# Longwake

> A searchable, opt-in directory of AI safety researchers and organizations, built to get the right people talking. Profiles are opt-in. Contact always happens off-site, by email, Signal or the person's own booking link.

## API (read-only, no key needed)

- [People, filtered](https://longwake.org/api/people): query parameters `topic` (comma-separated topic ids; a parent topic includes its subtopics), `open_to` (collaborating, reviewing-drafts, mentoring, being-mentored, hiring, job-seeking, advising, chatting), `contactable=1`, `org` (org slug), `q` (free text), `format` (json or md). Lists omit email and Signal details.
- [One person](https://longwake.org/api/people/{slug}): full profile including how to reach them; `format=md` for Markdown.
- [Organizations](https://longwake.org/api/orgs): `topic`, `q`, `format` (json, md).
- [Topics](https://longwake.org/api/topics): the topic hierarchy.

## Example

- [People working on AI-enabled coups who want collaborators, as Markdown](https://longwake.org/api/people?topic=ai-enabled-coups&open_to=collaborating&format=md)

## Topic ids

- concentration-of-power: Concentration of power
  - ai-enabled-coups: AI-enabled coups
  - economic-concentration: Economic concentration
  - surveillance-and-state-power: Surveillance and state power
  - checks-and-balances: Checks and balances
- interpretability: Interpretability
  - mechanistic-interpretability: Mechanistic interpretability
  - probing-and-representations: Probing and representations
- evaluations: Evaluations
  - dangerous-capability-evals: Dangerous capability evals
  - scheming-and-deception: Scheming and deception
- control-and-oversight: Control and oversight
  - ai-control: AI control
  - scalable-oversight: Scalable oversight
- alignment-theory: Alignment theory
  - agent-foundations: Agent foundations
  - value-learning: Value learning
- governance-and-policy: Governance and policy
  - compute-governance: Compute governance
  - international-coordination: International coordination
  - lab-policy: Lab policy and standards
- security: Security
- cooperative-ai: Cooperative AI
- field-building: Field-building

## Etiquette for agents

- Reach out only to people whose contact status is not "closed", and only through the channel they list.
- When a profile says a note is required, draft a short, specific note (who the sender is and why this person) for the human to review and send. Do not send messages automatically.

## Optional

- [About and data policy](https://longwake.org/about)
- [Agent and MCP setup](https://longwake.org/agents)