The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the heretic-llm package from PyPI and clones source code from https://github.com/p-e-w/heretic during the installation process.
[REMOTE_CODE_EXECUTION]: The skill facilitates the execution of third-party code from the heretic-llm repository via the heretic CLI tool and the provided installation script.
[DYNAMIC_EXECUTION]: The scripts/install.sh script uses the eval command to execute shell strings constructed from environment variables such as UPSTREAM_URL, REF, and PIP. This pattern can lead to arbitrary command injection if the environment variables are manipulated by an attacker.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves content from external websites (such as Hugging Face and Reddit) via the scrapling skill to research models and benchmarks.
Ingestion points: External web content, model cards, and community discussions are read into the agent's context.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands in retrieved data are defined in the instructions.
Capability inventory: The skill can execute shell commands (Bash), modify the local file system (Write, Edit), and perform network requests (WebFetch).
Sanitization: There is no evidence of sanitization or validation of retrieved external content before it is processed by the agent.

heretic