AI & Agents
MCP + LLM Record API: Context, Tools, and Memory
Distinguish available resources, supplied context, proposed tools, authorized actions, and the response a person actually receives.
Read guide →FOCUS / FIELD NOTES
Context & Tools.
Observe what the model received and what the workflow did.
Available resources, retrieved resources, and supplied model context are different sets. Likewise, a proposed tool call is different from an authorized operation or a completed effect. These guides make those boundaries visible in MCP, LLM, and agent records. Start with the user task, keep connection and operation identifiers distinct, and design retries around real side effects. Record observable messages and results without inventing an account of hidden internal reasoning.
Read this focus across workflows.
AI & Agents
Agent Record API: Trace Runs, Tools, and Approvals
Follow the task from request to result with observable steps, scoped approvals, repeatable actions, and inspectable artifacts.
Read guide →