The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes platform-specific tasks by running npx leapcat@0.1.1. This is the intended method for interacting with the Leapcat API and managing user sessions, trades, and portfolio data.
[EXTERNAL_DOWNLOADS]: Each command involves downloading or verifying the leapcat package from the npm registry. The skill uses a pinned version (0.1.1) to ensure execution environment stability and prevent automatic updates to unverified versions.
[DATA_EXFILTRATION]: The KYC skill includes an upload command that reads local files and transmits them to Leapcat's servers. While this involves moving local data to a remote endpoint, it is a required functionality for identity verification (Know Your Customer) and is constrained to user-provided file paths.
[PROMPT_INJECTION]: The skill processes untrusted external data from market quotes and search results, creating a surface for indirect prompt injection.
Ingestion points: Market data retrieved via market quote, market stocks, and portfolio positions commands.
Boundary markers: Instructions specify using the --json flag for data parsing, but no explicit delimiters are used to separate untrusted data from the agent's system prompt.
Capability inventory: The agent has the ability to execute shell commands via the npx interface.
Sanitization: There is no evidence of automated sanitization or filtering of the market data strings within the skill instructions.

leapcat