The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using the agent-validator CLI (e.g., agent-validator detect 2>&1 and agent-validator skip) to manage the validation gate lifecycle.
[PROMPT_INJECTION]: The instructions explicitly direct the agent to operate autonomously and bypass user confirmation ("Do not stop to ask permission, confirm next steps, or check in between tasks"). This design pattern is used to ensure continuous execution of the tech lead workflow.
[DATA_EXFILTRATION]: The skill exhibits an indirect prompt injection surface as it ingests and processes untrusted external data.
Ingestion points: Processes tasks, design documents, specifications, and proposals from the project environment.
Boundary markers: No explicit delimiters or boundary instructions are provided to the agent to segregate external content from core logic.
Capability inventory: Includes shell command execution via agent-validator and pull request management via codagent:finalize-pr.
Sanitization: No evidence of sanitization or validation of the external content before it is processed by the agent or subagents.

implement-change