Repositories / OpenClaw
Focused tour of agent execution, memory search, and tool orchestration; not a whole-project map.
Executes one embedded-agent attempt, preparing tools, session state, prompt context, and the guarded model stream.
Stored evidence from this immutable analysis snapshot
return await runEmbeddedAttemptExecutionPhase({attempt: params,...(activeContextEngine ? { activeContextEngine } : {}),agentDir,isRawModelRun,resolveActiveContextEnginePluginId,runAbortController,externalAbortController,
abortState, prepared: { bootstrap: preparedBootstrap, bundleTools: preparedBundleTools, sessionRuntime: preparedSessionRuntime, systemPrompt: preparedSystemPrompt, toolBase: preparedToolBase, toolCatalog: preparedToolCatalog, },