Back to skill
v1.0.4

Lever

ReviewClawScan verdict for this skill. Analyzed Apr 30, 2026, 3:02 PM.

Analysis

This appears to be a legitimate Lever integration, but it asks the agent to install and run an unpinned global CLI and grants broad OAuth-backed authority to change recruiting records without clear approval limits.

GuidanceBefore installing, confirm that you trust Membrane and the npm CLI package, pin or review the CLI version if possible, connect only a least-privileged Lever account, and require explicit confirmation before the agent creates, updates, archives, or deletes any Lever records.

Findings (8)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Agent Goal Hijack
SeverityLowConfidenceHighStatusNote
SKILL.md
clientAction.agentInstructions (optional) — instructions for the AI agent on how to proceed programmatically.

The skill allows instructions returned from an external connection flow to guide the agent's next programmatic steps. This is purpose-aligned for setup, but those instructions should not override the user's actual request or safety checks.

User impactA remote setup response could influence what the agent does next while connecting Lever.
RecommendationTreat returned agentInstructions as advisory only, and confirm with the user before taking sensitive or account-changing actions.
Tool Misuse and Exploitation
SeverityHighConfidenceHighStatusConcern
SKILL.md
Create User | create-user ... Update Opportunity Stage | update-opportunity-stage ... Archive Opportunity | archive-opportunity ... Delete Interview | delete-interview

The exposed actions can mutate important ATS data, including users, candidate stages, archived opportunities, and scheduled interviews. The artifacts do not define approval, dry-run, rollback, or per-action limits for these high-impact operations.

User impactThe agent could make changes that affect candidates, recruiters, hiring workflows, or scheduled interviews.
RecommendationRequire explicit user confirmation for every create, update, archive, or delete action, and ask the user to specify the exact Lever connection and record IDs before making changes.
Agentic Supply Chain Vulnerabilities
SeverityMediumConfidenceHighStatusConcern
SKILL.md
npm install -g @membranehq/cli@latest

The skill instructs installation of an unpinned npm package globally at runtime. Using @latest creates a moving dependency target and global installation increases host impact if the package or install chain is compromised.

User impactA future or compromised CLI release could be installed and run with the user's local privileges.
RecommendationPin the CLI to a reviewed version, avoid global installs where possible, and verify the package source before installation.
Unexpected Code Execution
SeverityMediumConfidenceHighStatusConcern
SKILL.md
Install the Membrane CLI so you can run `membrane` from the terminal:

```bash
npm install -g @membranehq/cli@latest
```

Although this is presented as an instruction-only skill with no install spec, the runtime instructions require installing and executing a CLI package. This is an install-to-run flow that is not captured by the declared install mechanism.

User impactInstalling the skill may lead to local command execution and persistent global tooling changes.
RecommendationOnly run the install command in an environment where global npm installs are acceptable, preferably after pinning the package version.
Cascading Failures
SeverityMediumConfidenceHighStatusConcern
SKILL.md
Update Opportunity Stage | update-opportunity-stage | Move an opportunity to a different pipeline stage
Archive Opportunity | archive-opportunity | Archive an opportunity with a reason
Delete Interview | delete-interview | Delete a scheduled interview

A mistaken action could propagate through hiring workflows, affecting candidate status, recruiter coordination, or scheduled interviews. The artifacts do not describe containment, review, or rollback safeguards.

User impactOne incorrect agent action could disrupt recruiting processes or affect candidates and hiring teams.
RecommendationUse read-only actions first, verify record IDs and intended changes, and require confirmation before workflow-changing actions.
Permission boundary

Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.

Identity and Privilege Abuse
SeverityMediumConfidenceHighStatusConcern
SKILL.md
Membrane handles authentication and credentials refresh automatically

The integration relies on delegated authentication and automatic credential refresh for Lever through Membrane. That is coherent with the purpose, but the artifacts do not clearly state OAuth scopes, privilege limits, token lifetime, or revocation guidance.

User impactConnecting an account may give the integration continuing access to sensitive Lever data and account actions.
RecommendationUse the least-privileged Lever account possible, review requested OAuth scopes during login, and revoke the connection when it is no longer needed.
Sensitive data protection

Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.

Memory and Context Poisoning
SeverityLowConfidenceMediumStatusNote
SKILL.md
The output contains the new connection id.

The skill uses a reusable connection identifier for the external Lever connection. This is expected for an integration, but the agent should avoid mixing connection IDs between users, tenants, or tasks.

User impactUsing the wrong saved connection context could cause the agent to read or change data in the wrong Lever account.
RecommendationConfirm the connection ID, tenant, and account context before accessing or modifying Lever records.
Insecure Inter-Agent Communication
SeverityMediumConfidenceHighStatusNote
SKILL.md
This skill uses the Membrane CLI to interact with Lever. Membrane handles authentication and credentials refresh automatically

The skill uses Membrane as an external broker between the agent and Lever. That is disclosed and purpose-aligned, but sensitive recruiting data and authentication flows may pass through this provider boundary.

User impactLever candidate, job, user, and interview data may be accessed through a third-party integration layer.
RecommendationReview Membrane's access model and privacy terms, and avoid sending unnecessary sensitive candidate data through the integration.