The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data (requirement documents) into a PowerPoint presentation. This is a standard attack surface where untrusted data could be used to influence the output content. However, the skill treats all input as text to be rendered in a binary document format (.pptx) and does not grant elevated privileges based on that content.
Ingestion points: The spec dictionary passed to requirement_review_deck in scripts/deckbuilder.py.
Boundary markers: The skill relies on the structure of the PowerPoint file format; there are no explicit NLP boundary markers for internal text content.
Capability inventory: The skill possesses the capability to write files to the local system using the save method in scripts/deckbuilder.py.
Sanitization: The library performs basic string sanitization and number parsing in the _num and _rr_lines methods in scripts/deckbuilder.py.
[COMMAND_EXECUTION]: The documentation in SKILL.md and README.md correctly instructs the agent to install the python-pptx library and run local validation scripts. These commands are standard for the skill's intended purpose of document generation and verification.
[DYNAMIC_EXECUTION]: The example scripts use sys.path.insert to load the local deckbuilder.py module. This is a necessary and standard pattern for utilizing local scripts within a skill's directory structure without requiring a system-wide installation.
[SAFE]: No evidence of data exfiltration, credential harvesting, obfuscation, or persistence mechanisms was found in the provided code or instructions.

ohos-req-review-ppt-gen