The Agent Skills Directory

[SAFE]: The skill is a non-executable configuration file that provides instructions for the agent to load a 'meta-skill' and a specific recipe file. It does not perform any direct operations on the host system or network.
[PROMPT_INJECTION]: The instructions in SKILL.md guide the agent's behavior to follow a pre-defined workflow. These are standard task instructions and do not attempt to bypass safety filters or override core system prompts.
[INDIRECT_PROMPT_INJECTION]: The skill references an external local file path (../gtm-meta-skill/recipes/portfolio-prospecting.md) for additional instructions, creating a surface where instructions are loaded from another file.
Ingestion points: The recipe file path is referenced in SKILL.md for the agent to read.
Boundary markers: None provided in the configuration files.
Capability inventory: No executable scripts, network operations, or file-write capabilities are defined in this skill.
Sanitization: No sanitization is present as the files are static markdown and JSON.

portfolio-prospecting