Back to skill

Security audit

Agentfinder

Security checks for vulnerabilities and agentic risk

Overview

This skill appears locally contained, but it overstates live multi-registry search while actually using a small hardcoded index.

Review this before relying on it for skill selection. It is best treated as a local demo or sample index, not a complete or current registry search engine. Install it only in an isolated Python environment and avoid entering sensitive problem descriptions unless you are comfortable sending them to the local service process.

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:4
Finding

Unpinned Third-Party Python Dependencies

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 4
Vulnerability Type: Unpinned third-party dependencies and non-reproducible installation
Risk Level: Medium

Vulnerable Code

yaml
metadata: {"openclaw":{"emoji":"🔎","requires":{"bins":["python"]},"install":[{"id":"pip","kind":"uv","packages":["fastapi","uvicorn","pydantic"]}]}}

Technical Analysis

The installation metadata requests fastapi, uvicorn, and pydantic without exact version constraints or cryptographic hashes. Consequently, installations performed at different times can resolve to different package versions.

This is a supply-chain weakness because the effective code installed with the Skill is not fully defined by the reviewed project. If a future dependency release is compromised, malicious, or incompatible, the package manager may resolve and install it automatically. Dependency code can execute during installation, import, application startup, or request processing.

The audit found no evidence that the named packages are currently malicious. The risk arises from mutable and unverified dependency resolution.

Attack Path

  1. An attacker compromises a listed package, its maintainer account, release infrastructure, or another dependency in its transitive dependency graph.
  2. The attacker publishes a malicious version that remains eligible because the Skill specifies no version constraints.
  3. A user installs the Skill using the declared uv installation procedure.
  4. The package manager resolves the compromised release and installs it.
  5. Malicious code executes during package installation, module import, Uvicorn startup, or API operation under the privileges of the user running the Skill.

Impact Assessment

Successful exploitation could permit arbitrary code execution with the privileges of the account installing or running the Skill. Depending on that account's permissions and environment, the compromised dependency could access application data ...[truncated 314 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin every direct dependency to an exact, reviewed version, for example:

    yaml
    "packages":["fastapi==<reviewed-version>","uvicorn==<reviewed-version>","pydantic==<reviewed-version>"]
    
  2. Generate and commit a lockfile that records the complete transitive dependency graph.

  3. Require cryptographic hashes for downloaded distributions where supported.

  4. Install only from an explicitly configured trusted package index over TLS.

  5. Run automated vulnerability and provenance checks against both direct and transitive dependencies.

  6. Review and deliberately update dependency pins instead of resolving unrestricted latest versions during installation.

  7. Run installation and the API server as a dedicated, non-privileged account in an isolated environment with restricted filesystem and network access.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 20)May include surrounding context.

Search across all registries

bash
curl -s -X POST http://localhost:8016/v1/search \
  -H "Content-Type: application/json" \
  -d '{"query": "security scanning"}' | jq

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 44)May include surrounding context.

Search across all registries

bash
curl -s -X POST http://localhost:8016/v1/search \
  -H "Content-Type: application/json" \
  -d '{"query": "security scanning"}' | jq

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 36)May include surrounding context.

Get recommendations for a problem

bash
curl -s -X POST http://localhost:8016/v1/recommend \
  -H "Content-Type: application/json" \
  -d '{"problem": "I need to validate data before passing it to another agent"}' | jq

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The docstring states that each adapter implements registry-specific search, but no such adapters exist and all searches use one in-memory list. In a skill-discovery context, this mismatch undermines provenance and can cause agents to overtrust the source, freshness, and registry isolation of results, creating an integrity and supply-chain selection risk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The module and skill metadata imply live cross-registry search, but the implementation only searches a hardcoded local list. This can mislead users and downstream agents into making trust or completeness decisions based on stale, incomplete, or fabricated registry coverage, which is a security-relevant integrity issue for a discovery tool.

Content

No source excerpt is available for this finding.