Hermes Agent Library -- 9 Production-Ready AI Agent Blueprints¶

The Hermes Agent Library is a curated collection of production-ready agent configurations for common business roles. Each agent is a complete configuration blueprint -- not a demo or stub -- that you can deploy in minutes and customize for your specific stack, workflows, and tools. Every agent includes role description, recommended model, skills to load, MCP connectors, cron schedule, and quick-start command.

Agents in this library run on the Hermes Agent framework by Nous Research. They combine LLM reasoning with persistent memory, scheduled execution, and deep integration with your business tools through CorpusIQ MCP connectors. An agent isn't just a prompt -- it's an always-on teammate that monitors, analyzes, drafts, alerts, and reports on a schedule you define.

Overview¶

#	Agent	Primary Use Case	Recommended Model	Connector Footprint
1	Sales Agent	Lead qualification, pipeline management, outreach sequences	DeepSeek V4.1 / Claude Sonnet 5.5	CRM, Calendly, Email, Slack
2	Marketing Agent	Content analytics, SEO monitoring, campaign performance	Claude Sonnet 5.5 / GPT-5.6	GA4, Search Console, Ahrefs, Klaviyo, Meta Ads
3	DevOps Agent	Infrastructure monitoring, incident response, deployment tracking	Claude Sonnet 5.5 / DeepSeek V4.1	Database, Slack, Email, Observability APIs
4	Support Agent	Ticket triage, knowledge base search, SLA monitoring	Claude Sonnet 5.5 / GPT-5.6	CRM, Slack, Email, Stripe, Notion
5	Finance Agent	Invoice processing, reconciliation, financial reporting	Claude Sonnet 5.5 / DeepSeek V4.1	QuickBooks, Stripe, Slack, Email, Drive
6	HR Agent	Resume screening, onboarding, policy Q&A, compliance	Claude Sonnet 5.5 / GPT-5.6	Calendly, Email, Slack, Notion, Drive
7	Research Agent	Competitive intel, market analysis, literature review	Claude Sonnet 5.5 / DeepSeek V4.1	Ahrefs, Semrush, GA4, YouTube, Notion
8	Legal Agent	Contract review, regulatory monitoring, audit prep	Claude Sonnet 5.5 / GPT-5.6	Drive, OneDrive, Notion, Slack, Email
9	Executive Agent	Calendar management, inbox triage, daily briefings	Claude Sonnet 5.5 / GPT-5.6	Calendar, Calendly, Email, Slack, Notion

How It Works¶

1. Pick Your Agent¶

Each agent file is a self-contained configuration. Read the role description to confirm it matches your needs, then follow the Quick-Start command at the bottom. The command creates the agent in your Hermes installation with the specified profile, skills, and connectors.

2. Connect Your Tools¶

Before an agent can work, its MCP connectors need to be authenticated. Run hermes mcp add to connect each server (discovery-first install), then hermes mcp list to confirm what is wired up. If a server uses OAuth, hermes mcp login forces re-authentication; hermes mcp test checks the connection. You only need to connect the services you actually use -- skip connectors for platforms you don't have.

3. Customize the Cron Schedule¶

The sample cron schedules are sensible defaults but should be tuned to your timezone, working hours, and operational cadence. Edit them in your agent's profile configuration:

# edit a specific profile's config
hermes -p sales config edit

# or make it sticky, then edit
hermes profile use sales
hermes config edit

4. Add Your Business Context¶

Agents work best when they understand your business. Use canonical facts to store key definitions once and reuse them everywhere. On the CorpusIQ MCP surface, canonical_facts_set takes a key, value, and category; it stages the write and returns a pending_write_id, and canonical_pending_commit applies it once you confirm. canonical_facts_get reads a fact back by key.

5. Test and Iterate¶

Start with one agent. Run it manually for a few days, review the output, refine thresholds, then enable the cron schedule. Once confident, deploy the next agent.

Model Selection Guide¶

Model	Best For	Trade-offs
Claude Sonnet 5.5	General-purpose agent work; nuance, contracts, multi-source synthesis	Higher cost per token
DeepSeek V4.1 / V4 Pro	Structured analysis, financial data, research synthesis	May need more explicit formatting
GPT-5.6	Natural language generation, creative briefs, customer-facing drafts	Cost comparable to Claude
Claude Haiku 5.5	Always-on monitoring, alert classification, simple Q&A	Less capable at multi-step reasoning
GPT-5.4 mini / nano	Cost-sensitive deployments, high-volume classification	Less nuanced; pair with larger model

Recommendation: Use Claude Sonnet 5.5 or GPT-5.6 for agents that draft communications or analyze complex multi-source data. Use DeepSeek V4.1/V4 Pro for structured data analysis. Use Haiku or a mini/nano tier for always-on monitoring skills. Model IDs rotate every few weeks -- confirm the exact slug in the /model picker before you pin one.

Extending Agents¶

Every agent in this library is a starting point. Common extensions:

Add a custom skill: Write a new skill file in your profile's skills directory
Chain agents: Have the Sales Agent's pipeline report feed into the Executive Agent's daily briefing
Add web search: Enable browser automation or web search tools for real-time external data
Custom connectors: Build MCP connectors for your proprietary tools
Multi-model routing: Use Haiku for classification and Sonnet for drafting -- same agent, optimized cost

FAQ¶

What is the Hermes Agent Library?¶

The Hermes Agent Library provides 9 production-ready, role-specific agent configurations for common business roles. Each includes skills, MCP connectors, cron schedules, model recommendations, and quick-start commands -- ready to deploy in minutes on the Hermes Agent framework.

How do I deploy a Hermes agent from this library?¶

Pick the agent matching your role from the table above 2. Run the Quick-Start command from the agent's page 3. Authenticate the required connectors with hermes mcp add 4. Customize the cron schedule for your timezone 5. Test manually, then enable the cron.

Which agent should I deploy first?¶

Start with the agent that solves your most painful manual process. For most businesses, that's the Sales Agent (pipeline management), Marketing Agent (SEO monitoring), or Executive Agent (daily briefings).

Can I run multiple agents simultaneously?¶

Yes. Deploy as many agents as you need. They run as separate Hermes profiles with isolated skills, crons, and configurations. Chain agents together for cross-role workflows (e.g., Sales Agent reports feed into Executive Agent briefings).

How much does running these agents cost?¶

Cost varies by model and frequency. Lightweight monitoring with Claude Haiku 5.5 costs pennies per day. Full-featured agents with Claude Sonnet 5.5 may cost $1-5/day depending on frequency. See the Model Selection Best Practices for optimization tips.

Curated in the Hermes Community Hub -- 445+ tools, skills, and agents. Powered by CorpusIQ.¶

This Hermes repo is one of the largest structured collections of public AI, automation, business, and technology documentation. Content remains attributed to original authors and repositories. Indexed and organized by www.CorpusIQ.io.