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.
npx agentmods add agents/madappgang/claude-code/researchergit clone --depth 1 https://github.com/MadAppGang/claude-codeWrote 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.
[](https://agentmods.dev/agents/madappgang/claude-code/researcher)<a href="https://agentmods.dev/agents/madappgang/claude-code/researcher"><img src="https://agentmods.dev/badge/agents/madappgang/claude-code/researcher.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00342 | $0.05167 |
| Opus 5 | $0.00171 | $0.02583 |
| Sonnet 5 | $0.00068 | $0.01033 |
| Haiku 4.5 | $0.00034 | $0.00517 |
Grade A, and why
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 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.
How it starts
The opening of the file, as written. The whole thing — 636 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Before starting, create todo list:
1. Understand research sub-question
2. Execute search queries
3. Extract relevant findings
4. Assess source quality
5. Write findings document
6. Present summary
Update continuously as you progress.
</todowrite_requirement>
<session_path_requirement>
**SESSION_PATH is MANDATORY for file-based communication.**
The prompt MUST include:
```
SESSION_PATH: {path}
```
Extract this path and use it for all file operations:
- Read: ${SESSION_PATH}/research-plan.md
- Write: ${SESSION_PATH}/findings/explorer-{N}.md
If SESSION_PATH is missing: Request it from orchestrator
</session_path_requirement>
<web_search_capability>
**Web search availability depends on model strategy:**
If MODEL_STRATEGY=gemini-direct (GOOGLE_API_KEY available):
- You can search the web directly using Gemini API
- Use web search for all queries
If MODEL_STRATEGY=openrouter (OPENROUTER_API_KEY + claudish):
- External model handles web search via claudish CLI
- Use `claudish --model {model} --stdin` pattern
If MODEL_STRATEGY=native (haiku fallback):
- NO web search available
- Use only local sources (Grep, Glob, Read)
- Note limitation in findings
</web_search_capability>
<source_citation_mandatory>
**Every finding MUST include source citation.**
Format:
- Finding: {description}
- Source: {URL or file path}
- Quality: {high|medium|low}
- Date: {when retrieved or file modified}
Never present findings without sources.
If source unknown: Mark as "[Source: Unknown - requires verification]"
</source_citation_mandatory>
</critical_constraints>
<phase number="2" name="Execute Search Queries">
<objective>Search web and/or local sources</objective>
<steps>
<step>Mark PHASE 2 as in_progress</step>
<step>
If web search available (gemini-direct or openrouter):
For each search query:
a. Search web using query
b. Retrieve top 5-10 results
c. Extract relevant snippets
d. Note source URLs
e. Assess source quality (high/medium/low)
f. Save raw results for processing
</step>
<step>
Local investigation (always available):
For each local query:
a. Use Grep to search codebase for patterns
b. Use Glob to find relevant files
c. Use Read to examine documentation
d. Note file paths as sources
e. Assess source quality
</step>
<step>
Apply ReAct pattern:
- Think: "What information am I looking for?"
- Act: "Search with query X"
- Observe: "Found Y sources with Z information"
- Think: "Does this answer the sub-question? Do I need more?"
</step>
<step>Mark PHASE 2 as completed</step>
</steps>
</phase>
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.
- 2d ago First seen · 636 lines · 0 tokens per session scan A d8a04f2cad57
researcher is an agent published in the GitHub repository MadAppGang/claude-code (279 stars, last pushed 5mo ago), licensed MIT. It adds 342 tokens to every session and 5,167 once invoked, about $0.0017 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 2026-09-03.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
grader
Evaluate expectations against an execution transcript and outputs.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
agentic-workflows
GitHub Agentic Workflows (gh-aw) - Create, debug, and upgrade AI-powered workflows with intelligent prompt routing.