Back to skill

Security audit

Clawddocs 1

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation helper skill with mostly inert scripts, but users should treat its copied configuration examples carefully.

Install only if you want Clawdbot documentation shortcuts, and verify answers against the official docs because the bundled scripts are minimal stubs. Do not commit real Discord or Telegram tokens, protect WhatsApp session folders, and change gateway examples to localhost or add authentication, TLS, and firewall restrictions before exposing a gateway on a network.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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)

T09 · Insecure Skill Coding Practices

Warning
Location
snippets/common-configs.md:37
Finding
Gateway Configuration Binds to All Network Interfaces## Vulnerability Details **File Location**: `snippets/common-configs.md:37-45` **Vulnerability Type**: Insecure network exposure **Risk Level**: Medium ### Vulnerable Code ```json { "gateway": { "host": "0.0.0.0", "port": 8080 } } ``` ### Technical Analysis The ready-to-use configuration binds the gateway to `0.0.0.0`, causing it to listen on every available IPv4 network interface. The example does not include authentication, TLS, firewall restrictions, or an explicit warning about remote exposure. Binding to all interfaces is not inherently vulnerable when appropriate controls exist, but presenting it as a default configuration without corresponding safeguards creates an unsafe deployment pattern. The actual exploitability depends on the gateway's authentication controls and the host's firewall and network configuration. ### Attack Path 1. A user copies the documented gateway configuration. 2. The gateway starts listening on port `8080` on all IPv4 interfaces. 3. The host firewall or surrounding network allows access to that port. 4. An attacker on a reachable local or external network discovers the service. 5. The attacker connects to the exposed gateway and attempts to access any endpoints not protected by adequate authentication or authorization. ### Impact Assessment The maximum impact depends on the gateway capabilities and its independent security controls. If sensitive endpoints are insufficiently protected, a remote attacker could access gateway data or operations available to the service. The configuration itself does not grant operating-system privileges, and the reviewed files do not establish that authentication is absent; therefore, compromise is conditional rather than guaranteed.
Remediation
## Remediation Suggestions - Change the default bind address to `127.0.0.1` so the gateway is locally accessible only. - Require users to opt in explicitly when remote access is necessary. - Add authentication and authorization requirements to any remote-access example. - Terminate TLS directly at the service or through a securely configured reverse proxy. - Restrict inbound access using host and network firewalls, preferably to an explicit allowlist. - Add a prominent warning explaining that `0.0.0.0` exposes the service on all IPv4 interfaces. - Document verification steps, such as inspecting listening sockets and testing access from an untrusted network segment.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The declared description presents a broad documentation expert with multiple capabilities, but the supplied code only echoes a URL based on a provided path. Its actual behavior is limited and does not perform real document retrieval, searching, navigation, version handling, or snippet generation. This is a material mismatch in primary functionality and scope.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding
The declared description promises a comprehensive Clawdbot documentation expert with multiple capabilities, but the supplied code chunk is only a minimal shell script related to tracking changes. Even though version tracking is one declared area, the actual code is far narrower than the stated primary purpose and lacks the other core described capabilities. This is a material description-to-behavior mismatch.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The instruction 'Use this skill to help users navigate, understand, and configure Clawdbot' after 'You are an expert on Clawdbot documentation' implies activation whenever a user asks about Clawdbot, but it does not define boundaries for when the skill should not activate. Because the file lacks exclusion conditions or negative examples, ordinary conversation mentioning Clawdbot could be interpreted as invoking the skill.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
This markdown file includes configuration snippets for Discord and Telegram tokens, but it does not warn users that these values are sensitive credentials or advise them not to commit, share, or expose them. Under the markdown-specific warning criterion, credential-handling behavior that affects privacy or system integrity should be disclosed.

Missing User Warnings

Medium
Confidence
80% confidence
Finding
The WhatsApp example specifies a local sessionPath, which implies storage of session material on disk, but the markdown does not mention that this directory may contain sensitive authentication/session data. For markdown files, behaviors that affect privacy or local system integrity should include user-facing warnings.

Static analysis

No suspicious patterns detected.