The Agent Skills Directory

[PROMPT_INJECTION]: The skill acts as a coordinator that processes outputs from multiple specialized sub-agents, creating a surface for indirect prompt injection.
Ingestion points: The coordinator receives architectural plans, RTL modules, and verification reports from agents such as soc-architect and soc-rtl-designer (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters used to isolate sub-agent outputs or prevent the coordinator from executing instructions potentially embedded in sub-agent data.
Capability inventory: The agent has the ability to execute local Python scripts (check_loop_state.py, update_state.py) and interact with EDA tools via MCP (SKILL.md).
Sanitization: No explicit sanitization or validation of data received from sub-agents is performed before it influences the pipeline state or downstream agent dispatches.
[COMMAND_EXECUTION]: The skill is configured to execute local project scripts and specialized hardware design tools.
Evidence: The instructions call for running .agents/scripts/check_loop_state.py and update_state.py to verify pipeline progress (SKILL.md).
Evidence: It interacts with MCP tools like soc-build.soc_sim and soc-openroad for synthesis and physical design tasks. These represent standard, intended development operations within the SoC workflow.

soc-pipeline