The agent that grows with you
Python aiai-agentai-agentsanthropicchatgptclaudeclaude-codecodexhermeshermes-agentllmnous-researchopenai
name: hermes-tenuo
repo: https://github.com/tenuo-ai/hermes-tenuo
sha: 8bce830c5df9dcd06d16688a954146d8edc0c562
description: >-
Security guardrails for Hermes tool calls: per-call authorization, so a
prompt injection cannot make the agent do more than you allowed. Each call
is checked against a signed, expiring permission slip (a warrant) before
the handler runs, down to the arguments (file paths, search queries,
spending limits), and every decision lands in a local audit log. Give cron
jobs, delegate_task subagents and gateway users their own policy.
Disclosure — hermes-tenuo is a no-op until a warrant is configured and blocks every tool call in a kanban worker that has no per-task warrant; it appends every tool call's name and full arguments to $HERMES_HOME/tenuo/audit.jsonl (disable with audit_log: false); when TENUO_CONNECT_TOKEN or TENUO_CONTROL_PLANE_URL are set, the bundled tenuo SDK (compiled Rust core) streams those decisions, including arguments, to that control plane.
title: Tenuo
maintainer: tenuo-ai
tier: community
category: tools
requires_hermes: ">=0.20"
docs_url: https://github.com/tenuo-ai/hermes-tenuo
version: "0.1.3"
capabilities:
provides_tools: []
provides_hooks:
- pre_tool_call
- post_tool_call
- on_session_start
- on_session_end
- subagent_start
provides_middleware: []
requires_env:
- TENUO_WARRANT
- TENUO_SIGNING_KEY