The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill's primary function is to generate new Python tool classes and JSON configurations at runtime based on discovered API documentation. These generated files are then executed locally during the validation phase using the 'python scripts/test_new_tools.py' command. This creates a risk where malicious or poorly formed content from external APIs could result in the execution of unsafe code.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from web search results and external API documentation (Phase 1.3) to define the behavior and schemas of new tools. There is a risk that malicious instructions embedded in the documentation of a discovered API could influence the agent during the tool creation process.
Ingestion points: External web search results and API documentation processed in Phase 1.
Boundary markers: The instructions do not specify any delimiters or boundary markers to isolate external documentation content from the agent's core instructions.
Capability inventory: The skill possesses capabilities for file system writes (tool creation), network operations (API discovery), and shell command execution (validation and Git operations).
Sanitization: There is no evidence of sanitization or safety-filtering for the content extracted from external documentation before it is used to populate Python templates.
[COMMAND_EXECUTION]: The workflow relies on executing shell commands for testing (python scripts/test_new_tools.py) and repository management (git, gh). While these are standard developer actions, their automation in a pipeline driven by external input increases the attack surface.

devtu-auto-discover-apis