hermes-kit
AGENTS.md1 starsChanged 50 days agoMITView on GitHub
Guide for AI coding agents working with or installing from hermes-kit. Hermes Agent plugins, Hermes-oriented skills, and docs (patched Hermes fork, Feishu customizations, memory method). Human overview: README.md. Details: docs/install.md. skills/ and plugins/ are produced by an export script from a private source repo (see ADR 0001). Each export replaces them in one commit. Edits made here directly are lost on the next export. To change a skill or plugin: open an issue or a PR against this repo. The…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- ai-hotspot-digestSkill
- convert-documents-to-markdownSkill
- cursor-cliSkill
- desktop-browser-chromeSkill
- feishu-chat-style-reviewSkill
- feishu-ui-feedbackSkill
- figma-copy-proofreadSkill
- hermes-daily-reviewSkill
- hermes-upstream-checkSkill
- writing-with-cursorSkill
The file
We split this AGENTS.md into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 57
No personal or company data in any file: no real chat or user IDs, emails, hostnames, home-directory paths, or people's names.
- Distinctive: line 64
When you change an English page, update its pair in the same change.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md
Guide for AI coding agents working with or installing from hermes-kit.
What this repo is
Hermes Agent plugins, Hermes-oriented skills, and docs (patched Hermes fork, Feishu customizations, memory method). Human overview: README.md.
Install per host
| Host | Install | Update | Remove |
| Hermes, plugin | hermes plugins install Adonis0123/hermes-kit/plugins/<name> then hermes plugins enable <name> | hermes plugins update <name> | hermes plugins remove <name> |
| Hermes, plugin pinned | hermes plugins install Adonis0123/hermes-kit/plugins/<name> --ref <40-char-sha> | reinstall with --force --ref <sha> | hermes plugins remove <name> |
| Hermes, all skills (tap) | hermes skills tap add Adonis0123/hermes-kit then hermes skills install <name> | hermes skills update | hermes skills tap remove <tap> |
| Hermes, one skill | hermes skills install Adonis0123/hermes-kit/skills/<name> | hermes skills update | hermes skills uninstall <name> |
| Claude Code | /plugin marketplace add Adonis0123/hermes-kit then /plugin install hermes-kit@hermes-kit | /plugin marketplace update hermes-kit | /plugin uninstall hermes-kit@hermes-kit |
| Codex, Cursor, others | npx skills add adonis0123/hermes-kit --skill <name> (--list to list) | rerun the add command | remove the skill folder |
| Patched Hermes fork | see docs/hermes-fork.md | hermes update | see the same doc |
Details: docs/install.md.
Layout
.
├── plugins/<name>/ GENERATED. Hermes plugins: plugin.yaml + __init__.py with register(ctx)
├── skills/<name>/SKILL.md GENERATED. Flat, one level deep (the Hermes tap only scans that)
├── docs/ Hand-written docs, English + .zh-CN.md pairs
│ ├── adr/ Architecture decision records
│ ├── assets/ README diagrams (light/dark, en/zh)
│ ├── templates/ Fictional MEMORY/USER/SOUL examples
│ └── <guide>.md GENERATED guides (listed in .export-manifest.json)
├── .claude-plugin/marketplace.json
├── .gitleaks.toml Generic privacy rules for CI (no personal words)
├── .privacy-allow Known false positives for the pre-export scan
├── scripts/check-doc-pairs.mjs
├── CONTEXT.md Glossary (EN + 中文 in one file)
├── AGENTS.md / CLAUDE.md / llms.txt
└── LICENSEGenerated content: do not hand-edit
skills/ and plugins/ are produced by an export script from a private source repo (see ADR 0001). Each export replaces them in one commit. Edits made here directly are lost on the next export.
To change a skill or plugin: open an issue or a PR against this repo. The maintainer applies the change in the source and re-exports.
The guides under docs/ that appear in .export-manifest.json are generated the same way. All other docs, CI, and root files are hand-written and can be edited here.
Naming rules
- Skill directory name =
nameinSKILL.mdfrontmatter: lowercasea-z0-9-, at most 64 characters. SKILL.mdfrontmatter:name,description(at most 1024 characters), optionallicense: MIT.- Plugin directory name =
nameinplugin.yaml.plugin.yamlhasname,version,description. - A skill may say "if
references/local-*.mdexists, read it".local-*files are never published; users create their own. - No personal or company data in any file: no real chat or user IDs, emails, hostnames, home-directory paths, or people's names.
Bilingual docs
- English is authoritative. Every
README.mdanddocs/**/*.mdhas a.zh-CN.mdpair next to it. - First line of each:
English | [简体中文](x.zh-CN.md)or[English](x.md) | 简体中文. - Exempt:
AGENTS.md,CLAUDE.md,CONTEXT.md(bilingual in one file),llms.txt,LICENSE,docs/templates/,skills/,plugins/. - When you change an English page, update its pair in the same change.
Validation
node scripts/check-doc-pairs.mjs # every doc has its .zh-CN.md pair
python3 -m json.tool .claude-plugin/marketplace.json # marketplace JSON is valid
gitleaks detect --source . --config .gitleaks.toml # privacy scan, same as CI
hermes plugins doctor plugins/<name> # plugin contract check (needs Hermes)CI runs the pair check (docs-pair-check.yml) and gitleaks (privacy-scan.yml) on every push and PR.
Docs index
| Doc | Kind | Topic |
| docs/install.md | hand-written | Install paths per host, pinning, updating |
| docs/hermes-fork.md | hand-written | The patched fork and how hermes update follows it |
| docs/feishu-customizations.md | hand-written | Behavior and config keys of each fork patch |
| docs/memory-method.md | hand-written | Write gate for MEMORY.md, USER.md, SOUL.md |
| docs/memory-provider-adoption-gate.md | generated | When to adopt an external memory provider |
| docs/feishu-gateway-blackbox-checklist.md | generated | Six manual Feishu checks with message-id evidence |
| docs/feishu-quiet-group-delivery.md | generated | Quiet group delivery: topics, cards, noise controls |
| docs/hermes-lock-screen-design.md | generated | Design of the hermes-lock-screen plugin |
| docs/cron-digest-pipeline-pattern.md | generated | Daily digest from a cron job |
| docs/upgrade-safe-customization.md | generated | Customizing Hermes so upgrades don't undo it |
| docs/adr/ | hand-written | Architecture decision records |
| CONTEXT.md | hand-written | Glossary |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

