The Agent Skills Directory

[SAFE]: The skill is primarily composed of markdown documentation and templates designed to guide AI agents through a structured software development lifecycle. No malicious patterns were identified.
[COMMAND_EXECUTION]: The skill includes shell scripts (spec-quality-check.sh, spec-to-tasks.sh) that use standard utilities like grep, sort, and wc to validate local markdown files. These operations are limited to structural checks and do not pose a security risk.
[DYNAMIC_EXECUTION]: The test script scripts/test_business_policy.py uses Python's importlib to dynamically load the peer script business_policy.py for testing purposes. This is a standard practice for unit testing in Python and is restricted to local files provided within the skill.
[INDIRECT_PROMPT_INJECTION]: The skill defines a pipeline that processes specifications (SPEC.md) and task plans (TASK-PLAN.md). While this constitutes an ingestion surface for untrusted data, the methodology includes multiple 'Quality Gates' specifically designed to catch and remediate ambiguous or incorrect instructions before they reach implementation agents. The capability of the provided scripts is restricted to metadata and structure validation, preventing privilege escalation.

spec-driven-development