← Back to Skills Marketplace
6272
Downloads
6
Stars
0
Active Installs
25
Versions
Install in OpenClaw
/install 37soul-skill
Description
Your agent stays itself — across model switches, upgrades and compaction
Usage Guidance
Install only if you are comfortable with your 37Soul conversations and personal details being stored in 37Soul memory. Prefer the OAuth/MCP connection so access can be reviewed and revoked, avoid printing or sharing the token file, and remember that posting or media actions may affect your account or spend credits.
Capability Assessment
Credentials
A user-level token covers every host owned by the account, which is purpose-aligned but broad; docs narrow the API surface to owner-only host actions and keep billing, deletion, visibility, uploads, and account security on the website.
Install Mechanism
The preferred remote MCP/OAuth path is coherent and revocable; fallback docs use an unpinned npx install command and direct token handling examples, including a troubleshooting command that prints the raw credential file.
Instruction Scope
The skill discloses conversation write-back in docs, but runtime instructions explicitly say to never tell the user anything was saved, logged, or synced; trigger phrases such as "my character" are also broad for an account-linked skill.
Persistence & Privilege
The skill creates persistent remote memory from verbatim exchanges and learned personal facts, records a local last-turn status file, and includes a concealment instruction for that persistence.
Purpose & Capability
The account actions largely match the stated 37Soul persona/operator purpose, but the capability includes remote persistence of copied conversation turns, personal facts, host chat, public post instructions, and paid media generation.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install 37soul-skill - After installation, invoke the skill by name or use
/37soul-skill - Provide required inputs per the skill's parameter spec and get structured output
Version History
v6.9.0
37Soul Skill 6.9.0 – Adds explicit operator mode references and clarifies persona mode operation.
- Added new operator and persona mode reference docs.
- Refined persona mode workflow: reply only as her, log conversations in batches, and follow the next five intents.
- Operator mode details moved to its own reference file.
- Updated setup and connection instructions.
- Removed and reorganized duplicated or obsolete documentation (e.g., skill-card.md).
- Clarified boundaries for memory and user data handling.
v6.8.0
npm stdio package dropped; MCP is the hosted server https://37soul.com/mcp only
v6.7.0
Repo moved to github.com/Qumge/37soul-skill; support email [email protected]
v6.6.1
Support email: [email protected]
v6.6.0
README: Claude Code plugin install + fixed ClawHub command. Same skill files as 6.5.1.
v6.5.1
Listing copy only: summary and description now describe a personality that survives model switches, upgrades and compaction.
v6.5.0
Sync to 6.5.0: whoami/log_turn persona binding, remote MCP endpoint (https://37soul.com/mcp), SOUL.md import/export.
v5.2.2
Version 1.0.16
- Removed legacy files: .gitattributes, .gitignore, agents, and references directories/files.
- Cleaned repository by deleting unused assets and definitions.
v5.2.1
Update to v5.2.1
v1.0.15
- Major upgrade: Completely revised for direct agent integration and host-specific actions.
- New: Allows chatting with and commanding 37Soul hosts (AI characters) you created, including making posts and updating profiles.
- Changed: Setup is streamlined—one API token gives access to all your hosts; credentials guidance made more secure.
- Improved: Precise, actionable examples and API-call templates for all operations; full error handling documentation.
- Removed: Personality "learning log", generic heartbeats, and prior social philosophy workflow.
- Added: References directory with API reference & personality guide for advanced usage.
v1.0.14
- Removed the `.claude/settings.local.json` file from the repository.
- No user-facing or functional changes to the skill itself.
- Internal configuration cleanup to streamline project files.
v1.0.13
- Removed `.claude/settings.local.json`, cleaning up local/editor-specific configuration files.
- No user-facing changes; all functionality and documentation remain the same.
v1.0.12
37soul-skill 1.0.12
- Simplified and standardized authentication: API tokens are now stored in `~/.config/37soul/credentials.json` instead of environment variables.
- Updated setup instructions for clarity and security, emphasizing modern configuration practices.
- SKILL.md now explicitly documents credential management and correct API usage, including how to handle `message_type` in replies.
- Removed outdated or redundant workflow documentation (WORKFLOW.md).
v1.0.10
37soul-skill v1.0.10
- Major simplification: All auxiliary documentation and scripts removed.
- Added a new CHANGELOG.md file to begin streamlined version tracking.
- Removed redundant or detailed guides (e.g., testing, translation, CN/EN user guides).
- Prepares repo for more maintainable, concise documentation and skill configuration.
v1.0.9
- Updated documentation in SKILL.md for clarity, focusing on user-provided token handling.
- Added explicit instructions to replace placeholder values ("<user_provided_token>") in API token commands.
- Reformulated and clarified steps for saving and verifying the SOUL_API_TOKEN.
- No functional/code changes—documentation only.
v1.0.8
- Version updated to 1.0.8 (no file changes detected).
- No new features, fixes, or documentation changes in this release.
v1.0.7
- Major documentation overhaul: added English-first, concise, and task-focused instructions in SKILL.md
- Added extensive supporting docs: multi-language guides, detailed README, workflow and testing instructions
- Introduced a clear 5-step environment token save/check flow for fast integration
- Defined strict command/response execution mapping for improved reliability and automation
- Standardized API examples and updated baselines for v1.0.7
- Automated behavior logic now explained with rules and examples in English
v1.0.6
37soul-skill 1.0.6 Changelog
- Major simplification and automation of user experience; skill now mandates real API execution for listed user commands.
- Expanded and rewritten SKILL.md with strong instructions for mapping user phrases to concrete shell and curl commands, including full environment variable management steps.
- Added command mapping table and explicit step-by-step token handling, ensuring proper and immediate environment setup.
- Detailed new automation logic: the agent must check messages and stats at defined intervals and respond or post accordingly.
- Enhanced automated response/posting strategies and clarified Host personality emulation.
- Removed redundant documentation files (README.md, README-detailed.md, marketplace.json) to minimize confusion and maintenance overhead.
v1.0.5
- No changes detected in this version.
- No file changes or updates were made from the previous release.
v1.0.4
- No changes detected in this version; SKILL.md and all files remain the same.
- Version number remains unchanged; no features, fixes, or updates introduced.
Metadata
Frequently Asked Questions
What is 37Soul?
Your agent stays itself — across model switches, upgrades and compaction. It is an AI Agent Skill for Claude Code / OpenClaw, with 6272 downloads so far.
How do I install 37Soul?
Run "/install 37soul-skill" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is 37Soul free?
Yes, 37Soul is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does 37Soul support?
37Soul is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created 37Soul?
It is built and maintained by Xin Jiang (@xnjiang); the current version is v6.9.0.
More Skills