ClawHub
Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Soul Archive

v2.2.8

Soul Archive -- A digital personality persistence system that builds your digital soul clone through everyday AI conversations (with user consent, auto-extra...

⭐ 2· 149·1 current·1 all-time
by@dqsjqian
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
The skill's name/description (digital personality persistence) aligns with what is present: Python scripts for extraction, chat prompt construction, reporting, and an optional crypto module. It reads/writes a dedicated ~/.skills_data/soul-archive/ directory and offers AES-256-GCM protection via an optional cryptography dependency — these are appropriate for the declared purpose.
ℹ
Instruction Scope
SKILL.md and the scripts instruct the agent to read conversation content, run the provided Python scripts, and read/write many personal-data JSON files under the user's home directory. That scope matches the feature set. Important operational note: the skill constructs System Prompts from archived personal data (soul_chat) — whether those prompts are sent to an external LLM is determined by the agent/platform. If your agent forwards prompts to cloud LLMs, this skill will cause sensitive local data to be included in outbound requests.
✓
Install Mechanism
There is no remote install/download: this is an instruction-only skill with bundled Python scripts. No network downloads or arbitrary installers are present in the provided manifest. The only external dependency is the optional 'cryptography' package for encryption, which is a standard Python package.
✓
Credentials
The skill declares no required environment variables; it documents an optional SOUL_PASSWORD environment variable for encrypted mode. That matches provided code (SoulCrypto supports password from env). There are no unrelated credentials requested. The amount of local filesystem access (home directory) is proportional to the purpose but is privacy-sensitive by design.
✓
Persistence & Privilege
The skill does not request always:true or any special platform-level persistence. It stores data under the user's home directory (~/.skills_data/soul-archive/) and manages its own files only. This is expected for a local archive tool; however, because data resides in the home folder it is accessible to other local processes and development tools.
Assessment
This skill appears to do what it says: extract and store multi-dimensional personal data locally and optionally encrypt it. Key things to consider before installing: - Encryption: enable AES-256-GCM protection (follow instructions) if you will store sensitive identity, relationship, health, finance, or emotional data. If enabled, keep the access key safe—losing it means permanent data loss. - Data exfiltration risk: the skill builds System Prompts from archived data. If your agent is configured to send prompts to cloud LLMs or external services, those prompts may include your archived personal data. Verify your agent/platform does not forward sensitive prompts before enabling auto-extraction or running Soul Chat. - Local exposure: data is stored under ~/.skills_data/soul-archive/ (Path.home()). Any other process or user account on the same machine can potentially access those files unless you use OS-level protections or encryption. Do not store the HTML report in the data directory (README warns against this). - SOUL_PASSWORD handling: the skill permits providing the encryption key via the SOUL_PASSWORD env var or --access-key; avoid putting long-lived secrets into shell history or shared environment variables. Prefer interactive entry when possible. - Dependency: the cryptography package is required only for encryption features; ensure you install it from a trusted source (pip from PyPI) if you enable protection. If you want stronger guarantees, ask the skill author to (1) document exact places where prompt data is constructed/sent, (2) provide an option to redact or limit which fields are included in live prompts, and (3) implement an explicit offline mode that refuses any outbound network activity. Otherwise the skill is coherent with its stated purpose but carries predictable privacy risks you should manage.

Like a lobster shell, security has layers — review code before you run it.

latestvk970k83m1022qpb3pxs1f5wssh844hnd

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Comments