The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze user-supplied information, including financial metrics (Total Shareholder Return, peer returns, capital decisions) and specific queries about CEO performance. This creates a potential surface where adversarial instructions could be embedded in the data being analyzed to influence the agent's evaluation output.
Ingestion points: SKILL.md (Workflow Inventory and Query Response Framework sections).
Boundary markers: No specific delimiters or warnings to ignore instructions within user-provided data are present in the framework instructions.
Capability inventory: The skill does not define any executable tools, file-system access, or network operations, limiting the impact of any potential injection.
Sanitization: No sanitization or validation protocols for the input data are defined within the skill instructions.
[SAFE]: The installation instructions in the README.md use a standard package execution command (npx) to add the skill from a repository belonging to the skill's author (simbajigege). This is a legitimate distribution method for the author's resources.
[SAFE]: All cited excerpts and external references in SKILL.md and the quotes/ directory point to the author's official GitHub repository and provide source-grounded documentation for the decision framework.

outsiders-ceo-capital-allocation-thorndike