The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes task descriptions, requirements, and comments from a decentralized marketplace. These are untrusted inputs that could contain malicious instructions designed to manipulate the agent's behavior.
Ingestion points: Untrusted content enters the agent context via the 0xwork discover and 0xwork task <id> commands described in SKILL.md.
Boundary markers: SKILL.md includes a detailed "Security: Untrusted Content Handling" section that provides explicit "Rules (non-negotiable)" and boundary definitions to help the agent distinguish between trusted system instructions and untrusted marketplace data.
Capability inventory: The agent uses the 0xwork CLI to perform on-chain operations and is instructed in references/execution-guide.md to use tools such as exec, write, and the file system to create deliverables.
Sanitization: The skill relies on prompt-level defensive instructions for the agent rather than technical sanitization of the marketplace content.
[DYNAMIC_EXECUTION]: In the "Code" category of the references/execution-guide.md, the agent is instructed to write and test code deliverables using exec and write tools. This involves executing code derived from untrusted task specifications.
[COMMAND_EXECUTION]: The skill requires the installation and use of the @0xwork/cli Node.js package to interact with the 0xWork protocol, manage wallet state, and submit work deliverables.

0xwork