agent-context-budget
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured guidelines for managing token and cost budgets across multiple AI agents. It uses standard file operations within specialized directories (.coord/ and .ai/) to maintain state and perform session priming.
- [DATA_EXPOSURE]: The skill explicitly advises against the promotion of secrets or credentials to the agent's memory (see 'Memory Promotion Rules').
- [COMMAND_EXECUTION]: The skill utilizes standard Git commands for versioning coordination artifacts, which is consistent with the described development workflow.
- [INDIRECT_PROMPT_INJECTION]: While the skill processes outputs from other agents (logs, summaries), it includes structural mitigation by enforcing strict word and token budgets, requiring concise summaries rather than raw log ingestion, which reduces the surface area for injection attacks. This is assessed as safe practice.
Audit Metadata