Back to skill

Security audit

embedded-register-verify

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because its reported capabilities do not fully match its stated firmware review purpose, and one finding says some unresolvable register writes may be marked as passed.

Consider whether the skill's actual register lookup and reference features fit your needs, and whether you are comfortable with the reported possibility that an unresolvable register write may be marked as passed. Skillspector's analysis is partial because a local path-like reference could not be resolved, so the evidence is incomplete.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Some functions provide relevant supporting checks, including name resolution, value-width validation, enum resolution, and read-only-write warnings. However, the supplied module primarily implements a general register-lookup/reference toolkit—the very use case the description directs elsewhere. Its SVD linting evaluates definition quality, not firmware writes. File reads are limited to SVD metadata and hashing and are not independently suspicious. This assessment concerns the supplied chunk; other code could implement the declared review workflow.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Unresolvable MODIFY_REG values can be classified as passed despite the documented uncertainty policy.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

When the MODIFY_REG set expression cannot be evaluated, _check_modify_value returns without calling add_undet. If no other check reports an issue, check_file classifies the write as passed even though its value, width, and enumeration legality remain unverified. This is particularly consequential in a firmware-verification skill: runtime-dependent or unsupported expressions can receive misleading assurance and allow unsafe register configurations to escape review.

Content

No source excerpt is available for this finding.

The module promises no I/O, but device_info() and macros_for() read the SVD file from disk.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The module docstring at L0003 explicitly states that functions only read Device and perform no I/O. Contrary to that contract, device_info() reads filesystem metadata and hashes device.path at L1052–L1055, while macros_for() also invokes the disk-reading hash helper at L1211; _sha256() opens and reads the file at L1087–L1089. These operations access the filesystem rather than operating solely on the supplied in-memory Device.

Content

No source excerpt is available for this finding.