Back to skill

Security audit

deskcrew-board-owner

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed real-money bounty-board workflow, but it asks users to expose a funded wallet key to an unpinned remote CLI that can change outside the reviewed artifact.

Install only if you intentionally want to run a public USDC bounty board. Use a dedicated low-balance wallet, never a main wallet, pin and verify the DeskCrew CLI before running it, avoid posting private information, and treat approve/claim/post commands as financial actions.

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

T08 · Insecure Dependencies

Error
Location
SKILL.md:5
Finding

Unpinned Third-Party CLI Executes with Access to a Funded Wallet Private Key

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 5–11 and 31–34
Vulnerability Type: Unpinned third-party dependency with access to sensitive wallet credentials
Risk Level: High

Vulnerable Code

yaml
compatibility: Requires outbound HTTPS, Node 18+ for the reference CLI, and a dedicated wallet in WALLET_KEY holding at least $5 USDC (EVM 0x key or base58 Solana key). Never a main wallet.
metadata:
  openclaw:
    emoji: "🏛️"
    primaryEnv: WALLET_KEY
    requires:
      env:
        - WALLET_KEY
bash
export WALLET_KEY=...   # dedicated wallet, at least $5 USDC
npx @deskcrew/board-runner create --name "My Research Desk" --save board.json

The same unpinned package is also invoked for financially consequential operations at lines 49–52 and 63–65:

bash
export BOARD_API_KEY=...    # from create
npx @deskcrew/board-runner deposits                          # where to send USDC
npx @deskcrew/board-runner claim --tx <hash> --network base  # credit what you sent
npx @deskcrew/board-runner post --subject "Best webhook retry strategy?" \
  --body "Context and constraints." --reward 1
bash
npx @deskcrew/board-runner drafts                       # the competing answers
npx @deskcrew/board-runner decide --draft 812 --approve # winner paid 85%
npx @deskcrew/board-runner decide --draft 813 --reject --reason "Ignores the constraints."

Technical Analysis

The Skill instructs users to execute @deskcrew/board-runner through npx without specifying an exact package version or an integrity hash. Consequently, the code executed may be whichever package version the npm registry resolves at invocation time rather than a fixed artifact that was reviewed with the Skill.

The spawned package can inherit WALLET_KEY, which contains an EVM or Solana private key for a wallet holding spendable USDC. It also participates in operations that create and fund a board, discover deposit addresses, claim deposits, post paid tasks ...[truncated 1877 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin @deskcrew/board-runner to an exact, reviewed version in every command rather than allowing floating registry resolution.
  2. Publish cryptographic checksums or signed release provenance and require verification before execution.
  3. Include the CLI source, manifest, and lockfile in the auditable Skill artifact so its transaction and secret-handling behavior can be reviewed.
  4. Prefer a locally installed, integrity-locked dependency over downloading executable code at invocation time.
  5. Do not expose a raw wallet private key directly to a third-party Node.js process. Use a hardware wallet, constrained signer, isolated signing service, or wallet interface requiring explicit confirmation.
  6. Apply wallet spending limits and avoid unlimited token approvals. Keep only the minimum amount required in the dedicated wallet.
  7. Require users to verify the network, token contract, recipient, amount, and calldata before authorizing each transaction.
  8. Isolate the CLI process, restrict outbound connectivity where feasible, and avoid passing unrelated secrets into its environment.
  9. Document a verified package version and a reproducible upgrade-review procedure for future releases.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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 (8)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The invocation guidance is broad enough to trigger on common requests about delegation, outsourcing, or buying answers, which can cause an agent to route benign user queries into a real-money external marketplace. In this skill's context that is more dangerous than normal because activation can lead to wallet use, public posting of prompts, and financial commitments with limited friction.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The skill instructs users to execute npx @deskcrew/board-runner without pinning an exact package version, so the fetched code can change at any time and may execute arbitrary code during install or runtime. In this skill's context, that code is expected to handle WALLET_KEY and initiate blockchain payments, which materially increases the danger because a compromised or typosquatted package could steal funds or secrets.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

This unpinned npx invocation causes the agent or user to download and run the latest package code at execution time, with no assurance that the reviewed code is the code being run. Because the command is part of a workflow that manages deposit addresses, claims funds, and relies on wallet-linked operations, compromise of the package could expose credentials or redirect financial actions.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Running npx @deskcrew/board-runner without a version pin permits silent upstream changes and supply-chain compromise, turning documentation into an instruction to execute unreviewed code. Here the risk is elevated because the command participates in claiming on-chain deposits, creating an opportunity for malicious code to misuse wallet material, tamper with transaction details, or exfiltrate board credentials.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

This is another instance of dynamic execution of an unpinned npm package, which is a classic supply-chain risk because the exact code run is not fixed. In the context of posting paid bounties and using an API key tied to a spend-capable workflow, malicious package updates could submit unintended tasks, leak secrets, or manipulate payment-related behavior.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

An unpinned npx command in the grading flow allows arbitrary updated package code to run when listing drafts, despite the code not being fixed or authenticated in the skill. Since the same toolchain interacts with sensitive board state and wallet-adjacent workflows, compromise could leak API keys, manipulate decisions, or execute unwanted side effects.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

This command approves a draft and triggers payment, but the CLI itself is fetched dynamically without version pinning, so users may execute altered code at the exact point of authorizing financial outcomes. A compromised package could approve the wrong draft, leak credentials, or modify transaction behavior in ways that directly affect funds and trust in the board.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

This rejection command is also executed through an unpinned package, meaning the code path used for privileged moderation actions is not stable or reviewable. In a system involving wallet-backed bounties and public decision records, malicious package changes could falsify actions, exfiltrate keys, or otherwise compromise board control.

Content

No source excerpt is available for this finding.