The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core functionality involves ingesting data from external, untrusted sources.
Ingestion points: The agent is instructed to visit external review platforms (e.g., App Store, Google Play, Reddit, Hacker News) and copy verbatim text into replica/reviews.csv (SKILL.md).
Boundary markers: The instructions do not define clear delimiters or provide the agent with specific warnings to ignore instructions that might be embedded within the collected review text.
Capability inventory: The skill has the capability to write to multiple files (feedback.md, fixes.md, features.csv) and execute a local Python script (reviews.py) as part of its workflow.
Sanitization: The included reviews.py script extracts snippets of the raw review text using regular expressions but does not perform sanitization, filtering, or escaping before the data is re-presented to the agent in feedback.md for analysis.
[COMMAND_EXECUTION]: The skill instructions specify the execution of a local Python script, reviews.py, to process the collected review data. This script uses only Python standard libraries and performs local file operations to generate reports.

replica-entrepreneur