Tophant Clawvault Operator

Operate ClawVault services, configuration, vault presets, and scanning from OpenClaw

This skill appears coherent for operating a local ClawVault installation, but consider the following before installing: - Prerequisites: confirm ClawVault was installed by the corresponding installer skill and that the installer's venv (the SKILL.md / SECURITY.md reference) exists as expected. The operator calls the 'claw_vault' module and relies on that installation. - File reads: the 'scan-file' command will read any path you pass it. Do not scan sensitive files unless you intend to expose their contents to the skill's runtime environment. - Process control: the skill runs subprocesses (pgrep, start via Python -m claw_vault, os.kill). Review platform policies if you need strict process isolation. - Local-only network: the skill claims to only contact 127.0.0.1; verify that the dashboard binds to localhost if you need to avoid remote access. - Review upstream packages: because the operator invokes the installed 'claw_vault' package, verify the integrity/source of that package (installed venv) to avoid running malicious code from upstream packages. - If you require strong isolation, run this skill in a disposable VM/container or audit the code and the installed clawvault package first. I have high confidence the package is internally consistent with its stated purpose. If you want, I can point out the exact lines that read/write the config, run pgrep/kill, or perform the local HTTP calls so you can review them yourself.

Type: OpenClaw Skill Name: tophant-clawvault-operator Version: 0.2.5 The tophant-clawvault-operator skill is a management tool for the ClawVault security service, providing commands to start/stop services, manage local YAML configurations, and scan text or files for sensitive data. The implementation in clawvault_ops.py uses standard Python libraries and subprocess calls to interact with the local claw_vault module and its configuration directory (~/.ClawVault/). All requested permissions (execute_command, read_files, write_files, network) are well-justified and restricted to local operations, with no evidence of data exfiltration, prompt injection, or malicious intent.