Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly a coherent research prompt, but it asks the agent to persistently update the installed skill after each project without clear user approval.
Review or remove the maintenance notes before installing if you do not want an agent to edit the installed skill over time. Also treat the uninstall command carefully, since it permanently deletes the skill directory; use a confirmed or interactive removal step instead.
The command rm -rf ~/.hermes/skills/social-org-research is a forceful recursive deletion instruction in end-user documentation. Even with a constrained path, such commands are risky because users may execute them without scrutiny, potentially against the wrong path, through copy/paste mistakes, shell expansion issues, or from privileged contexts.
rm -rf ~/.hermes/skills/social-org-research
The command rm -rf ~/.hermes/skills/social-org-research is a forceful recursive deletion instruction in end-user documentation. Even with a constrained path, such commands are risky because users may execute them without scrutiny, potentially against the wrong path, through copy/paste mistakes, shell expansion issues, or from privileged contexts.
rm -rf ~/.hermes/skills/social-org-research
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
If you already have this repo cloned to ~/.hermes/skills/social-org-research/, you're done — Hermes auto-discovers skill folders.
ls ~/.hermes/skills/social-org-research/SKILL.md
Inside any Hermes conversation, /skills list should show "social-org-research".
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
If you already have this repo cloned to ~/.hermes/skills/social-org-research/, you're done — Hermes auto-discovers skill folders.
ls ~/.hermes/skills/social-org-research/SKILL.md
Inside any Hermes conversation, /skills list should show "social-org-research".
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
If you already have this repo cloned to ~/.hermes/skills/social-org-research/, you're done — Hermes auto-discovers skill folders.
ls ~/.hermes/skills/social-org-research/SKILL.md
Inside any Hermes conversation, /skills list should show "social-org-research".
The uninstall step uses a recursive force deletion command without any warning, confirmation, or safety guidance. While the target path is relatively specific, destructive shell commands in installation docs are dangerous because users may run them blindly, with elevated privileges, or after accidental path expansion/editing mistakes.
The trigger phrases are broad enough to match ordinary research or due-diligence requests, which can cause the agent to invoke this skill when a more appropriate skill should handle the task. Because this skill also pushes specific output tooling and workflow requirements, over-triggering can misroute user intent, expand scope unnecessarily, and increase the chance of unsafe or unintended downstream actions.
The case study embeds example user prompts such as "Help me deep-research BottleDream" and follow-up asks that are phrased like ordinary user requests, with no visible scoping or exclusion boundaries. In skills that may be auto-selected or triggered from natural-language matches, broad example phrasing can cause unintended activation on unrelated but common research requests, leading to overbroad behavior, unnecessary data gathering, or execution of downstream workflow steps such as report/PDF generation.
Line L052 explicitly says the package is "English-only," which is a natural-language locale restriction. The file does not offer a user language choice or opt-in, so this conflicts with the policy requirement to avoid forcing a specific language unless clearly optional or justified.
No suspicious patterns detected.