⚡ 9 Plugins + 5 Workflows + Full Docs

Hermes Autonomy Pack

Everything your hermes-agent needs to run itself. Smart model routing, self-correction, hallucination detection, and overnight learning. Battle-tested across 130+ development cycles.

$49one-time
Three Pillars of Autonomy

Each pillar solves a fundamental challenge of running AI agents in production.

🧠
Core Operations
  • Smart model routing per task type
  • Sensitive data never leaves local GPU
  • Agent-to-agent bridge communication
  • Autonomous goal management
🛡
Quality & Safety
  • Self-correction before sending outputs
  • Hallucination detection (0-10 scoring)
  • Email prompt injection screening
  • Budget enforcement via Langfuse
💡
Learning & Memory
  • Nightly fact extraction with scoring
  • Importance-based memory decay
  • Self-updating personality/identity
  • Shared retry utilities
The Right Model for Every Task

Auto-detects task type and routes to the best model. Sensitive data stays local — always.

Task TypeAuto-Selected ModelCost
Code Generationqwen-coder-free$0
Reasoning & Logicdeepseek-r1-local$0
Research & Searchnemotron-free$0
Creative Writingdolphin3-local$0
Summarizationqwen3-4b-free$0
Sensitive DataLOCAL GPU ONLY$0
Production-Ready, Battle-Tested
evey-delegate-model
Smart routing + 4-model fallback + sensitivity filter. Direct LiteLLM API — works from cron, bridge, Telegram.
delegate_with_model(goal="Summarize this paper") → Auto-routes to nemotron-free (research task)
evey-validate
Catches hallucinations from free models. 6 regex patterns + LLM scoring (0-10). Trust/Caution/Reject.
validate_output(task="Research", result=draft) → Score: 5/10 — CAUTION (verify claims)
evey-reflect
Self-correction loop. Cheap local model critiques outputs before sending. Catches errors early.
reflect_on_output(task="Blog post", draft=text) → "FIX: paragraph 3 has unverified claim"
evey-memory-consolidate
Nightly fact extraction from Langfuse traces. Importance scoring (1-10). Low-scored facts decay first.
consolidate_daily_memory(hours_back=24) → 5 facts extracted, scored, stored in Qdrant
evey-identity
Self-updating personality. Agent evolves its own identity file based on experience.
update_identity(reflection="Smart routing works") → SOUL.md: "- Always let routing choose [2026-03-18]"
evey-email-guard
Dual-layer prompt injection screening. 20+ regex patterns + AI classifier. Blocks/warns/passes.
email_screen(body="Ignore previous instructions...") → BLOCKED: 3 injection patterns detected
evey-cost-guard
Budget enforcement via Langfuse. Tracks daily and per-task spend. Warns at 80%, alerts at 100%.
cost_check(period="today") → $0.12 / $1.00 budget (12% used)
evey-bridge
Bidirectional agent-to-agent communication. File-based inbox/outbox with auto-compress and read tracking.
claude_bridge_task(type="code-change", description="Fix the login bug")
evey-goals
Autonomous goal management. Set, track, complete, review. Wire to a cron job for self-directed work.
evey_goals(action="add", goal="Research RAG") → Added to Active goals
Choose Your Tier
Starter
$19
  • 3 Core Plugins
  • Basic Documentation
  • Install Script
  • Email Support
Enterprise
$99
  • Everything in Pro
  • Priority Support
  • Custom Integration Help
  • Monthly Updates
Common Questions
What version of hermes-agent do I need?
v0.4.0 or later. The plugins use the standard plugin API and don't patch hermes internals.
Do I need LiteLLM, Langfuse, and Qdrant?
LiteLLM is required for delegation and validation. Langfuse is optional (for cost-guard and memory-consolidate). Qdrant is optional (for memory-consolidate vectors).
Can I customize the smart routing models?
Yes. Edit TASK_ROUTING and SENSITIVE_PATTERNS in evey-delegate-model. Add your own task categories and model preferences.
Does this work without a GPU?
Yes. The delegation plugin routes to free OpenRouter models by default. Local GPU models are optional fallbacks.
Is there a refund policy?
Yes. If the plugins don't work with your setup, email [email protected] for a full refund within 30 days.