Open research toolkit for AI agents

Turn scattered evidence
into buyer context

Six research skills turn customer conversations, CRM notes, product docs, market sources, and competitor evidence into durable buyer intelligence your team and agents can use.

View on GitHub

Claude Code · Codex · Cursor · OpenClaw · any Agent Skills runtime

How it works

Evidence in. Buyer intelligence out.

Bring the sources you already have, add current market evidence, and build one shared context your people and agents can work from. Your website can help, but it is only one input.

01 / Evidence

Start with what buyers actually said

Bring interviews, calls, CRM notes, reviews, win-loss records, product docs, and the public sources that matter.

/customer-research ./interviews

02 / Research

Run focused research agents

Study your company, customers, market, and the alternatives buyers compare. Every conclusion stays connected to evidence.

/full-research

03 / Context

Create a shared decision document

Synthesize ICP, jobs, triggers, pains, objections, decision criteria, alternatives, proof, buyer language, and open questions.

./buyer-context.md

The skills

Four research agents. One synthesis layer.

Run one skill against the evidence you have, or let full-research coordinate the complete workflow and refresh the canonical buyer-context.md.

buyer-context

Synthesis

Build or refresh the canonical buyer intelligence document from the evidence and research already in scope.

full-research

Orchestrator

Coordinate all four research workstreams, reconcile conflicts, and synthesize a complete buyer context.

company-research

Company

Understand the offer, capabilities, delivery model, customers, constraints, and proof the business can support.

customer-research

Customers

Analyze interviews, calls, CRM notes, reviews, and support evidence without inventing quotes or prevalence.

market-research

Market

Research the category, demand drivers, buyer environment, constraints, language, and material market shifts.

competitor-research

Alternatives

Research vendors, internal builds, manual work, incumbents, and doing nothing through the buyer's criteria.

Give every agent the same buyer context.