The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's setup instructions (SKILL.md) direct users to execute a remote shell script via curl -fsSL https://cli.inference.sh | sh. This pattern is highly risky as it executes unverified code from the internet with the user's local system privileges.\n- [COMMAND_EXECUTION]: The skill uses the infsh/python-executor tool to run embedded Python code (SKILL.md) at runtime. This allows the agent to execute arbitrary Python logic, which in this case is used for generating positioning maps using matplotlib.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from the web during the competitive analysis process.\n
Ingestion points: Untrusted content is fetched via the tavily/search-assistant, exa/search, and infsh/agent-browser tools (SKILL.md).\n
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat fetched web content as untrusted data rather than instructions.\n
Capability inventory: The environment allows powerful actions including shell access via Bash and arbitrary script execution via python-executor.\n
Sanitization: No input validation or sanitization is performed on retrieved competitor data before it is processed into final analysis deliverables.

competitor-teardown