Knowledge Management

WarnAudited by ClawScan on May 10, 2026.

Overview

The skill is mostly a coherent local memory organizer, but it contains an unsafe eval that could run code from crafted local markdown/tag data.

Install only if you trust the local `km` binary and need this memory organization workflow. Do not process untrusted knowledge folders until the `eval` parsing is replaced, use `--dry_run` before cleanup, and enable cron only if you want ongoing automatic syncs.

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

A malicious local knowledge file could potentially run code on the user's machine when the skill processes tags.

Why it was flagged

The skill uses JavaScript `eval` to parse tag text. If the matched tag data is influenced by a crafted or corrupted local markdown file, running the relevant command could execute JavaScript with the user's local permissions.

Skill content
const tagsArray = eval(`[${tagsMatch[1]}]`); // Safe since we control the format
Recommendation

Replace `eval` with strict JSON/YAML parsing and validate that tags are simple strings before processing. Avoid running the skill on untrusted output directories until fixed.

What this means

Using cleanup on the wrong output directory or state file could remove local knowledge files the user expected to keep.

Why it was flagged

Cleanup is a disclosed, purpose-aligned destructive operation. The dry-run option helps, but users should understand it can remove files from managed folders.

Skill content
`--cleanup` — Delete orphan files (files in folders but not tracked in state) ... `km cleanup --dry_run`
Recommendation

Run `km cleanup --dry_run` first and keep the output directory scoped to the intended knowledge-management folder.

What this means

A different `km` binary on the system could be invoked instead of this skill's reviewed code.

Why it was flagged

The skill depends on a `km` executable, but the registry does not define an install mechanism. Users need to ensure the command they run is the reviewed local package, not another binary on PATH.

Skill content
Required binaries (all must exist): km ... Install specifications: No install spec — this is an instruction-only skill.
Recommendation

Install or link the CLI from the reviewed package path, verify `which km`, and avoid relying on an unknown pre-existing binary.

What this means

Sensitive information in OpenClaw memory may appear in organized folders and index files where future tools or users can read it.

Why it was flagged

The skill persistently copies and indexes memory content. This is central to the purpose, but it may duplicate private or sensitive memory entries into additional files.

Skill content
Automatically parses `MEMORY.md` and daily memory files, classifies entries by content type, and stores each as a timestamped markdown file in the appropriate folder.
Recommendation

Use a private workspace/output directory, review generated files, and avoid syncing memory that contains secrets or highly sensitive notes.

What this means

If scheduled, the skill may continue processing and writing memory-derived files after the initial setup.

Why it was flagged

The cron integration is optional and disclosed, but it creates recurring automated processing of memory files if the user installs it.

Skill content
openclaw cron add ... --name "Daily Knowledge Sync" ... --message "km sync --days_back 7"
Recommendation

Enable cron only if recurring sync is desired, and document how to remove or disable the scheduled job.