- Agent playbook — masterpiece UX for consuming agentsEnd-to-end flow for an assistant helping a donor. Markdown
- Brand voice — Better World (agents and humans)Tone for answers that quote us. Markdown
- Cause research hub (research-only)Research notes on causes we have not added. Not a ranking. Markdown
- Human decision desk (Wave CA)Decisions waiting on our human reviewer. Markdown
- Evaluator anchorsWhere GiveWell and HLI publish the figures we cite. Markdown
- FAQ for agents — common failure modesFailure modes and how to handle them. Markdown
- Honesty — what Better World never doesWhat we never do, as testable rules. Markdown
- How to query Better World (agents)Query patterns for the REST API. Markdown
- MCP / agent tool catalogEvery MCP tool with arguments and caveats. Markdown
- Exception review queue — Ho Hum (read-only view)The exception-review queue (read-only). Markdown
- Values clarification interview (agent script)A short interview to learn a donor's values and red lines. Markdown
- Vision: "Did it work?"One-page vision. Markdown
- Impact pagesNotes on modeled impact estimates. Markdown
Structured research library
Organization cards, intervention notes and per-unit leaderboards, each in Markdown with JSON-LD: organizations, interventions, lives saved, wellbeing (WELLBYs), income.