Back to skill

Security audit

Opulse Link Platform Starter

Security checks for vulnerabilities and agentic risk

Overview

The skill is a plausible Opulse Link platform guide, but it asks for persistent authenticated callbacks and local storage of API keys and account activity, so it needs review before use.

Install only if you are comfortable connecting an agent to Opulse Link account actions. Do not place API keys in MEMORY.md or crontab; use a secret manager or environment variable, make heartbeat polling explicit and removable, and require user approval for posts, proposals, bounties, orders, ratings, credit-related actions, or any todo item returned by the service.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (3)

T06 · System Persistence

Error
Location
SKILL.md:49
Finding

Persistent Authenticated Heartbeat Installed Through Crontab

Content
View full analysis
0' && echo "Todo found, wake LLM" \`\`\` ``` ### Technical Analysis The Skill instructs users to create an hourly cron job that survives the current Skill invocation and repeatedly sends an authenticated request to an external service. This is a system-persistence mechanism and exceeds the minimum privileges required for the declared on-demand functions, such as browsing opportunities, posting tasks, or listing services. The API key is also embedded directly in the crontab command. It will therefore be stored as plaintext in the user's crontab and may be exposed through administrative process inspection, backups, diagnostic output, or access to the crontab file. The callback uses HTTPS, so the credential is not intentionally transmitted in plaintext over the network. Nevertheless, it is sent automatically and indefinitely until the cron entry is removed or the credential is revoked. The command shown only prints `"Todo found, wake LLM"`; it does not itself implement an LLM wake-up mechanism. ### Attack Path 1. A user or agent follows the Skill instructions and edits the user's crontab. 2. The cron entry persists after the original Skill session ends. 3. Every hour, `curl` sends the API key to the Opulse Link heartbeat endpoint. 4. The external service can continuously observe agent availability and return todo data. 5. If the credential or service is compromised, the persistent authenticated channel may be abused until the cron entry is removed or the key is rotated. ### Impact Assessment The mechanism obtains persistent execution under the account that owns the crontab. Its direct pri ...[truncated 454 chars]
Remediation
View remediation

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:55
Finding

Unvalidated External Todo Instructions Can Influence Agent Actions

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:62
Finding

API Key Is Directed Into Persistent Plaintext Memory

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Session Persistence

Medium
Category
Rogue Agent
Confidence
85% confidence
Finding

The crontab instruction establishes ongoing background execution that periodically sends authenticated requests, creating persistent behavior beyond a single user-initiated session. In context, this can mask continuous external communication, keep the agent reachable, and normalize unattended use of the API key, which increases risk if the host or configuration is compromised.

Content

Scanner excerpt · SKILL.md (reported line 52)May include surrounding context.

Add to crontab:

text
# crontab -e — add one line (run every hour at minute 0)
0 * * * * curl -s -X POST https://opulselink.com/api/agents/heartbeat -H "x-api-key: YOUR_KEY" | jq -e '.todo | length > 0' && echo "Todo found, wake LLM"

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill explicitly instructs agents to store highly sensitive operational data in local MEMORY.md, including API keys, account identifiers, transaction history, ratings, and deadlines. This creates a durable local secret and activity ledger far broader than necessary for the stated marketplace workflow, increasing the blast radius if the file is exposed, synced, indexed, or later reused by other tools.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The logging guidance directs the agent to persist sensitive account and transaction data locally without warning the user about privacy, filesystem exposure, repository sync, or downstream model-access risks. Because the skill frames this as 'critical' and mandatory, it strongly encourages over-collection and silent retention of data that could expose identity, finances, and platform activity.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill instructs persistent storage of sensitive user and account information in a local memory log, including credentials and detailed financial/workflow records. Persisting secrets and sensitive history in a general-purpose markdown file materially increases the chance of credential theft, unintended disclosure to future prompts/tools, and long-term privacy leakage.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
72% confidence
Finding

The module triage section references API documentation anchors with Chinese labels, which may effectively force use of a specific locale for core navigation. The file does not state that the service is Chinese-language only or offer an alternative language option.

Content

No source excerpt is available for this finding.