Currently in open development testing — expect changes, quirks, and occasional downtime. Your data stays yours.

Live · open development

Give your agent a memory.

Thaladex remembers what your agents learn and hands it back next session. ChatGPT, Grok, Claude Code, Hermes, or any agent that can read a link. Nothing to install.

How it works

1 · Sign up

Email and password. Your first memory is created for you.

2 · Copy one line

Pick your agent, copy the connect link, paste it into the agent.

3 · Your agent remembers

It writes its first memory and reads it back. You watch it happen.

What it's for

Working memory

Decisions, preferences, what got finished — remembered across sessions, per agent.

Knowledge base

Documents and reference that shouldn't fade. Your agents read them when they need them.

For agents

curl -sS "$THALADEX_URL/v1/whoami" -H "Authorization: Bearer $THALADEX_TOKEN" curl -sS -X POST "$THALADEX_URL/v1/layers/$LAYER/remember" \ -H "Authorization: Bearer $THALADEX_TOKEN" -H 'Content-Type: application/json' \ -d '{"text":"Agent connected successfully."}' curl -sS -X POST "$THALADEX_URL/v1/layers/$LAYER/recall" \ -H "Authorization: Bearer $THALADEX_TOKEN" -H 'Content-Type: application/json' \ -d '{"query":"agent connected","top_k":5}'
MCP toolREST map
thaladex_whoamiGET /v1/whoami
thaladex_rememberPOST …/remember
thaladex_recallPOST …/recall
thaladex_feedbackPOST /v1/feedback

Full reference: /docs/api · OpenAPI: /openapi.json · MCP: /mcp