Workspace Review
ReviewAudited by ClawScan on May 10, 2026.
Overview
Prompt-injection indicators were detected in the submitted artifacts (system-prompt-override); human review is required before treating this skill as clean.
This skill is reasonable to use for a private OpenClaw workspace audit. Before installing or invoking it, understand that it may read and summarize profile, memory, and git-status information; do not share the resulting report publicly, and run the helper only against the intended workspace. ClawScan detected prompt-injection indicators (system-prompt-override), so this skill requires review even though the model response was benign.
Findings (3)
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.
If taken out of context, prompt-like reference text could influence an agent, but here it is clearly framed as documentation.
This prompt-control language triggered the static scan, but it appears in a documentation section explaining OpenClaw's automatic memory-flush feature.
Agent receives system prompt: "Session nearing compaction. Store durable memories now."
Treat these lines as reference documentation only; actual system and user instructions should remain authoritative.
Running the helper can reveal workspace file names, git status, and files that may contain key-like strings to the agent/session output.
The helper script can run local filesystem, git, find, wc, and grep checks. By default it is scoped to the OpenClaw workspace and only reports locally, but an explicit path argument could make it scan a different directory.
WORKSPACE="${1:-$HOME/.openclaw/workspace}" ... grep -r -l "sk-" --include="*.md" .Run it only on the intended workspace and review output before sharing it outside a private context.
Private memories, preferences, and workspace notes may be read by the agent and summarized in the review output.
The review explicitly includes user profile files, long-term memory, daily logs, and vector-indexed reference material. This is central to the skill's purpose, but it involves sensitive persistent context.
USER.md ← User profile ... MEMORY.md ← Curated long-term memory ... memory/ ← Daily logs + reference docs (vector-indexed)
Use the skill in a private session, avoid storing secrets in memory files, and be cautious before enabling or expanding memory indexing.