agent-swarm-deployer

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process large volumes of external, untrusted data by distributing it to sub-agents.
  • Ingestion points: Data is ingested from directories, CSV files, or JSON arrays using the Glob, Bash, and Read tools as described in SKILL.md and references/agent-brief.md.
  • Boundary markers: The skill uses structured briefs and ## Input Data headers to delimit content, which provides some context separation but does not prevent sub-agents from obeying instructions embedded within the data items.
  • Capability inventory: The sub-agents are tasked with complex operations (summarization, extraction, generation) while the Commander agent uses Bash for file manipulation and the Agent tool for parallel execution.
  • Sanitization: There is no evidence of input sanitization or escaping for the data embedded in the {ITEMS_AS_STRUCTURED_DATA} section of the agent briefs, allowing malicious instructions in the data to potentially influence sub-agent behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 07:25 AM
Security Audit — agent-trust-hub — agent-swarm-deployer