Security risk report for @forgecat/voltagent_awesome-codex-subagents_business-product v0.0.11
Source Integrity
Low
Profile references a public GitHub repository (VoltAgent/awesome-codex-subagents) with MIT license and clear attribution.
No typosquatted, obfuscated, or attacker-controlled package dependencies are declared; no external binary execution is requested.
Agent Intent
Low
Content describes legitimate business and product agent roles (analyst, marketer, PM, legal advisor, etc.) with clear, task-focused instructions aligned to their stated purposes.
No instructions attempt to manipulate the AI into ignoring system prompts, exfiltrating credentials, reading sensitive files, or hiding its own instructions.
No guidance poisoning: recommendations are standard professional practices (requirements clarity, scope control, user-impact prioritization) without deceptive or security-weakening directives.
Agents are instructed to verify claims, flag risks, and escalate unresolved decisions—reinforcing integrity rather than undermining it.
Details
Evidence
business-analyst: 'verify every requirement maps to a concrete behavior or outcome' and 'call out assumptions that must be confirmed by a human decision-maker'
content-marketer: 'verify every core claim maps to observable product behavior' and 'flag claims that require product/legal/engineering verification before publish'
legal-advisor: 'verify each flagged risk cites concrete text or behavior evidence' and 'call out areas requiring qualified counsel before release decisions'
All agents include quality checks emphasizing evidence, feasibility, and escalation—no hidden manipulation or credential exfiltration.
Permissions
Low
No tools or hooks are declared; the profile is a collection of agent role definitions and working instructions, not executable code with system access.
Two agents (technical-writer, wordpress-master) declare sandbox_mode: workspace-write, which is scoped to documentation and WordPress site editing—appropriate to their stated functions.
No shell, file_delete, unrestricted file_write, or network access is requested; authority is limited to read-only analysis and bounded workspace editing.
Details
Evidence
technical-writer sandbox_mode: workspace-write (for drafting release notes and docs)
ForgeCat