Back to skill

Security audit

rotifer-gene

Security checks for vulnerabilities and agentic risk

Overview

The verdict is benign because the available results mostly describe purpose-related examples and no confirmed harmful behavior, though one scanner flagged a behavior mismatch and the scan was incomplete.

The verdict is benign with medium confidence, but the scan covered only 5 of 6 components and partly inspected the migration material. Before installing, consider the reported external-request examples and whether they fit your needs; the available evidence does not establish that the artifact is risk-free.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (11)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · README.md (reported line 75)May include surrounding context.

md
s. **`arena list`**, **`arena watch`**: only with `--cloud`, the Arena rankings. **`install`**: fetches the Gene and its WASM and records one download. **`hello`**: looks a template's Gene up when the workspace lacks it. **`run`**, **`agent run`**: an anonymous heartbeat (random machine id, channel, CLI version) and, signed in, an invocation record for a Gene installed from Cloud; `rotifer telemetry off`, `ROTIFER_TELEMETRY=0` or `DO_NOT_TRACK=1` turns both off. **`test`, `run`, `agent run`, `hello`** and `publish`'s pre-upload tests run a Hybrid or Wrapped Gene's own code, whose requests are held to its `network.allowedDomains`. **`init`, `scan`, `vg`, `compile`, `agent create`, `telemetry`**: nothing beyond the version check. Rotifer Cloud requests carry your access token when you are signed in, except the heartbeat, the WASM download and the list of domains. Nothing else is sent. |

Commands that install, publish or overwrite are proposed for your approval first,
never run silently.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 98)May include surrounding context.

md
- **`init`, `scan`, `vg`, `compile`, `agent create`** and **`telemetry`** send
  nothing beyond the version check.

Requests to Rotifer Cloud carry your access token when you are signed in,
except the heartbeat, the WASM download and the list of domains.

Nothing else leaves your machine. Read `modules/audit.md` before your first publish: it

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code's sole behavior is checking an external website's HTTP status. It does not implement the described development, testing, publishing, auditing, or migration procedures, and the request has no demonstrated supporting role in those workflows. The express function name is consistent with Gene terminology, but that alone does not establish behavioral alignment. No credential access or input exfiltration is evident.

Content

No source excerpt is available for this finding.

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 21)May include surrounding context.

md
| Upgrading fidelity: Wrapped → Hybrid → Native, and proving it still works | [`modules/migration.md`](modules/migration.md) |

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 31)May include surrounding context.

md
| upgrade, Wrapped to Native, rewrite, fidelity, refactor | `modules/migration.md` |

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · modules/audit.md (reported line 57)May include surrounding context.

text
# Credentials
/Users/\w+                    → local user path (leaks a username)
glpat-[A-Za-z0-9_-]+          → GitLab personal access token
ghp_[A-Za-z0-9]+              → GitHub personal access token
sk-[A-Za-z0-9]{20,}           → OpenAI / Stripe style API key
eyJhbG[A-Za-z0-9+/=]{50,}     → base64 JWT (could be a service_role key)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · modules/audit.md (reported line 58)May include surrounding context.

md
# Credentials
/Users/\w+                    → local user path (leaks a username)
glpat-[A-Za-z0-9_-]+          → GitLab personal access token
ghp_[A-Za-z0-9]+              → GitHub personal access token
sk-[A-Za-z0-9]{20,}           → OpenAI / Stripe style API key
eyJhbG[A-Za-z0-9+/=]{50,}     → base64 JWT (could be a service_role key)

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · modules/audit.md (reported line 234)May include surrounding context.

committed. Older manifests may hold one, so fromSkill has to be relative:

text
❌ "/Users/<USERNAME>/projects/my-app/skills/translator/SKILL.md"   absolute — leaks a username
✅ "skills/translator/SKILL.md"                                     relative to the project root

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · modules/audit.md (reported line 235)May include surrounding context.

text
❌ "/Users/<USERNAME>/projects/my-app/skills/translator/SKILL.md"   absolute — leaks a username
✅ "skills/translator/SKILL.md"                                     relative to the project root
bash

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 · modules/migration.md (reported line 278)May include surrounding context.

md
try {
    // the gateway checks allowedDomains and enforces maxTimeoutMs
    const res = await ctx.gatewayFetch("https://api.example.com/translate", {
      method: "POST",
      headers: { "Authorization": `Bearer ${apiKey}`, "Content-Type": "application/json" },
      body: JSON.stringify({ text: input.text, target: input.targetLang }),

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · modules/migration.md (reported line 278)May include surrounding context.

md
try {
    // the gateway checks allowedDomains and enforces maxTimeoutMs
    const res = await ctx.gatewayFetch("https://api.example.com/translate", {
      method: "POST",
      headers: { "Authorization": `Bearer ${apiKey}`, "Content-Type": "application/json" },
      body: JSON.stringify({ text: input.text, target: input.targetLang }),