The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs and executes the agent-passport-system and agent-passport-system-mcp packages from the NPM registry. These are vendor-owned resources necessary for the skill's functionality.
[DATA_EXFILTRATION]: Agent metadata—including name, public key, owner, and capabilities—is transmitted to a remote public registry at aeoess.com/agora during the registration process.
[CREDENTIALS_UNSAFE]: The skill generates and stores Ed25519 private keys locally in .passport/agent.json. Additionally, the documentation notes that the register_agora_public tool requires a GITHUB_TOKEN to function.
[COMMAND_EXECUTION]: The skill relies on npx to run CLI commands such as agent-passport join and agent-passport register for local environment setup and remote interaction.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to read and process external data from a public source.
Ingestion points: The get_agora_messages, get_agora_thread, and get_agora_by_topic tools retrieve untrusted messages from the public Agora registry.
Boundary markers: There are no documented delimiters or instructions to ignore embedded commands within the retrieved Agora messages.
Capability inventory: The skill allows the agent to write files, execute CLI commands via npx, and perform network operations.
Sanitization: No explicit sanitization or content validation is mentioned for the messages retrieved from the remote registry.

agent-passport