The Agent Skills Directory

[SAFE]: No malicious patterns were detected. The skill is instructional in nature and focuses on establishing rigorous workflows, auditability, and task verification.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by design, as it is intended to process and act upon arbitrary user tasks that could potentially contain embedded instructions. However, the risk is mitigated by the skill's explicit focus on 'rigor', 'verification against predicates', and 'audit trails'.
Ingestion points: The skill ingests user-provided tasks and references a local file (../poteto-mode/references/t3-execution.md) to define the execution environment.
Boundary markers: While no specific text delimiters are mentioned, the skill uses the disable-model-invocation: true flag in its frontmatter, which serves as a platform-level boundary restricting the model's ability to call tools directly.
Capability inventory: The instructions describe managing task delegation (delegate_task, t3_thread_launch) and creating audit logs via the show-me-your-work skill.
Sanitization: The skill lacks explicit sanitization instructions but mandates verification of all outputs against artifact-based predicates rather than self-reports.

figure-it-out