The Agent Skills Directory

[SAFE]: The skill defines a clear, restricted scope of operation using a small set of file manipulation tools (Read, Write, Grep, Glob) and does not request network permissions.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it ingests untrusted data from multiple external files (requirements, technology selection, and technical review documents).
Ingestion points: Reads requirements-{id}.md, technology-selection-{id}.md, and technical-review-{id}.md in the current iteration directory.
Boundary markers: Absent; the instructions do not specify delimiters to separate the content of these files from the system instructions.
Capability inventory: The skill has access to Read, Write, Grep, Glob, and the history-manager tool. It lacks network access or shell execution capabilities.
Sanitization: None; it treats the content of the read files as direct input for the design process.
Despite this surface, the risk is minimal as the skill's output is restricted to generating markdown documentation and does not trigger high-privilege operations.

program-design