The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local parameter validation script scripts/validate_params.sh and a third-party polymarket CLI binary to perform market interactions and trading operations.
[EXTERNAL_DOWNLOADS]: The skill relies on external software dependencies, including the sports-skills Python package and the polymarket CLI binary hosted on GitHub. The CLI is sourced from a well-known organization (Polymarket), which is considered a trusted source for its own functionality.
[CREDENTIALS_UNSAFE]: Trading and on-chain commands require the user to configure a sensitive POLYMARKET_PRIVATE_KEY via environment variables or CLI commands. While standard for cryptocurrency interactions, this involves handling highly sensitive cryptographic material.
[PROMPT_INJECTION]: The skill includes functionality to fetch user-generated comments from the Polymarket platform via get_comments and get_user_comments, introducing a surface for indirect prompt injection.
Ingestion points: External user content is fetched from the Polymarket API and CLI (documented in references/commands.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat fetched comment data as untrusted or to ignore embedded instructions.
Capability inventory: The skill has powerful capabilities, including placing and canceling orders and managing contract approvals (create_order, cancel_order, approve_set).
Sanitization: There is no evidence of sanitization or filtering of the retrieved comment content before it is processed by the agent.

polymarket