The Agent Skills Directory

EXTERNAL_DOWNLOADS (HIGH): The skill mandates using web_fetch to retrieve content from several external URLs, including github.com/ApeWorX/uniswap-sdk, docs.apeworx.io, and github.com/ApeWorX/ape-tokens. These sources are not on the Trusted External Source list.
PROMPT_INJECTION (HIGH): This skill is highly susceptible to Indirect Prompt Injection. It explicitly instructs the agent to use fetched external content as authoritative guidance for writing code. An attacker controlling these external pages could inject instructions to divert funds, modify trade parameters, or exfiltrate sensitive data.
Ingestion points: web_fetch calls to GitHub and ApeWorX documentation.
Boundary markers: None. The instructions demand strict adherence to the fetched documentation.
Capability inventory: Writing and executing Python code, interacting with blockchain networks, and managing ape accounts.
Sanitization: None detected.
COMMAND_EXECUTION (HIGH): The skill involves generating and running code that interacts with the Uniswap protocol and local Ape accounts. Combining dynamic code generation with untrusted external documentation creates a significant risk of Remote Code Execution (RCE) via injected logic.
CREDENTIALS_UNSAFE (MEDIUM): The skill references ape accounts and "live network accounts." While no keys are hardcoded, the workflow involves the agent handling financial credentials. If the documentation fetching is compromised, these credentials could be exposed or misused.

uniswap