The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute local Python scripts (e.g., engagement-state.py, checkpoint-manager.py, output-publisher.py) for workflow orchestration, state persistence, and artifact publishing. These scripts are invoked from the local ${CLAUDE_PLUGIN_ROOT} directory.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from client feedback and external research, which could potentially contain malicious instructions aimed at influencing the agent's behavior.
Ingestion points: Processes 'Opinion' intake (hypotheses and evidence) and 'Client Validation' deliverables that include user-provided decisions.
Boundary markers: The skill does not explicitly define markers or delimiters to isolate untrusted data within its prompt logic.
Capability inventory: The skill has access to powerful tools including Bash, Write, Edit, and Task (for parallelization).
Sanitization: Employs a mandatory quality gate command (/digital-marketing-pro:check) and script-based manifest validation (engagement-state.py validate-part) to inspect outputs before finalization.

engagement-workflow