The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill depends on several external Python libraries including py-clob-client, requests, eth-abi, and eth-utils to handle API interactions and blockchain data encoding.
[COMMAND_EXECUTION]: The skill uses the bash tool to execute Python scripts for all its core functions. Documentation in SKILL.md includes templates for executing Python code snippets directly from the command line using the python -c flag, which involves interpolating dynamically generated data like signatures and token IDs.
[CREDENTIALS_UNSAFE]: API keys, secrets, and passphrases for the Polymarket CLOB are stored in a local JSON file at /data/workspace/.polymarket_creds.json. While this is standard for agent persistence, it concentrates sensitive access credentials in a single file within the workspace.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the ingestion of external data.
Ingestion points: Market descriptions, rules, and research data are pulled from the Polymarket API and external web searches.
Boundary markers: No explicit delimiters are used to separate untrusted market content from the agent's core instructions.
Capability inventory: The skill utilizes high-privilege tools like wallet_sign_typed_data and wallet_transfer to commit financial resources.
Sanitization: Market information is presented to the agent for analysis without robust filtering for potentially malicious instructions hidden in the market text.

@1368/polymarket-trade