agent-context-budget
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill frequently invokes the
agent-collabCLI utility to perform policy validation, checkpoint validation, and policy evaluation. These commands use variable inputs such as<policy-ref>and<checkpoint-ref>. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, potentially untrusted data to prepare context packets for sub-tasks.
- Ingestion points: The workflow involves reading inputs from
<policy-ref>,<checkpoint-ref>, and various 'evidence and artifact paths' (SKILL.md). - Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing the content of evidence files or policies.
- Capability inventory: The skill has the capability to execute shell commands (
agent-collab) and write files to the repository (e.g.,.coord/context_<NNN>.md,.coord/session_primer.md). - Sanitization: There are no explicit instructions for sanitizing file paths or validating the integrity of evidence content before it is processed or passed to subsequent roles.
Audit Metadata