Quickstart
From API key to first successful request in the shortest path.
Don't rebuild assessments, parsing, and access controls. Start with the Context API so agents work from a real person, not a prompt that forgets.
Agents don't need to reconstruct the user on every request. Context is already there.
Context API first. Stable errors. Clear field definitions. No guesswork.
Agents filter, judge, and move work forward from the same long-term context. Tokens multiply.
From API key to first successful request in the shortest path.
Pull aggregated context first. Expand to task endpoints when needed.
Stable contracts, predictable errors, shared definitions for real integration.
Let agents read long-term context first, then decide if deeper capabilities are needed.
Create an API Key
Call Context API
Expand by Task
Generate a bearer key and define access boundaries first.
Make profile, preferences, and signals the default context layer.
Add graph, alerts, and prediction only when the workflow needs them.
Answer from a real profile instead of relearning the user each time.
Plug into matching, ranking, and filtering workflows with user permission.
Embed long-term context into notifications, coaching, and recommendations.
Use ProfileClaw as the memory layer instead of scattering identity across prompts.
Start with aggregated context before task endpoints
Depend on stable types, not free-form error strings
Keep OpenAPI and docs on the same contract source
Design for agent continuity, not single requests
Start with docs, review pricing, and decide if you're building for personal workflows or product integration.