The Agent Skills Directory

[PROMPT_INJECTION]: The SKILL.md file contains instructions stating that the workflow "OVERRIDES the agent's default development behavior" once activated. This is a workflow steering mechanism intended to enforce the AI-DLC methodology.
[COMMAND_EXECUTION]: The skill includes a maintenance script scripts/check-upstream.sh that executes shell commands such as git, diff, tar, and mktemp. These are used by the developer or agent to verify and sync reference files with the upstream repository.
[EXTERNAL_DOWNLOADS]: The skill contains a script for fetching data from the official AWS Labs GitHub repository (https://github.com/awslabs/aidlc-workflows.git). This is a well-known and trusted source.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external user-provided documents (Vision and Technical Environment) to drive the inception and construction phases. This creates a surface where embedded instructions in those files could influence the agent's behavior.
Ingestion points: User-supplied vision and technical environment documents (templates in references/inputs/).
Boundary markers: The skill employs mandatory human approval gates ("Wait for Explicit Approval") and maintains an append-only audit.md log to track all user interactions and decisions.
Capability inventory: The agent has capabilities to write files to the workspace and execute shell scripts.
Sanitization: No specific content sanitization or XML tagging is described for the ingestion of these external documents.

aidlc-master