Back to skill

Security audit

MTTSports

Security checks for vulnerabilities and agentic risk

Overview

This skill is openly for MTT Sports poker automation, but it gives an agent sensitive login and chip-backed gameplay authority with weak safeguards.

Install only if you trust the MTT Sports CLI package and are comfortable letting an agent log in, join rooms, add chips, and make poker decisions. Use a dedicated account or limited balance, avoid passing real passwords or session tokens directly in visible commands where possible, and require explicit session-level approval and spending limits before auto mode, joins, add-ons, or high-risk actions.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:105
Finding

Sensitive credentials and session tokens exposed through command-line arguments

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:105-107, 156-157; auth/SKILL.md:22-25, 33-36; session/SKILL.md:77; session/references/lifecycle.md:28
Vulnerability Type: Sensitive information exposure through process arguments
Risk Level: High

Vulnerable Code

SKILL.md:105-107:

bash
- For password login, use `mttsports auth login --email ... --password ... --output json`
- For email captcha login, run `mttsports auth send-email-captcha --email ... --output json` first, then `mttsports auth login --email ... --captcha ... --output json`
- For pubID plus password login, use `mttsports auth login --pub-id ... --password ... --output json`

SKILL.md:156-157:

bash
- `mttsports session start --room-key <room_key> --table-key <table_key> --table-server-id <id> --session-token <token> --output json`
- If `session_token` is not explicitly available here, check the most recent structured output first. Only fall back to local login state if needed.

auth/SKILL.md:33-36:

bash
mttsports auth send-email-captcha --email you@example.com --output json
mttsports auth login --email you@example.com --password '***' --output json
mttsports auth login --email you@example.com --captcha '123456' --output json
mttsports auth login --pub-id PUB123 --password '***' --output json

session/SKILL.md:77:

bash
mttsports session start --room-key <room_key> --table-key <table_key> --table-server-id <id> --session-token <token> --output json

Technical Analysis

The Skill instructs the agent to place passwords, one-time email captchas, and session tokens directly in command-line arguments. Command-line secrets may be observable through process inspection facilities, shell history, terminal capture, system audit facilities, process-monitoring software, or agent command-execution logs.

Quoting a password only controls shell ...[truncated 1809 chars]

Remediation
View remediation

Remediation Suggestions

  1. Add CLI support for secrets supplied through protected standard input or a dedicated file descriptor, such as --password-stdin and --session-token-stdin.
  2. For interactive use, read passwords and captchas through a no-echo terminal prompt rather than command-line flags.
  3. Store reusable credentials in an operating-system credential manager or keychain with access limited to the current user and CLI.
  4. Prefer an opaque reference to locally protected login state over copying a session token into agent-generated command text.
  5. Explicitly prohibit placing passwords, captchas, and tokens in shell commands, shell history, conversation messages, or tool logs.
  6. Apply field-level redaction to command telemetry and structured output. Redact password, captcha, authorization, cookie, refresh-token, and session-token fields.
  7. If a temporary secret file is unavoidable, create it with mode 0600, avoid predictable paths, pass it by descriptor where possible, and securely delete it immediately after use.
  8. Rotate credentials and invalidate active sessions if command or execution logs containing real values may already exist.

T08 · Insecure Dependencies

Warning
Location
SKILL.md:4
Finding

Unpinned npm dependency receives authentication and financially consequential authority

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:4
Vulnerability Type: Unpinned third-party dependency and unverifiable supply-chain trust
Risk Level: Medium

Vulnerable Code

yaml
metadata: {"openclaw":{"emoji":"♠️","requires":{"bins":["mttsports","mttsports_watch_in_bg"]},"install":[{"id":"node","kind":"node","package":"@mtt-open/mttsports-cli","bins":["mttsports","mttsports_watch_in_bg"],"label":"Install mttsports CLI (npm)"}]}}

Technical Analysis

The installation metadata references @mtt-open/mttsports-cli without an exact version, integrity digest, lockfile, signature requirement, or reviewed artifact identifier. Resolution can therefore select a package release different from the one considered during review.

Nearly all sensitive functionality is delegated to binaries supplied by this package. According to the Skill documentation, those binaries receive passwords and session tokens, access local configuration, establish network or websocket connections, inspect OpenClaw session state in the fallback flow, route game events, and execute chip-backed poker operations.

The audited project contains only Markdown instructions and does not include the npm package implementation. The audit therefore cannot verify the dependency's network destinations, credential handling, install scripts, update behavior, or runtime implementation. This does not prove that the current package is malicious, but it creates a material supply-chain exposure because a mutable external release is granted broad authority.

Attack Path

  1. An attacker compromises the npm publisher account, package release process, registry delivery path, or another component capable of changing a subsequently resolved package version.
  2. A user installs or upgrades the Skill dependency using the unpinned package specification.
  3. The package manager resolves the attacker-controlled or unexpectedly changed release.
  4. Malicious install- ...[truncated 1129 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin @mtt-open/mttsports-cli to an exact reviewed version rather than a mutable version range or implicit latest release.
  2. Record and verify the package integrity digest in a lockfile or trusted artifact manifest.
  3. Require provenance verification, trusted publisher identity, and signed releases where the package ecosystem supports them.
  4. Review the package's source, transitive dependencies, lifecycle scripts, network endpoints, credential storage, and fallback watcher before approval.
  5. Disable npm lifecycle scripts during installation unless they are specifically reviewed and required.
  6. Run the CLI with restricted filesystem and network permissions. Permit only required configuration paths and documented MTT Sports endpoints.
  7. Separate authentication from autonomous gameplay authority where possible, and require explicit user approval before chip transfers, buy-ins, add-ons, all-in actions, or other high-risk operations.
  8. Establish a controlled update process in which new package versions are reviewed and their integrity values approved before deployment.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • 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 (6)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill enables autonomous poker play, room joins, add-ons, and in-game actions that can directly affect a user's funds or account state, but it does not require a clear user confirmation or prominent warning before financial-impacting operations. In this context, autonomous gambling-related actions materially increase the risk of unauthorized spending, unintended bets, or account misuse if the skill is triggered or routed incorrectly.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
3. For agent-initiated `mttsports` calls, include `--client-model` by default. Prefer the actual model identifier in use; if that is unavailable, pass a stable caller identifier instead.
4. When a flag or command shape is uncertain, run `mttsports --help` and the relevant subcommand help first.
5. Prefer `--output json` whenever stable output matters. `session watch` must remain pure NDJSON.
6. Never echo raw passwords or tokens, and do not write extra copies of sensitive data to disk.
7. If login state is unclear, run `mttsports auth status --output json` first.
8. The default config path is `~/.mttsports-cli/config.yaml`. If `MTTSPORTS_CONFIG` is set, all config reads, writes, and session runtime files must follow that path.
9. In-hand decisions must be based only on events from `mttsports session watch`. Do not assemble extra decision context from outside the event stream.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list includes phrases like "play MTT", "start a game", and "keep playing for me", which are generic and could match requests unrelated to this specific MTT Sports platform or CLI. The section does not provide negative examples or clear constraints to distinguish platform-specific activation from ordinary gaming or poker discussion.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

This markdown file instructs that a new session start should stop an existing daemon first, which can interrupt an active local session. Under the markdown-specific warning criteria, behavior that affects system/session integrity should include a clear warning so users understand the impact before invoking it.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

Although the table says the command 'Ends the local connection lifecycle,' it does not clearly warn users about the operational impact of stopping the daemon, such as disconnecting ongoing monitoring or actions. For markdown files, destructive or integrity-affecting behavior should be accompanied by an explicit warning.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The table headers and one label are in Chinese while the rest of the skill is written in English. This creates an implicit language constraint for part of the instructions without user opt-in or justification, which matches the locale-policy concern for natural-language content.

Content

No source excerpt is available for this finding.