self-configuration

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill facilitates legitimate self-configuration of Letta agents through the official Letta API (api.letta.com). It allows agents to adjust their model, context window, and persistent settings as intended by the developer.
  • [EXTERNAL_DOWNLOADS]: The skill uses npx tsx to execute its internal TypeScript utility scripts (scripts/update-agent-settings.ts, scripts/update-compaction-prompt.ts). This is a standard method for running modern JavaScript/TypeScript tools within a development environment.
  • [COMMAND_EXECUTION]: The instructions include patterns for using curl and local scripts to interact with management endpoints. These operations are restricted to the agent's own ID and the current conversation context.
  • [PROMPT_INJECTION]: The skill provides instructions for the agent to modify its own system prompt and compaction prompt. While these are high-privilege operations that change the agent's core behavior, they are the primary stated purpose of the skill and include safety guidelines recommending user confirmation for persistent changes.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 04:06 PM
Security Audit — agent-trust-hub — self-configuration