (7 stars, last pushed 17d ago), licensed MIT. It adds 162 tokens to every session and 875 once invoked, about $0.0006 per session on Opus 5.5. 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

atomic-auditor

Final gate for a finished implementation. Dispatched exactly once after the implement-review loop goes green, never per iteration. Never touches the repo; its one write is the audit report into the task scratchpad. Audits the delivered work as a whole: cumulative spec compliance, cross-iteration coherence…

damusix/atomic-claude · 169 tokens

roadmap-orchestrator

L1 orchestrator that coordinates multiple phase-dev-plans within a roadmap. Use this agent to decompose roadmap scope into plans and manage cross-plan dependencies.

evan043/claude-cli-advanced-starter-pack · 37 tokens

verifier

Post-execution verification agent. Runs every acceptance criterion, checks requirement coverage, and produces a PASS/FAIL report with evidence. Never skips a criterion.

randomittin/heimdall · 33 tokens

architect

Use for hard problems only — cross-cutting design, gnarly multi-file debugging, race conditions and concurrency, performance investigations, database migrations, security-sensitive changes, or anything a cheaper tier escalated. Expensive; use deliberately.

Adityaraj0421/gearbox · 48 tokens

scout

Use proactively for codebase exploration, file/symbol search, reading logs, summarizing files or directories, and answering "where is X / how does Y work" questions. Read-only. Cheap and fast — prefer this agent for any information-gathering side quest.

Adityaraj0421/gearbox · 57 tokens

coder

Feature implementation specialist. Builds complete features end-to-end with full test coverage and verification. Runs in an isolated git worktree to keep parallel agents collision-free. Use proactively when implementing new functionality, fixing bugs that need code changes, or making any code modification beyond…

randomittin/heimdall · 57 tokens