Project Memory - 项目记忆管理
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The skill generally matches its stated purpose (a local project-memory manager) but has multiple mismatches and a serious safety issue: it writes/deletes files under the user's home directory and does not sanitize project names, which could allow path-traversal deletion of arbitrary paths.
This skill appears to do what it claims (local project-based memory stored under ~/.openclaw/projects), but do not install or enable it without addressing the filesystem-safety issues. Specific things to consider before installing: - Path traversal / destructive operations: The included code uses project names directly in paths and performs recursive deletions and moves. A project name containing ../ or other special characters could cause the skill to operate on (and delete) files outside ~/.openclaw/projects. Treat this as high-risk for data loss. - Undeclared env var: The code reads OPENCLAW_AGENT_ID to mark who created an entry; this env var is not declared in skill metadata. It’s not critical but should be documented or removed. - Documentation vs implementation mismatch: SKILL.md mentions automatic backups and auto-save behaviors that are not implemented in handler.js; don’t rely on backups being made. Recommended actions to mitigate before using: - Review and/or modify the code: validate/sanitize project names (reject path separators, control characters), normalize paths and verify the resolved path starts with the intended PROJECTS_DIR, and avoid allowing names that could escape the base directory. - Replace fs.rmSync calls with safer routines that verify the target path is inside PROJECTS_DIR before recursive deletion; require explicit user confirmation in the UI layer (not just a boolean flag passed programmatically). - Implement (or remove) the advertised automatic backup behavior and update documentation accordingly. - Run the skill as a non-privileged user and make a manual backup of important data first. Consider running inside a sandbox/container if you need to evaluate it. Because of the destructive filesystem risk and documentation mismatches, classify this skill as suspicious until the above issues are fixed.
Static analysis
Static analysis findings are pending for this release.
VirusTotal
No VirusTotal findings for this skill version.
Risk analysis
No visible risk-analysis findings were reported for this release.