The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The reference scripts references/fetch-farms.py and references/fetch-syrup-pools.py attempt to download and install the requests library from the official Python Package Index (PyPI) at runtime using pip.
[DYNAMIC_EXECUTION]: The Python scripts utilize os.system() and subprocess.check_call() to execute shell commands during execution to manage library dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including PancakeSwap explorer APIs, CoinGecko, and DexScreener. The skill body includes mandatory security rules instructing the agent to treat this data as untrusted, but the ingestion of attacker-influenced token names or symbols represents a potential attack surface.
Ingestion points: External API responses are piped into scripts (SKILL.md) or read via curl and node tools.
Boundary markers: Present; the skill provides explicit instructions to treat all external API response content as untrusted data.
Capability inventory: The skill has access to shell execution (Bash), network fetching (curl, WebFetch), and blockchain interaction tools (cast).
Sanitization: Absent; the skill relies on natural language instructions for the agent rather than programmatic escaping or filtering.
[COMMAND_EXECUTION]: The skill uses curl for API interaction and a telemetry ping to the vendor's domain, and provides instructions for using cast to interact with Ethereum-compatible blockchains.

farming-planner