Put an agent on your website
Use the website embedding guide at /embed-your-agent for guided owner setup, complete copyable snippets, published-site checks and platform paste locations. Choose an agent you own; visitors use its real Mind and configured capabilities.
Make the boundary explicit
Your application owns its user experience and workflow. The model produces its responses. thinqOS can maintain a Mind and expose capabilities through MCP and API interfaces. Your application decides when those capabilities are used.
Saving context does not make it appear in every response automatically. Capture, retrieval, identity and prompt assembly must all be connected. Test each boundary with one fact you can trace.
Connect and verify one fact
The commands below are the shortest real loop: store one fact, confirm it landed, then recall it. Run them in order against your own account. Every response quoted here came from running them against production on September 8, 2026, not from a specification.
Start by creating an API key in Settings, under API keys. There is no command-line path to mint one, so this first step needs a browser. Export it as THINQOS_API_KEY and the rest is curl.
Two things will surprise you, so they are worth stating before you hit them. Writing is REST and reading is not: an API key is deliberately refused on identity-resolving REST endpoints, which returns "API key authentication is not supported on identity-resolving REST endpoints; use a Clerk session token". Recall for a key-holder runs over MCP instead, which is the third command below. And extraction is asynchronous, so a fact you just stored reports extraction_status "pending" and will not appear in recall until that finishes. A successful write is not yet a learned belief.
Keep durable instructions separate from inferred lessons. Treat retrieved material as evidence with a source, not a higher-priority instruction. A document or outside message must not acquire authority just because it was remembered.
Use both kinds of retrieval
A question about a general preference calls for different evidence than a request for an exact invoice number. Semantic search finds related meaning. Lexical search finds words and identifiers. A useful connection preserves access to original records as well as selected Mind context.
Measure misses, contradictions, latency and total cost on your own tasks. A retrieval function returning successfully is not evidence that it found the needed fact.
Keep an exit path
Mind exports support a signed archive and documented open formats including JSON-LD, YAML-LD and a Markdown vault. Complete export and verified deletion are still being built, and we publish the limits as we test them. Verify which layers and fields your use requires.
Portable data is not a portable running system. Tool connections, application behavior and the cognition implementation require their own migration plan. Start with the setup guide and inspect the export before depending on it.
Commands
Store one fact
curl -X POST https://app.thinqos.com/api/imprints/native-memory \
-H "X-Ingest-Key: $THINQOS_API_KEY" \
-H "Content-Type: application/json" \
-d '{"source":"my-app","source_external_id":"note-1",
"label":"Deployment rule",
"text":"Staging deploys are gated on a manual approval step."}'
# 201 {"imprint":{"id":"...","extraction_status":"pending",...},
# "queued_for_extraction":true}Confirm it landed, and watch extraction
curl https://app.thinqos.com/api/imprints \
-H "X-Ingest-Key: $THINQOS_API_KEY"
# 200 - find your id. extraction_status goes pending -> done.
# belief_count stays 0 until extraction finishes.Recall it (MCP, not REST)
curl -X POST https://app.thinqos.com/mcp \
-H "Authorization: Bearer $THINQOS_API_KEY" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
"params":{"name":"recall",
"arguments":{"query":"deployment approval"}}}'
# 200 - Mind context for the query. Recall reflects your fact
# only after its extraction_status reaches done.