The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill accesses a sensitive private key file at ~/.config/ol-skill-feedback/private-key.pem to authenticate as a GitHub App for issue reporting.\n- [COMMAND_EXECUTION]: The skill executes a local Python script located at ~/.config/ol-skill-feedback/get-token.py to obtain installation tokens for the GitHub App.\n- [COMMAND_EXECUTION]: The skill utilizes the gh CLI to create issues. It interpolates user-provided variables including {skill_name}, {summary}, and the rendered issue body directly into the shell command parameters.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with GitHub (github.com), which is a well-known service, to create issues in the OntoLedgy/ol_ai_context_library repository.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes untrusted data from conversation history.\n
Ingestion points: The observed_output and expected_output parameters ingest potentially adversarial text from other skills or user inputs.\n
Boundary markers: No explicit delimiters or instructions are used to wrap the untrusted content when it is interpolated into the GitHub issue body.\n
Capability inventory: The skill can execute local scripts and perform network operations via the GitHub CLI.\n
Sanitization: There is no explicit logic described to sanitize or escape the content before it is passed to the shell command for issue creation.

skill-feedback