The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/verify_intake.py executes the command npm run build using the subprocess.run function. This is intended to verify that the project remains buildable after intake operations. The execution is performed with shell=False and uses an absolute path for the npm executable resolved via shutil.which, minimizing common command injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user prompts to generate structured product documentation and routing configurations, creating a potential surface for indirect prompt injection.
Ingestion points: Untrusted data enters the skill via the user_prompt and prompt_file parameters in scripts/load_intake.py and scripts/write_translation.py.
Boundary markers: No explicit delimiters or boundary markers are used when interpolating user-provided text into the markdown and JSON artifacts.
Capability inventory: The skill has the capability to write files to the project directory, copy files using shutil.copy2 in scripts/ingest_mockups.py, and execute build commands in scripts/verify_intake.py.
Sanitization: The skill does not perform sanitization or validation of the content of the user prompts before including them in the generated project documentation.

heyeddi-intake