linear-local-dev-loop
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from local repositories and Linear API responses, creating an attack surface where external content could influence agent behavior. * Ingestion points: Reads repository code, configuration files, and Linear API documentation snapshots (SKILL.md, references/official-docs.md). * Boundary markers: Explicitly instructs the use of isolated workspaces, labeled test records, and a 'protected live-test gate' that fails closed by default (SKILL.md). * Capability inventory: The agent is granted tools for filesystem modification (Write, Edit) and network access (WebFetch), which could be misused if influenced by malicious input. * Sanitization: Mandates replacing customer data with synthetic equivalents and redacting sensitive fields in all output (SKILL.md).
- [EXTERNAL_DOWNLOADS]: The skill references and fetches metadata for official Linear integration tools and documentation from established service domains. * Evidence: References official Linear API documentation and TypeScript SDK metadata from linear.app and official package registries (SKILL.md, references/official-docs.md).
Audit Metadata