celestia
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of documentation and reference material for the Celestia-node software. No executable code, scripts, or automated configuration changes are included in the skill package.
- [SAFE]: External references point to well-known and legitimate sources, including the official Celestia GitHub organization (celestiaorg) and reputable developer repositories.
- [SAFE]: No obfuscation, prompt injection patterns, or data exfiltration vectors were identified in the metadata or instructional content.
- [INDIRECT_PROMPT_INJECTION]: The skill describes node software that ingests data from a P2P network (headers and shares). However, the documentation explicitly details verification mechanisms such as Merkle proofs and fraud proofs (BEFP) used by the software to maintain data integrity. As this is a documentation-only skill, the attack surface for the AI agent itself is negligible.
Audit Metadata