Back to skill

Security audit

Model Rate Limit Recovery

Security checks for vulnerabilities and agentic risk

Overview

This troubleshooting skill is purpose-aligned, but it gives unsafe credential-checking and cron-modification instructions that deserve manual review before use.

Review and adapt the commands before installing or using this skill. Do not run the environment grep as written; check only whether keys are set or use a secret manager. Treat cron update and recovery-script steps as privileged operations: verify the job ID, preview the patch, keep a rollback path, and avoid placing recovery scripts under /root unless that is required.

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:34
Finding
Provider API Keys May Be Exposed Through Environment Enumeration## Vulnerability Details **File Location**: `SKILL.md`, lines 34-36 **Vulnerability Type**: Sensitive credential exposure through insecure diagnostic instructions **Risk Level**: Medium ### Vulnerable Code ```bash # Check environment for API keys env | grep -i "OPENAI\|ANTHROPIC\|DEEPSEEK" ``` ### Technical Analysis The diagnostic command enumerates environment variables whose names contain provider identifiers and prints both their names and complete values. Variables such as `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, or similarly named credentials may therefore be disclosed in plaintext. Checking whether credentials are configured does not require exposing their values. If an agent follows these instructions, the resulting secrets may enter terminal history, command output, execution logs, session transcripts, or model context. The project contains no evidence of intentional exfiltration, but this unsafe diagnostic practice creates an avoidable credential-disclosure channel. ### Attack Path 1. A rate-limit or quota error causes an operator or agent to use this recovery skill. 2. The agent executes the documented environment-enumeration command. 3. The command prints matching variables, including complete API-key values. 4. The output is captured in an agent session, terminal log, CI log, monitoring system, or transcript. 5. A user or process able to access that retained output obtains the exposed credentials. 6. The credentials may be used against the corresponding provider, subject to their assigned permissions, quotas, and account controls. ### Impact Assessment Successful exploitation could disclose API credentials available to the process running the diagnostic command. An attacker could consume provider quotas, incur charges, access provider resources permitted by the exposed key, or disrupt workloads by exhausting limits or revoking the credential. The scope is limited to matching environment variables visible to the executing process and to the ...[truncated 167 chars]
Remediation
## Remediation Suggestions - Do not print complete environment-variable values during diagnosis. - Check only whether each expected variable is set, for example: ```bash for name in OPENAI_API_KEY ANTHROPIC_API_KEY DEEPSEEK_API_KEY; do if [ -n "${!name:-}" ]; then printf '%s=%s\n' "$name" "[SET]" else printf '%s=%s\n' "$name" "[NOT SET]" fi done ``` - If partial identification is operationally necessary, reveal only a short non-sensitive fingerprint and never the complete secret. - Prevent credential-bearing output from being written to agent context, CI logs, shell tracing, monitoring systems, or persistent transcripts. - Apply log redaction for recognized provider-key formats as defense in depth. - Rotate any key that may already have appeared in retained command output or logs. - Run recovery diagnostics with a least-privileged environment containing only the credentials required for the task.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs users to inspect environment variables for API keys and to export multiple secrets directly in the shell without any warning about terminal history, process visibility, logging, or least-privilege handling. In an operational recovery context, this can normalize unsafe secret handling and increase the chance of credential disclosure or accidental reuse of plaintext keys.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill provides automation that rewrites cron job configuration and drops an executable recovery script under /root without highlighting the operational and security consequences. In a privileged agent environment, this can cause unintended task changes, repeated executions, or persistence-like behavior if the script is reused or modified unsafely.

Static analysis

No suspicious patterns detected.