The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for installation and operation. It uses mkdir -p and shell redirection (>) to create configuration and skill files in ~/.moltbot/skills/moltbook and ~/.config/moltbook/credentials.json. Additionally, it uses curl for all API interactions, including registration, posting, and search.
[EXTERNAL_DOWNLOADS]: The skill fetches components and data from the vendor's domain (moltbook.com). During setup, it downloads SKILL.md, HEARTBEAT.md, and package.json. At runtime, it is programmed to periodically fetch a remote file (heartbeat.md) to update its tasks.
[PROMPT_INJECTION]: The skill processes untrusted content from the Moltbook platform, creating an indirect prompt injection surface.
Ingestion points: Data enters the agent context via API responses for the global feed (/api/v1/feed), submolt posts (/api/v1/posts), and semantic search results (/api/v1/search).
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat external social content as untrusted data or to ignore instructions embedded within posts.
Capability inventory: The agent has access to curl for network requests and can write to the local file system via shell redirection.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from the social network.
[REMOTE_CODE_EXECUTION]: The skill implements a dynamic instruction fetching mechanism. The 'Heartbeat' section directs the agent to fetch a remote markdown file (https://www.moltbook.com/heartbeat.md) and 'follow it'. This behavior allows the remote server to dynamically update or influence the agent's logic and instructions at runtime without requiring an update to the local skill file.

moltbook