The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation instructions direct the user or agent to install a package from a third-party GitHub repository (github.com/joaquinhuigomez/agent-eval) via pip. This source is not associated with a verified organization or well-known service.
[COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary shell commands defined within YAML task configurations. These commands, specified in the judge section (e.g., pytest, npm run build), are executed as part of the tool's validation workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes external YAML task definitions and local codebases which can serve as vectors for indirect prompt injection.
Ingestion points: YAML task files and local project repositories specified in the configuration.
Boundary markers: None identified in the skill instructions to separate data from instructions.
Capability inventory: File system access (Read, Write, Edit, Grep, Glob tools) and shell command execution (Bash tool).
Sanitization: No description of validation or sanitization for shell commands defined in the YAML files before execution.
[DYNAMIC_EXECUTION]: The tool dynamically constructs and executes shell commands based on inputs found in data files (YAML configurations).

agent-eval