Back to skill
Skillv0.1.1
ClawScan security
Crm · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 2, 2026, 2:08 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- Files and runtime instructions implement a local, markdown-based personal CRM and the required actions (file I/O, parsing CSV/vCard, indexing, reminders) are coherent with the described purpose; there are only minor documentation/path inconsistencies to fix before use.
- Guidance
- This skill appears to do what it claims: a local markdown-based CRM implemented by included Python scripts. Before installing, do the following: 1) Review and fix the documented paths in SKILL.md (there are copy-paste duplications like memory/contacts/memory/contacts/ and inconsistent Heartbeat paths). 2) Place the scripts under memory/contacts/scripts (as the setup describes) so their relative CONTACTS_DIR resolves correctly. 3) When importing data, use --dry-run first to confirm results. 4) Only enable the Heartbeat/periodic checks if you want automated runs — otherwise run followups/remind manually. 5) Ensure memory/contacts isn't unintentionally synced to remote/shared storage (these files contain personal contact data). 6) Verify the manifest/signature if you need cryptographic assurance (README provides steps). If you want, run the tools in a sandbox or test workspace first.
Review Dimensions
- Purpose & Capability
- okThe skill is a markdown-based personal CRM and the included Python scripts implement adding, querying, importing/exporting, indexing, follow-ups and reminders via local file operations. There are no requests for unrelated credentials, network endpoints, or system-level privileges—capabilities requested match the stated purpose.
- Instruction Scope
- noteSKILL.md instructs copying templates and scripts into memory/contacts and running the CLI tools. The scripts operate on local files under CONTACTS_DIR (derived from __file__), which is consistent with the copy instructions. However the documentation has some copy-paste/path errors (e.g., 'memory/contacts/memory/contacts/scripts/...' duplication and inconsistent example paths in Heartbeat lines). Heartbeat instructions propose periodic execution of followup/reminder scripts — that is expected for reminder functionality but you should explicitly opt-in and confirm the paths are correct before enabling automation.
- Install Mechanism
- okThis is an instruction-only skill (no install spec). Code files are included in the package; there are no external downloads or extracted archives referenced. No package manager installs or remote code pulls are performed by the skill itself.
- Credentials
- okThe skill declares no required environment variables or external credentials. Scripts only read and write local markdown and index files and use PyYAML (yaml). The export tool validates output paths to prevent arbitrary workspace escapes (allows workspace, /tmp, or home), which is reasonable for export features. No secret exfiltration or unrelated env access is present.
- Persistence & Privilege
- notealways:false and no install-time persistence is requested. The skill recommends adding checks to HEARTBEAT.md for periodic runs (1–2x daily), which would cause automated executions if the user opts in. Autonomous invocation by the model is the platform default; combined with Heartbeat-style automation this means scripts could run on a schedule if you enable that — confirm you want periodic checks before adding them.