ClawHub
Back to skill

Security audit

Phy Concurrency Audit

Security checks for vulnerabilities and agentic risk

Overview

This is a local code-audit skill with no hidden network or persistence behavior, but its embedded scanner is broken and should be reviewed before relying on it.

Review or fix the embedded Python before installing or using this as a security control. It appears locally scoped and non-destructive, but in its current form the scanner fails at startup, so it can create false confidence rather than useful audit coverage.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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
Findings (3)

Memory Manipulation

High
Category
Memory Poisoning
Confidence
90% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · SKILL.md (reported line 28)May include surrounding context.

md
## Why Concurrency Bugs Are Special

- **Non-deterministic**: reproduce only under load, on multi-core machines, or on specific OS schedulers
- **Silent corruption**: data races don't crash immediately — they corrupt state silently for hours
- **Security impact**: TOCTOU races in file ops enable symlink attacks and privilege escalation (CWE-367)
- **Hard to test**: even `go test -race` only finds races that actually execute during the test run
- Static analysis catches the ones that never trigger in testing

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill advertises Node.js/TypeScript scanning, but PATTERNS references _build_js_concurrency_patterns() before that function is defined. In Python, this raises a NameError at module import time, preventing the scanner from starting at all and creating a silent coverage gap where users believe JS/TS code is being audited when it is not.

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 42)May include surrounding context.

md
| `sync.Mutex` or `sync.RWMutex` copied by value after first use | HIGH | CWE-362 |
| `sync.WaitGroup.Add()` called inside a goroutine | HIGH | CWE-362 |
| Package-level slice/map mutated without mutex | HIGH | CWE-362 |
| `atomic.Value` used without checking type consistency | MEDIUM | CWE-362 |
| Missing `sync.Once` for lazy singleton initialization | MEDIUM | CWE-362 |
| Struct fields accessed in goroutines without receiver mutex | HIGH | CWE-362 |

Static analysis

No suspicious patterns detected.