The Agent Skills Directory

EXTERNAL_DOWNLOADS (MEDIUM): The README.md recommends installation using npx skills add tumf/clawdbot-config. The account tumf is not on the trusted sources list, posing a risk of executing untrusted code during the installation process.
DATA_EXFILTRATION (MEDIUM): The config-schema.md defines a configuration for the iMessage channel that targets ~/Library/Messages/chat.db. This path contains highly sensitive user message history. Accessing this file exposes private data to the agent.
COMMAND_EXECUTION (MEDIUM): The schema includes configuration for a tools.exec setting and a Docker setupCommand. These settings enable the execution of arbitrary system commands, which presents a significant risk if the agent is manipulated by untrusted input.
PROMPT_INJECTION (LOW): The skill defines configurations for messaging channels (WhatsApp, Telegram, Discord) that serve as ingestion points for untrusted external data. This architecture is vulnerable to indirect prompt injection. Mandatory Evidence Chain: 1. Ingestion points: messaging channel configurations in references/config-schema.md. 2. Boundary markers: No delimiters or ignore instructions are specified in the schema. 3. Capability inventory: exec tool, browser tool, and Docker setupCommand defined in references/config-schema.md. 4. Sanitization: No sanitization or validation logic is present in the provided schema.

clawdbot-config