The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a Python orchestration script (saga_driver.py) using the Bash tool. It explicitly mandates the use of the --allow-skip-permissions flag, which is designed to suppress platform-level security prompts for file system modifications, enabling the agent to write files without user review.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data to generate development artifacts. 1. Ingestion points: The skill reads external SPEC component contracts (docs/06_SPEC/), free-text user prompts, and implementation plans (IPLAN-*). 2. Boundary markers: The instructions lack explicit delimiters or warnings to ignore potentially malicious instructions embedded within these source artifacts. 3. Capability inventory: The skill has extensive capabilities including shell command execution (Bash tool) and the ability to write or modify multiple project files (docs/07_TDD/*.yaml, docs/07_TDD/TDD-00_index.md). 4. Sanitization: No evidence is provided that the skill sanitizes or validates the content extracted from upstream documents before using it to drive the automation pipeline.

doc-tdd-autopilot