Create or improve SOUL.md files for OpenClaw agents through guided conversation. Use when designing agent personality, crafting a soul, or saying "help me create a soul". Supports self-improvement.
Security Analysis
medium confidenceSoulCraft is an instruction-only persona-design helper that reads and edits SOUL.md/IDENTITY.md in the workspace — its declared capabilities and resource needs are generally coherent, with a small documentation mismatch about workspace access.
The name/description (create or improve SOUL.md files) match the SKILL.md and README: guided conversation, question bank, and example SOUL.md files. No unexpected credentials, binaries, or external services are requested.
Runtime instructions explicitly tell the agent to read existing SOUL.md and IDENTITY.md and to write/edit SOUL.md; they also recommend offering to commit to git if the workspace is a repo. This file I/O and optional git activity are consistent with the skill's purpose, but users should be aware the skill expects workspace read/write access.
No install spec or code files are included; the skill is instruction-only, which minimizes install risk. Nothing is downloaded or executed from external URLs.
No environment variables, credentials, or config paths are requested. The README mentions 'OpenClaw agent with workspace access' (implying filesystem access), which is proportionate to the declared purpose of editing soul files.
always is false and disable-model-invocation is default (agent may invoke autonomously). The skill does not request persistent system-wide privileges nor modify other skills' configs. It does suggest committing changes to git in the workspace, which is a normal action for a file-editing skill.
Guidance
This skill is coherent for crafting SOUL.md files, but it expects read/write access to the agent's workspace and may offer to commit changes to git. Before enabling it: (1) review SKILL.md and the included examples to ensure the editing behavior matches your expectations; (2) ensure your workspace (repo) is backed up or under version control so edits can be reviewed/reverted; (3) grant workspace access only to agents/skills you trust; and (4) if you need stricter controls, require the skill to propose edits first and get explicit approval before writing or committing changes. If the skill ever asked for unrelated credentials, network endpoints, or to edit files outside the agent's workspace, that would be a red flag.
Latest Release
v1.0.0
Initial release: guided SOUL.md creation, improvement mode, self-reflection mode, IDENTITY.md integration
More by @kesslerio
Published by @kesslerio on ClawHub
