Quickstart
The four calls that make any AI useful on its hardest questions. Every example uses an identity API key (atamaia_…) — mint one under Admin → API Keys.
1. Hydrate — one GET returns the full working context
curl -s "https://api.atamaia.ai/api/hydrate?aiName=ash&projectId=1&preset=lean" \
-H "Authorization: Bearer $ATAMAIA_API_KEY"The envelope's data is the HydrationContext: identity, memories, active projects, tasks, standing rules, hints, and the previous session's handoff.
2. Ask any model — the context is the prompt
// ctx from step 1 feeds the system prompt; nothing else to assemble.
const reply = await anthropic.messages.create({
model: "claude-opus-4-6",
system: ctx.systemPrompt,
messages: userMessages,
});3. Write back what matters
client.create_memory(
1,
title="Deploy target is blue-green",
content="Prod deployments roll through staging first.",
type="Instruction",
provenance="Asserted",
importance=7,
)The same story in C#
var ctx = await client.HydrateAsync(aiName: "ash", projectId: 1, generateSystemPrompt: true);
var reply = await anthropicClient.CreateMessageAsync(new() {
Model = "claude-opus-4-6",
System = ctx.SystemPrompt,
Messages = userMessages,
});
await client.CreateMemoryAsync(1, new CreateMemoryRequest {
Title = "Deploy target is blue-green",
Content = "Prod deployments roll through staging first.",
Type = "Instruction",
Provenance = "Asserted",
Importance = 7,
});Prefer MCP? The same key works over Streamable HTTP at https://api.atamaia.ai/mcp — see MCP Integration.