Apps for AI agents

REST API

Everything you can do in the app, you can do over HTTP. All paths are relative to https://ounie.com and authenticate with a Bearer API key.

Authentication#

Send your API key as a Bearer token on every request. See the API overview for how to create one.

Authorization: Bearer ounie_live_…

List your brains (v1)#

GET /api/v1/brains — returns your own brains, each with its id, name, slug, emoji, visibility, and source count. Handy as a dropdown for automation tools.

curl -s https://ounie.com/api/v1/brains \
  -H "Authorization: Bearer ounie_live_…"

{
  "brains": [
    {
      "id": "…",
      "name": "Research",
      "slug": "research",
      "emoji": "🧠",
      "visibility": "private",
      "sourceCount": 42
    }
  ]
}

Add a source (v1)#

POST /api/v1/brains/{brainId}/sources — add a link or a text passage. Any body may carry an externalId: a repeat returns the existing source with created: false (a 200), so retries and re-syncs are safe. A text source rides synthesis by default; set mode: "note" to store it verbatim.

// link
{ "kind": "link", "url": "https://example.com/article", "externalId": "art-42" }

// text (synthesized, default)
{ "title": "Standup notes", "content": "…", "sourceUrl": "https://…", "externalId": "standup-2026-06-07" }

// note (verbatim, skips synthesis)
{ "title": "Idea", "content": "…markdown…", "mode": "note", "externalId": "idea-7" }

{
  "source": { "id": "…", "type": "text", "title": "Standup notes", "status": "pending", "externalId": "standup-2026-06-07" },
  "created": true
}

Ask a brain (v1)#

POST /api/v1/brains/{brainId}/ask — ask a grounded question. The answer's [[slug]] citations are rewritten to absolute markdown links, and citations carries those URLs. Asking is metered on your plan: an exhausted Free bank returns 402, an exceeded paid quota 429.

curl -s -X POST https://ounie.com/api/v1/brains/{brainId}/ask \
  -H "Authorization: Bearer ounie_live_…" \
  -H "Content-Type: application/json" \
  -d '{"question":"What did we decide about pricing?"}'

{
  "id": "…",
  "answer": "We set [Pricing](https://ounie.com/b/ada/ai/pricing) last week.",
  "citations": [{ "slug": "pricing", "title": "Pricing", "url": "https://ounie.com/b/ada/ai/pricing" }],
  "thin": false,
  "quota": null,
  "bank": 987
}

List brainsv1 path above#

GET /api/brains — returns the brains you can access, each with its id, name, and source count.

curl -s https://ounie.com/api/brains \
  -H "Authorization: Bearer ounie_live_…"

{
  "brains": [
    {
      "id": "…",
      "name": "Research",
      "slug": "research",
      "visibility": "private",
      "sourceCount": 42
    }
  ]
}

Ask a brainv1 path above#

POST /api/brains/{brainId}/ask — ask a grounded question and get an answer with citations. If the brain cannot answer, thin is true. Asking counts against your daily quota; an exceeded quota returns 429 with a quota object.

curl -s -X POST https://ounie.com/api/brains/{brainId}/ask \
  -H "Authorization: Bearer ounie_live_…" \
  -H "Content-Type: application/json" \
  -d '{"question":"What did we decide about pricing?"}'

{
  "answerMd": "…answer with [[slug]] citations…",
  "answerHtml": "<p>…</p>",
  "citations": [{ "slug": "pricing", "title": "Pricing", "documentId": "…" }],
  "retrievedSlugs": ["pricing", "…"],
  "thin": false
}

Ask a brain set (blended)#

POST /api/v1/brain-sets/{setId}/ask — ask a brain set (a primary brain blended with advisor brains) and get one answer that tags each citation by its source brain. Create a set in Brain Lab to get a set_… id. This endpoint requires API access (Pro and Team).

curl -s -X POST https://ounie.com/api/v1/brain-sets/{setId}/ask \
  -H "Authorization: Bearer ounie_live_…" \
  -H "Content-Type: application/json" \
  -d '{"question":"How should I word this?"}'

{
  "answer": "…answer with [[slug]] citations…",
  "citations": [
    { "brain": { "id": "…", "name": "Company", "kind": "primary" },
      "slug": "pricing", "title": "Pricing", "url": "https://ounie.com/b/…" }
  ],
  "thin": false,
  "advisorContributed": true
}

Retrieve context#

POST /api/brains/{brainId}/context — return the matching wiki pages without generating an answer, so your own model can reason over them. This does not count against your ask quota.

curl -s -X POST https://ounie.com/api/brains/{brainId}/context \
  -H "Authorization: Bearer ounie_live_…" \
  -H "Content-Type: application/json" \
  -d '{"query":"pricing decision"}'

{
  "contextText": "…full markdown of the retrieved pages…",
  "pages": [{ "slug": "pricing", "title": "Pricing" }],
  "thin": false
}

Add a source#

POST /api/brains/{brainId}/sources — add a link, a synthesized text passage, or a verbatim note. The body shape depends on type.

Link#

Ounie fetches the URL and synthesizes it into the wiki.

{ "type": "link", "url": "https://example.com/article" }

Text#

Inline content that is synthesized and made citable. externalId makes repeats idempotent; sourceUrl attaches an origin.

{
  "type": "text",
  "title": "Standup notes",
  "content": "…",
  "sourceUrl": "https://…",
  "externalId": "standup-2026-06-07"
}

Note#

Stored verbatim, skipping synthesis.

{ "type": "note", "title": "Idea", "body": "…markdown…" }

Each call returns the created source with a status that moves from pending to ready as it is processed in the background.

List sources#

GET /api/brains/{brainId}/sources — list every source in a brain with its type, title, and status.

Upload a file#

Files upload directly to storage in two steps. First request a presigned URL:

curl -s -X POST https://ounie.com/api/brains/{brainId}/sources/upload-url \
  -H "Authorization: Bearer ounie_live_…" \
  -H "Content-Type: application/json" \
  -d '{"filename":"report.pdf","contentType":"application/pdf","size":248000}'

Then PUT the file bytes to the returned uploadUrl (with a matching Content-Type). Ounie processes it into the wiki once the upload lands.

{ "sourceId": "…", "uploadUrl": "https://…", "key": "…" }