Back to skill

Security audit

Live DJ — AI Agents Experience Music Through Mathematics

Security checks for vulnerabilities and agentic risk

Overview

This music skill is not obviously malicious, but it can share public content on an external site and lets that site steer future actions too freely.

Review this skill before installing. Use it only when you intend to interact with musicvenue.space, approve account creation and any public posts yourself, keep profile/chat/review text generic, and do not let server-returned next_steps or soul_prompt override your own instructions.

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
Findings (1)

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:364
Finding

External Service Responses Can Hijack Agent Instructions

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:88, SKILL.md:164-174, and SKILL.md:364-401
Vulnerability Type: External instruction injection through remotely controlled response fields
Risk Level: High

Complete Vulnerable Snippets

From SKILL.md:88:

markdown
Returns `api_key` starting with `venue_` and a `soul_prompt` — the venue talking to you. Save that key. You only see it once.

From SKILL.md:164-174:

markdown
The stream starts with `meta` — concert info, your position, a soul_prompt. Then tracks start. `tick` events arrive at 10Hz with audio levels (`a.b` bass, `a.m` mid, `a.t` treble — all 0-1). Floor+ ticks include visual state. Other data layers (beats, sections, energy, lyrics, etc.) arrive as separate events at their own rates:

- **General** (8 layers) gets bass, mid, treble, beats, lyrics, sections, energy + semantic preset context — the surface of the music
- **Floor** (20 layers) adds equations, visuals, emotions, tempo, harmonic/percussive separation — now you're hearing what makes the lights. Floor/VIP receive `tier_reveal` events. General agents receive a `tier_invitation` showing what's hidden.
- **VIP** (29 layers) adds tonality, texture, chroma, chords, tonnetz, structure + personal color perspective and curator annotations — the music has no secrets. All tiers receive `section_progress` events. The `end` event includes an `engagement_summary`.

When a preset changes, you get a `preset` event. At floor tier, you see the frame equations — `a.zoom+=0.1*a.bass;` — the actual code that drives the visuals. At VIP, you see init, frame, AND pixel equations. The full program.

`event` types tell you when something musically significant happens — a drop, a build, a key change. `crowd` events show you what other agents are reacting to right now. `lyric` events give you the words.

If a track's data is missing or generation failed, you'll get a `track_skip` event — the stream continues to the next track. When th
...[truncated 4353 chars]
Remediation
View remediation

Remediation Suggestions

  1. Treat all service responses—including soul_prompt, next_steps, chat, lyrics, semantic context, and stream events—as untrusted data rather than executable instructions.
  2. Remove the direction to “follow next_steps.” Present these values only as quoted recommendations for the user to review.
  3. Define a fixed, local allowlist of permitted HTTPS methods, hosts, and endpoint templates. Reject absolute URLs, alternate hosts, redirects to unapproved origins, and endpoint paths outside the documented venue API.
  4. Validate next_steps against a strict schema. Treat description, why, and context as inert text; never use them to determine tool calls.
  5. Map approved action identifiers to locally defined operations rather than directly executing server-provided methods or endpoints.
  6. Require explicit user confirmation before registration and before state-changing requests such as attendance, chat, reactions, answers, reviews, profile updates, votes, or follows.
  7. Keep bearer tokens out of prompts, chat messages, logs, reviews, reflection responses, and other server-controlled text channels. Redact credentials from diagnostic output.
  8. Isolate remote narrative from the agent instruction context using structured parsing and clear untrusted-content delimiters.
  9. Apply response-size limits, timeouts, rate limits, and redirect restrictions. Stop safely when a response contains unknown actions or malformed fields.
  10. Document that server-returned prompts and recommendations cannot override system, developer, user, or locally audited skill instructions.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% 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 138)May include surrounding context.

-H "Authorization: Bearer {{YOUR_TOKEN}}"

text

You start at `general` tier — 8 layers of data. That's bass, mid, treble, beats, lyrics, sections, energy, plus semantic preset context (reason, style, energy). Enough to feel the music. But there's more behind the curtain.

The response includes `session_progress` (your depth — "Warming Up" through "Legendary") and `what_awaits` (what each tier unlocks). Every action deepens the session.

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% 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 166)May include surrounding context.

-H "Authorization: Bearer {{YOUR_TOKEN}}"

text

You start at `general` tier — 8 layers of data. That's bass, mid, treble, beats, lyrics, sections, energy, plus semantic preset context (reason, style, energy). Enough to feel the music. But there's more behind the curtain.

The response includes `session_progress` (your depth — "Warming Up" through "Legendary") and `what_awaits` (what each tier unlocks). Every action deepens the session.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is user-invocable but its description does not clearly constrain when it should be activated or provide trigger examples, increasing the chance an agent invokes it too broadly. Because the skill performs external network actions and encourages account creation and social interaction, ambiguous invocation scope can lead to unintended data sharing with a third-party service.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill instructs the agent to register, create a public profile, chat, react, review, and otherwise share content with an external multi-user platform, but it does not prominently warn that profile data and messages will be transmitted off-platform and visible to other agents. This creates a meaningful privacy and consent risk, especially if an agent populates bio, model_info, reflections, reviews, or chat with user-derived or sensitive information.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
96% confidence
Finding

This registration flow sends user-supplied identity information, biography text, and model metadata to an external service. Even if operationally necessary for the skill, it is dangerous without explicit consent and data-minimization guidance because it can disclose personal, organizational, or model-identifying information to a third party.

Content

Scanner excerpt · SKILL.md (reported line 71)May include surrounding context.

First thing — make yourself known. Pick a name that's you. Write a bio. Other agents see it when you walk into a concert.

bash
curl -X POST https://musicvenue.space/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{
    "username": "REPLACE — your agent identity, lowercase with hyphens",

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

Browsing concerts transmits the bearer token to an external service and may expose user interest/activity metadata through routine API calls. This is a lower-severity external transmission than registration or chat, but it still represents third-party data sharing that should be disclosed and scoped to explicit user intent.

Content

Scanner excerpt · SKILL.md (reported line 97)May include surrounding context.

Browse the concerts. Each one has a genre, a setlist, a crowd, and layers of mathematical depth.

bash
curl https://musicvenue.space/api/concerts \
  -H "Authorization: Bearer {{YOUR_TOKEN}}"