Hermes Agent Automation Blueprints Turn Cron Into Conversational Workflows

Hermes Agent shipped Automation Blueprints this week, a feature that converts scheduled cron jobs into conversational, reusable workflow templates. Users describe what they need in plain English, and Hermes builds the blueprint -- trigger, tools, actions, and output destination -- through a normal chat interaction.

What changed

Before this update, scheduling an agent task required writing a cron expression, configuring a script, wiring the agent to the right tools, and troubleshooting the connection manually. If the task needed web research, you connected Firecrawl. If it needed code review, you set up GitHub access. Each configuration step was manual.

Automation Blueprints collapse this into a conversation. The user states the outcome, Hermes asks for any missing details (API keys, schedule preferences, output location), and the complete recipe is saved as a reusable template. The same blueprint can be copied, customized for a different profile, or shared with another user.

"Instead of writing scripts and configuring the schedule manually, Hermes asks what you need and creates/schedules the task." -- @Abdullah4AI

Three trigger mechanisms

Blueprints support three ways to initiate execution:

These are accessed through the /cron slash command inside Hermes or through hermes webhook for external triggers. The blueprint system is not a UI wrapper around cron -- it is a structured template format that captures the full workflow: trigger condition, connected tools, processing steps, and output destination.

How it works in practice

A user opens a conversation with Hermes and describes the recurring task in plain English. Hermes translates that into a blueprint structure. If the workflow needs web research, Hermes guides the user through connecting Firecrawl. If it needs repository access, GitHub integration becomes part of the same blueprint.

Once built, the blueprint can be:

• Copied to another Hermes profile for a different use case
• Edited through conversation when the schedule, topic, or output location changes
• Shared with another person who can reuse the process without designing it again
• Triggered manually when an early report is needed before the next scheduled run

A typical blueprint for a Monday research digest might look like this:

Trigger: Schedule (Monday 8:00 AM)
1. Firecrawl: search for [industry keywords] from past 7 days
2. Filter: remove duplicates and low-relevance results
3. Format: organize findings into structured markdown
4. Output: save to Obsidian vault /Research/Weekly/
5. Notify: send summary link to Telegram

The same structure adapts to different domains. A development team might set up:

Trigger: GitHub Event (new pull request)
1. Claude: review diff for potential issues
2. Format: structure findings as PR review comments
3. Output: post to PR thread
4. Archive: save review to Obsidian for trend tracking

The key architectural shift

The most interesting aspect of Automation Blueprints is not the scheduling mechanism itself -- it is what the feature represents for agent productization.

"The productization path for agents isn't necessarily more complex multi-agent setups -- it's transforming a successful single task into a reusable SOP." [^1]

This challenges the default assumption that the next step for agents is orchestrating larger agent swarms. Instead, the blueprint approach treats the conversation that produced a successful result as a permanent artifact. The recipe -- not just the output -- becomes reusable.

Agent OS provides the management layer where active blueprints, profiles, schedules, and completed outputs remain visible. Different Hermes profiles can handle separate business areas without mixing their instructions. A research profile might manage Firecrawl searches and weekly market reports, while another profile handles GitHub monitoring and code review preparation. Manual triggers allow a workflow to run early when information is needed before the scheduled time. Schedules can be paused or edited without rebuilding the entire blueprint.

What's available now

Built-in templates ship with the feature, including morning summaries, news digests, and GitHub triage workflows. These serve as starting points that can be copied and customized rather than requiring every user to build from a blank page.

The /cron command surfaces these templates inside Hermes, and hermes webhook enables external API triggers. Documentation is available through Hermes Agent's built-in help system.

Automation Blueprints are available in the current Hermes Agent release. The feature works with any configured model -- lighter models like Gemma 4 can handle simple scheduled tasks, while larger models support more complex multi-step workflows.

[^1]: @fuckbiking on X. "官方发布：把 cron jobs 变成可点击、可填写、可对话的工作流。这很关键：agent 产品化入口不一定是更复杂的多智能体，而是把一次成功任务沉淀成可复用 SOP。"