Orchestration
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The
catalog.jsonfile contains instructions for the user to install a plugin from an external GitHub repository (https://github.com/austintgriffith/ethskills) and a package viaclawhub install ethskills. - [INDIRECT_PROMPT_INJECTION]: The skill instructions in
catalog.jsonsuggest that the user tell the agent to "Read https://ethskills.com/orchestration/SKILL.md before scaffolding". This directs the agent to ingest content from an external, third-party URL into its active context. - Ingestion points: The agent is instructed to fetch a remote markdown file (
SKILL.md) fromethskills.com. - Boundary markers: Absent. The instructions do not include delimiters or warnings to the agent to treat the remote content as untrusted or to ignore embedded instructions.
- Capability inventory: The skill body and demo code demonstrate capabilities for shell command execution via
yarn(e.g.,yarn deploy,yarn fork), which could be targeted by instructions embedded in the remote file. - Sanitization: Absent. There is no evidence of filtering or validation for the remote content before it is processed by the agent.
- [COMMAND_EXECUTION]: The
catalog.jsonfile defines an installation commandcurl -s https://ethskills.com/orchestration/SKILL.md. While this specific command only fetches a file, the demo code illustrates the use of shell commands for contract deployment and infrastructure management.
Audit Metadata