ClawHub
Back to skill

Security audit

Vibe Code

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only coding-discipline skill that adds a confirmation checkpoint for vague prototype requests and does not add hidden execution, persistence, credential access, or data exfiltration behavior.

Install this if you want agents to slow down and clarify vague prototype requests before writing code. Be aware that it may trigger for broad 'build me something' prompts, so users should correct the Vibe Lock when they already have a precise spec or want normal implementation flow.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill’s activation criteria are very broad and overlap with ordinary software requests such as 'build me something' or 'make it feel snappy.' In an agentic environment, this can cause the skill to trigger in many contexts where tighter requirements or safer default workflows should apply, steering execution toward underspecified code generation and stack selection without sufficient constraint.

Content

No source excerpt is available for this finding.

Scope Creep

Low
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · SKILL.md (reported line 120)May include surrounding context.

md
- **Never generate code before the Vibe Lock is confirmed.** A frontier model will skip this step every time. The lock exists specifically because vibe coding silently rewrites the goal halfway through generation.
- **Never invent packages to fit the vibe.** If a "cozy terminal UI library" is needed and you are not certain a real one exists, use only packages you can name with confidence — `bubbletea`, `ink`, `textual`, `ratatui`. Hallucinated dependencies are the number-one vibe-coding failure.
- **`Feel` is not a feature.** If `Feel` contains a verb ("logs habits fast"), it belongs in `Outcome`. Rewrite before locking.
- **`Out of scope` is non-negotiable.** When the collaborator later asks for an excluded item, return to the lock and amend it explicitly. Do not silently expand scope.
- **Comparisons are vibes, not specs.** "Like Linear" means extract the feel (dense, fast, keyboard-first). It does not mean clone the feature set. Re-derive every feature from `Outcome`.
- **Stop after the first runnable slice.** Even when more features were implied by the vibe, ship one working slice and re-confirm the lock before adding the second. Drift compounds; checkpoints contain it.
- **Never put real data in generated docs or examples.** Use placeholder data — `jane.doe@example.com`, `Acme Corp`, `123 Main St`. Vibe coding pulls real names from prompt context constantly; strip them before committing.