hal-vault
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Recommends downloading the
hal-vaultbinary or source code from the author's official GitHub repository (github.com/ofoxai/hal-vault). - [COMMAND_EXECUTION]: Instructs the agent to perform CLI operations including vault initialization, secret storage (
add), and retrieval (get), as well as usingssh-keygento create dedicated encryption keys. - [INDIRECT_PROMPT_INJECTION]: The skill manages a surface where the agent processes untrusted user data (credentials).
- Ingestion points: User-provided secrets shared in conversation (SKILL.md).
- Boundary markers: Explicit 'safety contract' requiring masked output and prohibiting raw secret display (SKILL.md).
- Capability inventory: Shell execution of
hal-vault,curl,sed, andssh-keygen(SKILL.md, references/cli-reference.md). - Sanitization: Input is validated for UTF-8 compatibility by the CLI tool.
Audit Metadata