workspace-adoption

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill configuration in assets/opencode.jsonc specifies third-party plugins to be added to the workspace, including @dietrichgebert/ponytail, @tarquinen/opencode-dcp, and @slkiser/opencode-quota. These packages originate from unverified sources on public registries and represent a potential supply chain risk.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from local workspace files to guide its configuration logic, creating an entry point for indirect instructions to influence agent behavior.
  • Ingestion points: The skill reads AGENTS.md, markdown files in the context/ directory, and workspace .opencode/ configuration files (SKILL.md).
  • Boundary markers: The instructions lack specified delimiters or explicit warnings to the model to ignore instructions embedded within the ingested files.
  • Capability inventory: The skill has the capability to write to the file system (staging and applying configuration changes) and query available models via tools.opencode.models.
  • Sanitization: There is no evidence of sanitization, validation, or filtering of the text content within the ingested workspace files.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 01:03 PM