# OpenSOP: Hermes Agent plugin

> Turn your best agent workflow into a file anyone can run.

Canonical: https://hermeswatcher.com/plugins/opensop  
Tier: Community  
Official page (Nous Research): https://hermes-agent.nousresearch.com/docs/plugins/opensop  
Last updated in catalog: 2026-11-17

Turn your best agent workflow into a file anyone can run. An OpenSOP process file sets, step by step, what is a script (must be certain) and what needs the agent (judgment). Hermes starts a run with sop_run, does its steps and submits them with sop_submit; the next script step checks the work, so scripts decide done. The run lives on disk, not in the conversation: Hermes is re-anchored on its open step every turn and across compaction, nudged to finish it before wrapping up, and refused when it tries to decide a step reserved for a person (seat belts, not a sandbox). Any agent or person can pick the run up with the opensop CLI, and every step leaves a receipt. Requires the opensop CLI 0.10.0+ on PATH.

## Good to know before installing

Sop_run, sop_submit and sop_retry go through Hermes's approval prompt (process hash, inputs, run id, run folder) because a run's script steps run as shell commands with your permissions, outside Hermes's terminal approvals; a run this profile started or continued through an approved call does not prompt again. llm and judgment steps can drive coding-agent CLIs (claude, codex, cursor-agent, kimi) with the permission the file declares, so only run process files you trust. The plugin has no credentials of its own. It runs opensop with a scrubbed environment: PATH, HOME, LANG, LC_*, TMPDIR, SSH_AUTH_SOCK, OPENSOP_*, the variables the process file references (env.X / ${env.X}), and ANTHROPIC_API_KEY only when the file has an llm or judgment step on a claude* model (the CLI calls the Anthropic API directly for those). The approval prompt lists which variables will be forwarded. Hermes's other provider keys never reach script steps or the agent CLIs they drive. Run folders are written under OPENSOP_LOCAL_HOME, the nearest .opensop cell or ~/.opensop-local; the plugin keeps a session-to-run map and approved run folders in plugin state. Tool calls are bounded by the plugin's timeout setting (plugins.entries.opensop.settings.timeout, default 1800 s); on timeout the CLI and its child processes are stopped. The plugin makes no network calls, never auto-approves or changes approval settings, starts no Hermes children, runs nothing in the background and has no self-updater.

From the plugin's catalog listing.

## How to install

1. Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.

```bash
hermes plugins install opensop
```

2. Said no? Turn it on later:

```bash
hermes plugins enable opensop
```

3. Restart Hermes so it loads the plugin.

Some plugins also need a sign-in or settings: see the official page. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide: https://hermes-agent.nousresearch.com/docs/user-guide/features/plugins

## Facts

- Category: Automation & multi-agent, Security & privacy
- Maintained by: opensop (Community)
- Needs: No API keys declared
- Adds tools: `sop_run`, `sop_status`, `sop_submit`, `sop_retry`
- Works on: linux, macos
- Requires Hermes: `>=0.23.0`
- Version: 0.1.0
- In the catalog since: 2026-11-17

## Links

- [Full setup and settings (official page)](https://hermes-agent.nousresearch.com/docs/plugins/opensop)
- [Source code](https://github.com/opensop/hermes-opensop#readme)
- [All plugins](https://hermeswatcher.com/plugins.md)

---

HermesWatcher is an independent guide. It is not affiliated with or endorsed by Nous Research. The [official Hermes Agent documentation](https://hermes-agent.nousresearch.com/docs) and [repository](https://github.com/NousResearch/hermes-agent) from Nous Research are the authority; each page here lists the official sources it was written from. If this site and the official docs disagree, trust the official docs. Index for agents: https://hermeswatcher.com/llms.txt
