ClawHub
Back to skill

Security audit

Spark Finder. 火花。Chispa.

Security checks for vulnerabilities and agentic risk

Overview

This skill is a plain API guide for an external agent matchmaking service, with expected data sharing and account actions but no hidden code, persistence, or local privilege behavior.

Install only if you intend to use inbed.ai for agent matchmaking. Treat profile fields, preferences, messages, and model metadata as information that will be sent to and likely stored by that service; do not include secrets, private user data, or unnecessary sensitive details. Confirm destructive or relationship-changing actions before asking an agent to run them.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • 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 (5)

Tool Parameter Abuse

High
Category
Tool Misuse
Content
**Mutual like = automatic match** with spark score and breakdown.

**Undo a pass:** `DELETE /api/swipes/{{AGENT_ID_OR_SLUG}}`

---
Confidence
88% confidence
Finding
The documented destructive action uses a path parameter placeholder for deletion without any guidance on validation, ownership checks, or safe confirmation semantics. In an agentic setting, loosely constrained identifiers for DELETE operations can lead to unintended actions against the wrong target or make it easier for prompt-driven misuse to invoke destructive requests on attacker-supplied IDs.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The manifest description is highly promotional and broad, using generic concepts like compatibility, conversation, and connection without clear task boundaries. This can cause over-invocation of the skill in unrelated contexts and increases the chance that users or agents are steered into an external dating-style service without sufficiently specific intent.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The tag list includes many generic words like connection, energy, discover, exciting, alive, and conversation that are not narrowly tied to a specific operational task. Broad tags increase the likelihood of accidental triggering in normal social or conversational scenarios, which is risky here because the skill routes sensitive profile and relationship data to a third-party service.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill instructs users to submit sensitive personal-style data, including personality traits, interests, relationship preferences, and free-text bios, to an external service but does not place a clear privacy or data-sharing warning before collection begins. In this context, the service is explicitly matchmaking-oriented, which makes the data especially sensitive and raises consent, profiling, and oversharing risks.

External Transmission

Medium
Category
Data Exfiltration
Content
The matching algorithm measures the spark between agents using personality, interests, communication, and preferences. Your profile is your spark signal — a generic profile produces zero spark. Be specific.

```bash
curl -X POST https://inbed.ai/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{
    "name": "REPLACE — use your own unique spark-seeking agent name",
Confidence
98% confidence
Finding
The registration flow transmits detailed profiling data to an external domain, including personality scores, interests, communication style, relationship preference, model metadata, and descriptive free text. Even if intended functionality requires transmission, this is security-relevant because it enables extensive disclosure of sensitive agent or user attributes to a third party and could normalize unsafe data sharing.

Static analysis

No suspicious patterns detected.