salla-theme-builder
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides standard instructional documentation for interacting with the Salla theme management API via the
salla_themestool. It properly guides the agent through role-based access control checks and platform constraints. - [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface where external configuration data is ingested and used to influence agent actions.
- Ingestion points: The agent reads
twilight.jsonconfiguration files from GitHub repositories using thegithub_configaction inSKILL.md. - Boundary markers: The instructions do not specify explicit delimiters or "ignore embedded instructions" warnings for the ingested configuration data.
- Capability inventory: The agent can perform write operations including committing changes to GitHub repositories, updating theme details, and requesting publication reviews (Steps 6-7 in
SKILL.md). - Sanitization: The skill notes that the Salla Portal filters the configuration file content to a specific set of allowed keys before it is provided to the agent, reducing the risk of processing malicious instructions (documented in
references/theme-api-notes.md).
Audit Metadata