The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/run_local_validator.py executes local Python validation tools using subprocess.call(). The commands are constructed using a list of arguments and a base path defined by the $UIUX_SKILL_SYSTEM_ROOT environment variable. While this allows the skill to perform its auditing functions, it involves executing shell-level commands based on the skill's instructions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core function involves reading and analyzing untrusted external data from a project's workspace, which could contain malicious instructions designed to subvert the audit process.
Ingestion points: The skill reads TASK-PLAN.md, progress-state.json, agent-progress-screen.html, and workflow-log.jsonl from the project directory (as identified in SKILL.md and references/decision-framework.md).
Boundary markers: Absent. The skill instructions do not specify the use of delimiters or 'ignore' instructions to separate untrusted artifact content from the agent's control logic.
Capability inventory: The skill has the capability to execute local subprocesses (via scripts/run_local_validator.py) and write validation reports or log entries to the project folder.
Sanitization: Absent. Although SKILL.md mentions producing a 'sanitized report,' there is no technical evidence of escaping or validation logic to prevent content from project files from influencing the final output or downstream tools.

workflow-compliance-supervisor