Back to skill

Security audit

ClaimCheck

Security checks for vulnerabilities and agentic risk

Overview

ClaimCheck serves a legitimate privacy purpose, but overstates which traffic is protected and when stored secrets are encrypted.

Review before installing. Use a pinned, reviewed release, set SANITIZE_VAULT_PASSWORD on Linux/macOS, and confirm your client's endpoint is actually scrubbed. Use credentials intended for the configured provider and inspect sanitized output before sharing sensitive material. Review the disclosed startup service or logon task if you do not want continuous operation.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Rogue AgentSelf-Modification, Session Persistence
Findings (6)

Chaining Abuse

High
Category
Tool Misuse
Confidence
94% confidence
Finding

Piping the HTTP response directly into bash turns fetched content into executable instructions without an intervening verification boundary. This is the same supply-chain execution issue identified by SC2, not independent evidence of malicious chaining or a second attack.

Content

Scanner excerpt · SKILL.md (reported line 47)May include surrounding context.

md
Linux (per-user, no sudo; x64 + arm64 prebuilt):

    curl -fsSL https://raw.githubusercontent.com/saasclawai-org/ClaimCheck/main/tools/install.sh | bash

Proxy as a systemd --user service: run `./install.sh --proxy --upstream <base-url> [--port 8766]`.

Session Persistence

Medium
Category
Rogue Agent
Confidence
80% 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 49)May include surrounding context.

md
curl -fsSL https://raw.githubusercontent.com/saasclawai-org/ClaimCheck/main/tools/install.sh | bash

Proxy as a systemd --user service: run `./install.sh --proxy --upstream <base-url> [--port 8766]`.

Repo: https://github.com/saasclawai-org/ClaimCheck — full configuration reference in docs/CONFIGURATION.md.

The proxy setup forwards requests and API credentials to a third-party upstream without warning about the privacy and credential-sharing implications.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The workflow instructs users to run a proxy targeting api.z.ai and retain their real API key, which passes through unchanged. It does not warn that sanitized request content and authentication credentials will reach that upstream, while the earlier blanket assurance that nothing leaves the machine (L028–L029) can obscure this distinction.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 126)May include surrounding context.

md
Any OS (manual): run the binary as the proxy:

ClaimCheck --proxy --upstream https://api.z.ai/api/coding/paas/v4/ --port 8766

Point your client's Base URL at the proxy — it mirrors the upstream path:

The documented proxy operation contradicts the explicit claim that ClaimCheck makes no network calls.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The unconditional claims that nothing leaves the machine and there are no network calls conflict with the documented proxy mode, which forwards sanitized requests and authentication to an external provider. In a privacy-focused skill, this can cause users to misunderstand the trust boundary and transmit data under incorrect assumptions; sanitization is explicitly imperfect and some detection tiers are disabled by default.

Content

No source excerpt is available for this finding.

External Script Fetching

Low
Category
Supply Chain
Confidence
97% confidence
Finding

The installation command immediately executes a remotely fetched script from a mutable main branch without an independent integrity check or review step. A compromised repository or malicious upstream change could execute code with the user's permissions, which is particularly consequential on a machine holding the secrets and reversible vault this skill is meant to protect.

Content

Scanner excerpt · SKILL.md (reported line 47)May include surrounding context.

md
Linux (per-user, no sudo; x64 + arm64 prebuilt):

    curl -fsSL https://raw.githubusercontent.com/saasclawai-org/ClaimCheck/main/tools/install.sh | bash

Proxy as a systemd --user service: run `./install.sh --proxy --upstream <base-url> [--port 8766]`.