Scribe

WarnAudited by ClawScan on May 10, 2026.

Overview

Scribe is a coherent local note-taking skill, but it broadly collects private chats, logs, preferences, drafts, and memory into persistent notes with limited scoping controls.

Install only if you explicitly want a local journal of broad OpenClaw and Cursor activity. Before running it, review which files it will scan, consider disabling Cursor history or draft scanning, and avoid scheduled runs until you are comfortable with where the generated notes are stored and how long they will be kept.

Findings (5)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Private chats, preferences, personal memory files, and draft writing may be copied into new note files where they can be retained or later read by other tools or agents.

Why it was flagged

The skill is designed to aggregate highly personal local context into persistent notes. The artifacts do not show category opt-outs, redaction, retention controls, or safeguards for how these notes may be reused later.

Skill content
Scribe automatically scans and analyzes: ... **Chat history** ... **Memory files** ... **Behavior files** - BEHAVIOR.md, DESIRES.md, TASTES.md, PREFERENCES.md ... **Drafts** ... Saves to Notes
Recommendation

Only run this on data you are comfortable journaling. Add explicit include/exclude controls, redaction for secrets, and retention guidance before using it routinely.

What this means

A daily or weekly run may include older log entries than expected, potentially copying more historical or sensitive activity into notes.

Why it was flagged

The daily/weekly modes imply time-bounded scanning, and the function computes a cutoff, but the shown log loop appends matching log entries without enforcing that cutoff.

Skill content
cutoff_time = datetime.now() - timedelta(days=days_back)

for log_file in self.logs_dir.glob("*.log"):
Recommendation

Enforce the cutoff before adding log entries, or clearly document that log scanning is not time-limited.

What this means

Cursor chats from unrelated workspaces may be included in generated notes, exposing more local history than a user may expect for an OpenClaw summary.

Why it was flagged

The code reads Cursor's user-level global and workspace storage databases rather than a narrowly selected project or conversation.

Skill content
self.cursor_storage = Path.home() / "Library" / "Application Support" / "Cursor" / "User"
self.global_db = self.cursor_storage / "globalStorage" / "state.vscdb"
self.workspace_storage = self.cursor_storage / "workspaceStorage"
Recommendation

Ask the user to select specific Cursor workspaces or add a flag to disable Cursor history scanning by default.

What this means

If enabled, the skill can keep producing private activity summaries every day without a fresh manual review each time.

Why it was flagged

The documented cron integration is user-directed, but it enables recurring autonomous collection of sensitive local activity into notes.

Skill content
### Cron Job Integration

Add to your OpenClaw cron jobs: ... "message": "Run scribe.py --mode daily" ... "cron": "0 0 * * *"
Recommendation

Use scheduled runs only after reviewing the data sources and output location, and periodically audit or delete generated notes.

What this means

Users have less information for verifying who maintains the skill or where updates come from.

Why it was flagged

The artifacts do not provide a verifiable source or homepage for provenance, though the included script is visible in the supplied review material.

Skill content
Source: unknown
Homepage: none
Recommendation

Prefer a verified repository/homepage and review the included script before installing or updating.