skill-escalation
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the GitHub CLI (
gh issue create) to submit improvement proposals to the author's repositories (philtzjp/skillsandphiltzjp/startingpoint). - [INDIRECT_PROMPT_INJECTION]: The skill has a workflow that ingests data from untrusted external sources and uses it to perform actions.
- Ingestion points: External information gathered through "Web searches" (Web 検索) as mentioned in the usage section of
SKILL.md. - Boundary markers: None identified. The skill does not explicitly instruct the agent to use delimiters or ignore instructions within the search results.
- Capability inventory: The skill uses the GitHub CLI (
gh issue create) to write data to external repositories (inSKILL.md). - Sanitization: No specific sanitization or validation steps for the external data are described before it is used to populate issue templates.
Audit Metadata