tomevault-io Hook Authoring
Hook Points
Meerkat provides 8 hook points in the agent lifecycle:
- RunStarted — When the agent run begins
- PreLlmRequest — Before sending to the LLM
- PostLlmResponse — After receiving LLM response
- PreToolExecution — Before executing a tool call
- PostToolExecution — After tool execution completes
- TurnBoundary — At the boundary between turns
- RunCompleted — When the agent run completes successfully
- RunFailed — When the agent run fails
Execution Modes
- Foreground: Blocks execution, can modify/deny. Use for policy enforcement.
- Background: Runs concurrently, fire-and-forget. Use for logging, analytics.
- Observe: Receives events but cannot block. Use for monitoring.
Decision Semantics
Hooks return one of:
- Allow: Proceed normally
- Deny: Block the operation with a reason
- Rewrite: Modify the operation (e.g., rewrite tool arguments)
Patch Format
For PreToolExecution, hooks can rewrite tool arguments using JSON patches.
Failure Policy
- fail-closed: Hook failure blocks the operation (safer)
- fail-open: Hook failure allows the operation (more resilient)
Converted and distributed by TomeVault — claim your Tome and manage your conversions.