The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The code example for the calculate tool in SKILL.md uses the eval() function. This is a dangerous pattern that allows arbitrary Python code execution if the input expression, which is generated by the agent in response to user input, is maliciously crafted.
[EXTERNAL_DOWNLOADS]: The skill uses the requests library to fetch data from well-known external services, including the GitHub API (api.github.com) and PyPI (pypi.org). While these are trusted services, the practice of fetching external content is documented as an external data dependency.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes content from external sources and feeds it into the reasoning loop without sanitization.
Ingestion points: get_github_issues, search_pypi, and get_github_repo tools in examples.md.
Boundary markers: No clear delimiters or instructions to ignore embedded commands are present in the agent prompts or tool outputs.
Capability inventory: The agent can make network requests via requests and potentially execute arbitrary code if the eval-based calculate tool is implemented.
Sanitization: There is no evidence of filtering, escaping, or validation of the data retrieved from external APIs before it is used in the prompt.

dspy-react