gems
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use common shell tools such as
sed,git status, andgit logto manage the ledger file and verify repository state. These actions are limited to the local filesystem and project environment, serving the primary purpose of context maintenance. - [INDIRECT_PROMPT_INJECTION]: The skill maintains an attack surface for indirect prompt injection because it ingests data from user messages and local files to generate a ledger that acts as a 'source of truth' for future sessions.
- Ingestion points: Reads and processes content from
.gemkeeper/ledger.mdand the current user conversation history. - Boundary markers: Relies on Markdown structural elements like headers and bullet points to organize and delimit ledger entries.
- Capability inventory: The agent is permitted to write to the local filesystem and execute specific shell commands to update the ledger.
- Sanitization: No explicit instructions are provided for sanitizing or escaping content extracted from user messages or project files before writing to the ledger.
Audit Metadata