The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the repository (README, landing pages, package.json) to auto-draft marketing materials, which could be exploited to influence agent behavior through hidden instructions in those files.
Ingestion points: SKILL.md instructions for codebase scanning and scripts/context_validator.py for file reading.
Boundary markers: None specified in the instructions for separating external content from agent instructions.
Capability inventory: Local file system read/write access via the agent and the included Python script.
Sanitization: No specific filtering or escaping is applied to the content read from external files beyond regex pattern matching.
[COMMAND_EXECUTION]: The skill includes a local script scripts/context_validator.py intended to be executed by the agent to score marketing documents.
The script uses standard Python libraries (re, pathlib, json) to perform its analysis and does not initiate network connections, execute arbitrary shell commands, or perform privilege escalation.

marketing-context