requests
Fetch public research sources and API responses when approved.
Agent Kits
Research · medium tier
AI Research Analyst Agent Kit
Produce cited research briefs, competitor scans, source tables, and watchlist updates.
Use this kit for source-grounded research workflows that combine web search, hosted browser sessions, notes, and recurring reports.
What it does
Useful before any auth, better with the right tools
The kit starts by installing editable operating files into the agent workspace. It can work from pasted context immediately, then ask for tool access only when the current job benefits from it.
- Citation and source-quality rules
- Competitor watch and research brief runbooks
- Web search, hosted browser, and recurring report recommendations
Starter prompts
Research this topic and produce a cited brief with a source table.
Compare these competitors by positioning, pricing, features, and proof.
Create a recurring watchlist plan for this market.
Toolbelt
Not a naked prompt. A prepared working environment.
Research libraries for web-page extraction, PDFs, tables, citations, and clean Markdown evidence packs.
beautifulsoup4
Extract structured text from HTML pages.
markdownify
Convert web pages into citation-friendly Markdown notes.
pypdf
Extract text from PDFs, papers, filings, and uploaded source documents.
pandas
Compare evidence tables, source lists, and research matrices.
Recommended connections
Proactive, permission-aware setup
OpenClaw Setup records what the agent should ask for and why. Native OAuth and API-key installers can attach to this surface later without changing the kit contract.
Web search
Search access lets the agent gather public sources and compare evidence.
Cited public-source research.Safety model
External actions stay behind approval
Approval required
- using logged-in websites
- publishing reports
- messaging external parties
Blocked claims
- guarantees complete coverage
- can access paywalled or private sources without user access