OpenClaw Agent Orchestrator

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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 malformed or malicious agent ID could cause deletion or rewriting of unintended local directories, especially if path traversal such as '../' is accepted before the OpenClaw CLI validates the ID.

Why it was flagged

The user-provided agent ID is embedded directly into a filesystem path, then the target directory is recursively deleted if it exists. The script does not validate the ID or prove the resolved path remains inside the intended workspaces directory.

Skill content
TARGET_WORKSPACE="${OPENCLAW_HOME}/workspaces/${AGENT_ID}" ... if dst.exists():
    shutil.rmtree(dst)
Recommendation

Validate agent IDs against a safe pattern, resolve and check that the target path stays under the intended OpenClaw workspaces directory, and require confirmation or backups before recursive deletion.

What this means

Private workspace files, saved prompts, instructions, or other runtime state could be duplicated into a durable subagent and reused later without the user realizing exactly what was copied.

Why it was flagged

By default the script recursively copies almost all files from the main OpenClaw workspace into the new durable agent workspace, with only a small skip list.

Skill content
SOURCE_WORKSPACE="${5:-$OPENCLAW_HOME/workspace}" ... for current_root, dirnames, filenames in os.walk(src, topdown=True ... shutil.copy2(source_file, target_file)
Recommendation

Document what is copied, add exclusions for secrets and sensitive files such as .env, keys, logs, and credentials, and ask the user to choose or confirm the source workspace before copying.

NoteHigh Confidence
ASI10: Rogue Agents
What this means

Created agents and bindings may remain available after the current task and could affect future OpenClaw behavior.

Why it was flagged

The skill intentionally creates durable OpenClaw agent entries. This is disclosed and purpose-aligned, but it is persistent runtime state rather than a temporary chat-only action.

Skill content
For durable subagents in this deployment, use `openclaw agents add` and verify the result.
Recommendation

Use it only when durable agents are actually desired, verify created agents and bindings, and provide cleanup instructions for agents that are no longer needed.

What this means

The skill may fail or behave differently depending on the local environment, and users may not realize it requires local CLI execution.

Why it was flagged

The metadata does not declare required binaries, while the included scripts depend on bash, python3, and the openclaw CLI.

Skill content
Required binaries (all must exist): none ... No install spec — this is an instruction-only skill.
Recommendation

Declare required binaries and runtime assumptions in metadata, especially the openclaw CLI and python3.