The Agent Skills Directory

[COMMAND_EXECUTION]: The SKILL.md file contains instructions to execute local Python scripts (scripts/hiring_plan_modeler.py and scripts/comp_benchmarker.py) in the Quick Start section. These scripts are missing from the provided files, preventing an audit of the commands they execute or their interaction with the system.
[DATA_EXFILTRATION]: The multi-agent workflow includes tasks that transmit company identifiers, growth stages, and role lists to external platforms like Levels.fyi, Glassdoor, and LinkedIn. Sending specific internal organizational data to third-party endpoints presents a risk of sensitive data exposure.
[PROMPT_INJECTION]: The skill is designed to ingest and synthesize data from external websites, such as Glassdoor reviews and LinkedIn signals, which are untrusted sources. This surface is vulnerable to indirect prompt injection if an attacker embeds malicious instructions in public reviews or profiles.
Ingestion points: External data retrieved via sub-agent tasks targeting Glassdoor and LinkedIn.
Boundary markers: No explicit markers or instructions to disregard embedded commands were found in the synthesis logic.
Capability inventory: The skill includes documentation for executing local Python scripts as part of its core functionality.
Sanitization: There is no evidence of sanitization or validation for the content retrieved from external research tasks before it is processed by the agent.

chro-advisor