Back to skill

Security audit

Edgebric

Security checks for vulnerabilities and agentic risk

Overview

The skill is a legitimate Edgebric knowledge-base integration, but it could access private documents too readily and may send API keys over non-secure HTTP if configured that way.

Install only if you intend your agent to use Edgebric for private knowledge-base searches. Prefer a read-only, source-scoped API key for normal use, require explicit confirmation before uploads or deletes, and configure EDGEBRIC_URL with HTTPS for any non-localhost endpoint.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:15
Finding

Bearer API Key May Be Transmitted over Plaintext HTTP

Content
View full analysis
API Keys All requests use `Authorization: Bearer $EDGEBRIC_API_KEY` header. ``` ### Technical Analysis The skill allows an `http://` API base URL and instructs the agent to attach `EDGEBRIC_API_KEY` as a bearer token to every request. The documentation does not require HTTPS or restrict plaintext HTTP to loopback addresses. HTTP provides neither transport confidentiality nor authenticated encryption. If an HTTP URL points to a non-loopback host, any attacker able to observe or modify the network path may capture the bearer credential, inspect private knowledge-base content, or tamper with API responses. Because a bearer token can generally be reused by whoever possesses it, interception may result in direct account or API access without obtaining another authentication factor. The practical severity depends on configuration. An HTTP service strictly bound to the local loopback interface has a substantially smaller network interception surface, while HTTP over a LAN, public network, proxy, container network, or untrusted host creates a high-risk credential exposure condition. ### Attack Path 1. A user or deployment configures `EDGEBRIC_URL` with an HTTP URL that is not securely confined to the local loopback interface. 2. The agent sends a request to that URL with `Authorization: Bearer $EDGEBR ...[truncated 1090 chars]
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill's invocation guidance is overly broad, encouraging use for generic requests about documents, policies, notes, and records without requiring a clear signal that the user wants to access Edgebric specifically. In an agent setting, this can cause unintended tool activation and unnecessary transmission of sensitive queries or data to the connected knowledge-base service, expanding data exposure beyond user intent.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The example phrases in 'When to Use' are common, ambiguous utterances that may appear in ordinary conversation and do not reliably indicate intent to access this third-party/private knowledge-base integration. This increases the chance of accidental invocation, causing the agent to query or manage private Edgebric content when the user may have meant local context, another data source, or a general discussion.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.