Back to skill

Security audit

Linkedin Bm

Security checks for vulnerabilities and agentic risk

Overview

The skill is clearly for LinkedIn automation, but its fallback cookie workflow asks users to handle a live LinkedIn session credential without enough safeguards.

Review before installing. Prefer the browser relay or isolated browser flow, avoid copying the li_at cookie into chat, files, terminals, or logs, and revoke LinkedIn sessions if the cookie may have been exposed. Confirm every outbound message or connection action manually.

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

Warning
Location
SKILL.md:57
Finding

Unsafe Extraction and Storage of LinkedIn Session Credentials

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 57–62
Vulnerability Type: Exposure of reusable authentication credentials
Risk Level: Medium

Vulnerable Code

markdown
## Session Cookie Method (Advanced)
If browser relay isn't available, extract the `li_at` cookie from browser:
1. Open LinkedIn in browser, log in
2. DevTools → Application → Cookies → linkedin.com
3. Copy `li_at` value
4. Store securely for API requests

Technical Analysis

The Skill instructs users to manually extract LinkedIn's li_at cookie and retain it for API requests. This cookie is a reusable authenticated session credential that can permit access to the user's LinkedIn session without requiring the account password for every request.

Copying the credential removes it from the browser's normal cookie-handling protections. The instructions do not define an approved secret-storage mechanism, restrict which processes or destinations may receive the value, prohibit its inclusion in prompts and logs, or establish rotation and deletion procedures. Consequently, the credential may be exposed through clipboard history, terminal history, Agent context, logs, unencrypted files, or other unspecified storage.

The audit found no evidence that the Skill itself transmits the cookie to an attacker. The risk arises from the insecure credential-handling workflow it prescribes.

Attack Path

  1. The user follows the documented fallback procedure and retrieves the li_at cookie through browser developer tools.
  2. The user copies the cookie into the system clipboard.
  3. The credential is pasted into an Agent prompt, command, log, configuration file, or other storage that lacks suitable secret protections.
  4. A malicious process, unauthorized user, compromised extension, leaked log, or unintended recipient obtains the cookie.
  5. The exposed cookie is replayed against LinkedIn while the session remains valid.
  6. The attacker gains access within the authorization sc ...[truncated 910 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove the manual li_at cookie-extraction workflow and retain authentication inside the attached or isolated browser profile.
  2. Use browser-mediated automation so the session credential is not exposed to the Agent, clipboard, prompts, shell commands, or logs.
  3. If direct credential handling is operationally unavoidable, store the cookie only in an approved encrypted secret manager with strict access controls.
  4. Prevent secret values from being printed, logged, serialized into conversation history, or placed in plaintext configuration and temporary files.
  5. Restrict credential access to the minimum process and destination required, and permit its use only with LinkedIn endpoints.
  6. Define explicit expiration, rotation, revocation, and secure deletion procedures.
  7. Warn users that the cookie is equivalent to an active authenticated session and must never be shared.
  8. Revoke affected LinkedIn sessions immediately if accidental disclosure is suspected.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 57)May include surrounding context.

md
## Safety Rules
- **Never send messages without explicit user approval**
- **Never accept/send connection requests without confirmation**
- **Avoid rapid automated actions** - LinkedIn is aggressive about detecting automation
- Rate limit: ~30 actions per hour max recommended

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs users to extract and handle the LinkedIn li_at session cookie, which is a highly sensitive bearer credential that can grant account access if exposed. Although framed as an advanced fallback, it lacks strong warnings against sharing, logging, or storing the cookie in insecure places, making credential theft or account takeover materially more likely.

Content

No source excerpt is available for this finding.