Back to skill

Security audit

Embedded OTA / Bootloader Toolkit

Security checks for vulnerabilities and agentic risk

Overview

The skill received a suspicious verdict because scanners reported a potentially unsafe partition-header generator and a report-page script injection weakness, while analysis was incomplete.

Treat this as a review item before installing. The skillspector scan was partial: it inspected 6 of 7 components, and obfuscated text in ota_audit.py could not be fully evaluated. Consider avoiding installation until the partition-header issue is addressed and the incomplete analysis is resolved; if you use it, be cautious with generated HTML reports and with outputs from overflowing layouts.

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
ota_audit.py
Finding

Attacker-controlled source filenames execute script in generated HTML reports

Content
View full analysis

Impact

Opening the report executes attacker-supplied JavaScript in the report's browser context. It can alter the displayed audit findings and read or transmit report content available in that document; additional access depends on the browser and hosting context.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • 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
Findings (5)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 17)May include surrounding context.

md
- `ota_audit.py` — audit side: statically scans upgrade code and reports brick-vulnerable spots by severity (no compiling, no hardware)

Self-Modification

High
Category
Rogue Agent
Confidence
90% confidence
Finding

Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.

Content

Scanner excerpt · ota_audit.py (reported line 322)May include surrounding context.

python
SIG_OTA_CTX = re.compile(
    r"(?<![A-Za-z0-9])(?:"
    r"bootloader|boot_load|boot_(?:jump|app|entry|loader|image|flag|mode|update|upgrade|request)"
    r"|iap|ota|dfu|fota|firmware|fw_update|upgrade|update_fw|app_update|img_update|self_update"
    r")\w*",
    re.IGNORECASE)

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

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 26)May include surrounding context.

md
- **Dead on first interrupt after upgrade**: forgot `SCB->VTOR` relocation before jumping to APP — all APP interrupts land in the bootloader vector table. Rule 1.
- **Bricked by banked upgrades**: jumping with an empty/half-flashed APP partition — instant HardFault. Rule 3.
- **"Boots but misbehaves"**: received data written to flash without verification — transmission errors became firmware. Rule 4.
- **Watchdog kills upgrade mid-flash**: multi-second erase/write without feeding the dog — every failure is another brick chance. Rule 6.
- **"Upgrade succeeded but firmware is broken"**: NOR flash can only write 1-to-0; writing a page without erasing first corrupts it. Rule 14.
- **Don't know what goes in the header**: an industrially adequate 64-byte image header format + generator + verifier.

The partition-header generator promises a compile-time bounds assertion but omits it precisely when the layout overflows flash.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

When a layout exceeds flash capacity, layout_to_c() substitutes a warning comment for the compile-time assertion, removing enforcement exactly when it is needed. The CLI also writes this header before returning a failure status, so workflows that ignore that status can consume an unsafe partition table. In an OTA/bootloader toolkit, downstream firmware trusting these bounds could attempt out-of-range flash operations and cause upgrade failures or device unavailability; the warning and nonzero exit status reduce, but do not eliminate, this risk.

Content

No source excerpt is available for this finding.