Back to skill

Security audit

Hippocampus Openclaw Onboarding

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent onboarding guide, but it asks users to run an unpinned external npm package with a bootstrap credential and local configuration write access.

Review before installing. Use only a verified official package/source, prefer an exact pinned version, avoid putting long-lived tokens in shell history, and confirm the bootstrap token is short-lived, single-use, scoped to the intended workspace, and revocable. After setup, inspect `~/.hipokamp/config.json` and its permissions.

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

Warning
Location
SKILL.md:24
Finding
Unpinned Third-Party Package Execution with Bootstrap Credential Access<![CDATA[ ## Vulnerability Details **File Location**: `SKILL.md`, lines 24-27 **Vulnerability Type**: `T08: Insecure Dependencies` **Risk Level**: Medium ### Vulnerable Code ```markdown 3. Copy the generated bootstrap one-liner. 4. Run: `npx hipokamp-mcp setup --bootstrap-token <token> --gateway <gateway-origin>` 5. Let `hipokamp-mcp` write local config under `~/.hipokamp/config.json`. ``` ### Technical Analysis The documented command invokes `hipokamp-mcp` through `npx` without pinning an exact package version or requiring integrity or signature verification. Depending on the local npm configuration and cache, `npx` may retrieve the package from a remote registry and immediately execute its installation or runtime code with the invoking user's privileges. The package receives a sensitive bootstrap token through a command-line argument and is expected to write configuration beneath the user's home directory. Consequently, a compromised, replaced, or incorrectly identified npm package would have both local code-execution capability and direct access to an authentication credential. The package name `hipokamp-mcp` also differs in spelling from the “Hippocampus” product name used throughout the document. The available project material does not establish that this name is malicious or incorrect, but the discrepancy increases package-identity and typo-squatting uncertainty. No malicious behavior in the package itself can be confirmed because its source is not included in the audited project. ### Attack Path 1. An attacker compromises the npm account or release process for `hipokamp-mcp`, or publishes a confusingly named package that users resolve instead of the intended package. 2. The user follows the onboarding instructions and runs the unpinned `npx hipokamp-mcp` command. 3. `npx` retrieves the mutable package version currently selected by the configured registry. 4. npm executes package-controlled installation or runtime code with the permissions of th ...[truncated 951 chars]
Remediation
<![CDATA[ ## Remediation Suggestions 1. Confirm and document the canonical npm package name, preferably using a verified organizational scope such as `@vendor/package`. 2. Pin the command to a reviewed, immutable version rather than allowing automatic resolution to the latest release. 3. Publish the official registry and source repository URLs so users can verify package provenance. 4. Verify package integrity or signatures before execution and document the expected digest where operationally practical. 5. Prefer a lockfile-based installation workflow in which dependency versions and integrity hashes are committed and reviewed. 6. Avoid placing bootstrap credentials directly in command-line arguments because they may be exposed through shell history or process inspection. Accept the token through protected standard input, an interactive prompt, or a permission-restricted temporary credential file. 7. Ensure bootstrap tokens are narrowly scoped, short-lived, single-use, and immediately revocable. 8. Document the precise files and configuration changes the setup utility performs, and recommend execution from a minimally privileged user account or isolated environment. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Session Persistence

Medium
Category
Rogue Agent
Content
## Preferred Flow

1. Register or sign in through the Hippocampus portal.
2. Create a root OpenClaw agent in the dashboard.
3. Copy the generated bootstrap one-liner.
4. Run:
   `npx hipokamp-mcp setup --bootstrap-token <token> --gateway <gateway-origin>`
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.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding
The skill instructs users to run `npx hipokamp-mcp` without pinning an exact package version. This causes execution of whatever version is currently resolved from the registry, creating a supply-chain risk if a future release is compromised, typo-squatted, or behavior changes unexpectedly.

Missing User Warnings

Low
Confidence
88% confidence
Finding
The skill explicitly tells the user to let the tool write configuration into `~/.hipokamp/config.json` but does not clearly warn that local files will be modified. While not directly exploitative on its own, silent or under-disclosed disk writes can mislead users and reduce informed consent around persistence of credentials and configuration.

Static analysis

No suspicious patterns detected.