Back to skill

Security audit

Super Agent Knowledge

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent knowledge-management helper, but it asks users to install and run an unpinned separate package and recommends recurring auto-fix file changes.

Review the actual `knowledge` package and `scripts/know` implementation before installing or scheduling it. Keep `KNOWLEDGE_DIR` limited to a dedicated knowledge folder, run maintenance in audit/dry-run mode first if available, and avoid cron or heartbeat auto-fix until version provenance and backup behavior are clear.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:11
Finding

Unpinned and Mismatched External Skill Dependency

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 11-18
Vulnerability Type: Unverified external dependency installation
Risk Level: Medium

The audited artifact identifies itself as super-agent-knowledge, but its installation instructions direct users to install a separately named and unpinned Skill, knowledge:

bash
clawhub install knowledge

This installs scripts/know — add to PATH or use full path:

bash
~/.openclaw/skills/knowledge/scripts/know

Technical Analysis

The implementation of scripts/know is not included in the audited artifact. Consequently, its behavior, permissions, integrity, and security properties cannot be verified from the supplied project files.

The dependency is referenced only by the name knowledge, with no immutable version, integrity digest, repository identity, or trusted registry declaration. This name also differs from the artifact slug super-agent-knowledge, as declared in both SKILL.md and _meta.json. This creates a supply-chain risk in which installation could resolve to an unintended, replaced, or malicious package. Because no version or digest is pinned, the effective installed code may also change after this artifact has been reviewed.

No evidence establishes that the external package is currently malicious. The confirmed weakness is that the artifact asks users to install and execute code outside the audit boundary without sufficient identity or integrity controls.

Attack Path

  1. A user reviews or loads the super-agent-knowledge documentation.
  2. The user follows the documented command clawhub install knowledge.
  3. The package manager resolves the separately named and unpinned knowledge package from its configured registry.
  4. An attacker who controls, replaces, or successfully imitates that package supplies a malicious scripts/know executable.
  5. The user invokes know, or schedules the documented `know tidy --fix ...[truncated 872 chars]
Remediation
View remediation

Remediation Suggestions

  1. Include the complete scripts/know implementation within the audited artifact so its behavior is covered by the same review.
  2. Make the installed package name match the declared artifact slug, or explicitly document and justify the relationship between super-agent-knowledge and knowledge.
  3. Pin the dependency to an immutable, reviewed version rather than resolving the latest package by name.
  4. Require an integrity checksum, signed release, or equivalent provenance verification before installation.
  5. Specify the trusted registry or canonical repository from which the dependency must be obtained.
  6. Review the executable's filesystem access, command execution, network behavior, and input handling before recommending scheduled execution.
  7. Avoid adding the tool to cron until the exact executable version and integrity have been verified.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill recommends know tidy --fix for heartbeats or nightly cron and describes auto-fix behavior that can normalize tags, move files, remove empty content, and rebuild indexes without any explicit warning about modifying user data. In an agent context, this increases the risk of unintended file changes or deletion, especially if the configured knowledge directory is broader than expected or the command is run automatically without review.

Content

No source excerpt is available for this finding.