Should a workflow start scheduled?
No. Run it manually first, then add a schedule only after inputs, outputs, boundaries, and stop rules are boring and repeatable.
From setup to repeatable work
Hermes Agent workflows
Workflows should start narrow: one trigger, one workspace, one provider path, one safety boundary, and one observable output. Expand after the first version is boring and repeatable.
Agent Guide is an independent editorial resource. It is not affiliated with, endorsed by, or sponsored by Nous Research, Hermes Agent, or Hermes/Hermes brand owners. Product names and marks belong to their respective owners.
Intent
hermes-agent-workflows-overview
Sources
6
Schema
2
Links
6
Starter workflow gallery
Daily briefing
Collect selected sources, summarize changes, and produce a morning note.
PR review assistant
Review diffs against project rules, then leave a draft summary for a human.
Research tracker
Monitor official docs, releases, issues, and community patterns for guide updates.
Messaging assistant
Handle Telegram or Discord workflows with scoped credentials and clear cutover rules.
Workflow readiness test
- Can you run it manually once?
- Can you name the files, channels, providers, and keys it can touch?
- Can you detect a bad run before it affects users?
- Can you disable the workflow without deleting unrelated state?
Template shape
Every workflow page should keep the trigger, inputs, boundary, output, and stop rule visible near the top.
Workflow: daily briefing
Trigger: manual first, scheduled later
Inputs: selected URLs or repositories
Boundary: Docker backend, read-only source access
Output: Markdown report
Stop rule: no outbound messages until reviewedWorkflow quality gate
| Gate | Pass condition | Failure response |
|---|---|---|
| Manual run | The workflow succeeds once without automation. | Do not schedule it yet. |
| Boundary | Files, env vars, providers, and channels are named. | Move to Docker or narrow mounts before retrying. |
| Review step | A human can inspect output before external action. | Remove outbound actions or add an approval point. |
| Stop switch | The workflow can be disabled without deleting unrelated state. | Do not connect shared tokens or production cron. |
Workflow pages to publish next
Cron jobs and daily briefings
A narrow scheduled workflow with explicit inputs, reviewable output, and a stop switch.
Telegram/Discord workflow
A messaging workflow with allowlists, pairing, token scope, and cutover safety.
Google Workspace workflow
A Gmail, Drive, Calendar, or Docs workflow with account boundaries and human approval.
Obsidian workflow
A second-brain workflow with vault boundaries and write-review rules.
Founders and small teams
A cautious use-case map for small teams before turning workflows into external automation.
Agent Guide judgment
A workflow is not ready because Hermes can perform the task once. It is ready when the trigger, input boundary, output destination, review step, cost limit, failure response, and stop switch are all explicit.
The best first workflows are internal and draft-first: research briefs, issue summaries, PR review notes, meeting prep, and private reports. External messages, customer support, legal language, and financial actions should stay human-approved.
Workflow smoke test
- Run the workflow manually with harmless input.
- Inspect the output before it reaches another person or system.
- Record provider, cost, files touched, and whether any tool ran.
- Add scheduling only after the manual run passes twice with predictable output.
Workflow feature signals
| Official surface | Useful workflow | Guardrail |
|---|---|---|
| Messaging gateway | Ask for a private daily brief from Telegram or Slack. | Allowlisted users and draft outputs. |
| Cron | Scheduled research digest or issue monitor. | Cost cap, logs, private destination, disable command. |
| Skills | Reusable PR review, documentation, or research routine. | Skill audit before production workspace. |
| Tool gateway | Web search, extraction, browser, image, or TTS tasks without many provider accounts. | Know which external tool route is active and what data is sent. |
Official sources reviewed
| Source | Used for | Last checked | Confidence |
|---|---|---|---|
| Hermes Agent documentation | Hermes Agent feature scope, documentation structure, and official source navigation. | 2026-06-05 | high |
| MCP with Hermes Agent | MCP tool-surface, integration, and skill workflow safety context. | 2026-06-05 | high |
| Bundled skills catalog | Skill discovery, reuse, and audit guidance. | 2026-06-05 | high |
| Hermes Agent tool gateway docs | Tool gateway routing, cloud browser/search/image/TTS surface, and setup-order caution. | 2026-06-05 | high |
| Hermes Agent messaging docs | Gateway-supported messaging surfaces, 20+ platform context, setup boundaries, and chat-platform workflow risk. | 2026-06-05 | high |
| Public Hermes Agent cron issue reports | Publicly reported cron friction patterns; not treated as official product status. | 2026-06-05 | medium |
Known caveats: Hermes Agent is moving quickly. Treat commands and support status as current only as of the verification date, then check the official docs before changing production systems.
FAQ
Operator checklist
Get the Agent Guide launch checklist
Receive the smoke-test order for install path, sandbox boundary, provider setup, source review, and production checks.