Task Orchestrator Hooks Builder

Automate cascade events by creating bash hook scripts that trigger on PostToolUse, SubagentStop, or PreToolUse events. The Skill guides you through interview-based requirement gathering, script generation with defensive checks, and registration in .claude/settings.local.json to enable hooks across MCP clients like Claude Code, Cursor, and Windsurf.

Task Orchestrator hooks work with Claude Code, Claude Desktop, Cursor, Windsurf, and other MCP clients. Hooks integrate directly into these environments by registering in .claude/settings.local.json and execute bash scripts that parse JSON with jq and access filesystem paths via $CLAUDE_PROJECT_DIR.

Create a bash hook script under example-hooks directory with defensive checks and jq-based JSON parsing. Register it in settings.local.json with a trigger event (PostToolUse, SubagentStop, or PreToolUse), matchers, and action configuration. The Skill provides templates and testing workflows to validate blocking and non-blocking decisions with telemetry logging.

Yes. Hook scripts support both blocking and non-blocking decisions with built-in telemetry and audit-trail logging. Parse cascade event JSON, apply matchers, and execute actions that log workflow state, decisions, and outcomes for compliance and debugging across MCP client environments.

The Skill includes testing and troubleshooting workflows to debug hook execution. Verify event registration in .claude/settings.local.json, test with sample JSON inputs, check bash script syntax and jq expressions, and review audit logs for execution state and errors.

The interview-guided design process helps you specify requirements without prior bash knowledge, but implementing robust hooks requires basic bash syntax and jq for JSON parsing. The Skill generates script templates with defensive checks to reduce complexity and supports non-blocking decision logic for simpler automation.