The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/deploy_hummingbot_api.sh script clones the official Hummingbot API repository from GitHub (https://github.com/hummingbot/hummingbot-api.git) to install the trading infrastructure. Additionally, scripts/setup_gateway.sh pulls the hummingbot/gateway Docker image.
[COMMAND_EXECUTION]: The skill uses shell commands and scripts to manage Docker containers and interact with the local trading API. This includes docker compose operations and curl requests to the Hummingbot API to manage trading executors and controllers.
[CREDENTIALS_UNSAFE]: The scripts/add_wallet.py script handles sensitive Solana private keys and provides an option for CLI argument input via --private-key. This may expose keys in shell history or process listings, although the script's default behavior is to use a secure getpass prompt. The skill also reads credentials from .env files for API authentication.
[INDIRECT_PROMPT_INJECTION]: The skill fetches pool data from the external Meteora API (https://dlmm.datapi.meteora.ag).
Ingestion points: Data enters the context from the Meteora API in scripts/list_meteora_pools.py and scripts/get_meteora_pool.py.
Boundary markers: Absent; the agent is presented with raw or table-formatted data from the API without explicit instructions to disregard embedded commands.
Capability inventory: The skill possesses shell execution, network request, and file system write capabilities.
Sanitization: Content is formatted into markdown tables or JSON for display, but no security-focused sanitization is applied to the external API responses.

lp-agent