Apollo Stem
v2.0.0像干细胞一样自我更新和分化：让AI能够定期更新核心能力，同时保持已有的有效经验。
⭐ 0· 51·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The SKILL.md describes self-update triggers and a rebuild workflow; the included shell script implements checks against the agent workspace (skill files, workflow state, learnings) which is consistent with assessing update readiness and planning 'differentiation'. No unrelated services, binaries, or secrets are requested.
Instruction Scope
Instructions and the script access files under /root/.openclaw/workspace (skills, .workflow/state.json, .learnings/LEARNINGS.md) and will read many SKILL.md files to evaluate status. This is coherent with the skill's purpose, but it does mean the skill will inspect other skills' contents and the workspace; review those files for sensitive data before running in a privileged context.
Install Mechanism
There is no install spec and no external downloads; the only artefact is an included shell script. This minimizes supply-chain risk.
Credentials
The skill requires no environment variables, credentials, or network access. It only reads/writes within the agent workspace, which is proportionate to its stated monitoring purpose.
Persistence & Privilege
always:false and the skill is user-invocable (normal). The script writes its own state file to /root/.openclaw/workspace/.stem/state.json — this is reasonable for its function but does require filesystem write access to the workspace and assumes a /root path; confirm the execution environment and permissions are appropriate.
Assessment
This skill appears coherent and implements a workspace health/check routine without requesting credentials or network access. Before installing, confirm: (1) running agent has permission to read other skills and to write in /root/.openclaw/workspace (or adjust paths to a less privileged workspace); (2) any sensitive data is not present in SKILL.md files or .learnings/LEARNINGS.md (the script will read them); and (3) you are comfortable with the skill creating its own state file under the workspace. If you want tighter boundaries, modify the script to use a non-root workspace path and/or to exclude specific skill directories from inspection.Like a lobster shell, security has layers — review code before you run it.
latestvk97453gxvsh92tnth8eev5dppd84dw9k
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🔬 Clawdis