Prompt Firewall
From its catalog listing
Community security-preview plugin that detects and quarantines selected prompt-injection patterns in tool results and evaluates risky tool actions; limited to hook/Core enforcement semantics.
Good to know before installing
Rewrites tool results the agent sees: spans that match its injection heuristics are replaced with `[QUARANTINED:<category>]` markers, and output past its scan budget (or output it fails to scan) is withheld with a notice; may block or escalate to the Hermes approval prompt writes to protected config/instruction/credential paths, endpoint changes and commands that disable prompt-firewall; adds a fixed trust-boundary notice to every turn; appends metadata-only audit records (tool name, target path, reason) to `<HERMES_HOME>/logs/prompt-firewall.jsonl`. No network.
From its catalog listing.
How to install
- Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install prompt-firewall - Said no? Turn it on later:
hermes plugins enable prompt-firewall - Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide (opens in a new tab).
- Category
- Security & privacy
- Maintained by
- Gaijin-01 (Community)
- Needs
- No API keys declared
- Adds hooks
pre_tool_calltransform_tool_resultpre_llm_call- Needs Hermes
- 0.22.7 or newer
- Version
- 1.1.4
- In the catalog
- 2026-11-10
Known issues (from its catalog listing)
- Security preview, not a sandbox; detection is heuristic and incomplete.
- Approval decisions depend on Hermes Core and can be bypassed by auto-approval settings.
Using the Hermes desktop app? See where plugins live

Full setup & settings ↗Source code ↗
More in security & privacy
All security & privacy plugins
New Hermes Agent updates every day on X. Follow @HermesWatcher