Hooks

Your own shell commands at fixed points: to check, block or add context.

Event	When it runs
SessionStart	a conversation starts
UserPromptSubmit	you send a message, before PHRONEN reads it
PreToolUse	before a tool runs — can block it
PostToolUse	after a tool ran
Stop	PHRONEN is about to finish its turn

.phronen/settings.jsonjson

{
  "hooks": {
    "PreToolUse": [
      { "matcher": "Bash", "hooks": [{ "type": "command", "command": "./scripts/check-command.sh", "timeout": 30 }] }
    ],
    "PostToolUse": [
      { "matcher": "Edit|Write", "hooks": [{ "type": "command", "command": "npx prettier --write \"$(jq -r .tool_input.file_path)\"" }] }
    ]
  }
}

The hook gets a JSON object on stdin: session_id, cwd, event, and — as they apply — tool_name, tool_input, tool_result, prompt.
Exit code 2 blocks the action; what the hook wrote to stderr is shown to PHRONEN as the reason.
Printing {"decision": "block", "reason": "…"} on stdout does the same; {"additionalContext": "…"} adds context instead.
Any other exit lets the action proceed. matcher is a tool name or glob; several are separated by |. timeout is in seconds (default 60).

Was this page helpful?

PreviousMCP servers Next Memory (PHRONEN.md)