Covalent

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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.

What this means

Installing a global CLI gives that package local execution ability, and the latest version may change over time.

Why it was flagged

The skill directs installation of an unpinned latest version of an external global CLI. This is expected for a Membrane integration, but the CLI package code is not part of the provided artifacts.

Skill content
npm install -g @membranehq/cli@latest
Recommendation

Install the CLI only from the official Membrane source, avoid elevated privileges, and consider pinning or reviewing the package version in sensitive environments.

What this means

The agent may be able to use the authenticated Membrane/Covalent connection according to the permissions granted during login.

Why it was flagged

The integration uses Membrane login and credential refresh. This is purpose-aligned, but it grants delegated account access that users should understand before connecting.

Skill content
Membrane handles authentication and credentials refresh automatically ... membrane login --tenant --clientName=<agentType>
Recommendation

Authenticate only to the intended Membrane tenant and Covalent connection, review requested permissions, and revoke the connection when no longer needed.

NoteMedium Confidence
ASI01: Agent Goal Hijack
What this means

Remote setup responses could influence the agent's next steps if treated as authoritative.

Why it was flagged

The setup flow may return external service-provided instructions for the agent. This can be useful, but such instructions should remain subordinate to the user's request.

Skill content
clientAction.agentInstructions (optional) — instructions for the AI agent on how to proceed programmatically.
Recommendation

Treat returned agentInstructions as untrusted guidance and require explicit user approval before installs, credential changes, or data-mutating actions.

What this means

Users may not know whether the skill is limited to data retrieval or may also manage account/business objects.

Why it was flagged

The description mentions broad management objects, while the visible body focuses on blockchain-data retrieval. This mismatch could confuse users about the skill's actual authority.

Skill content
Covalent integration. Manage Organizations, Projects, Pipelines, Users, Goals, Filters. ... Covalent is a unified API that provides access to blockchain data
Recommendation

Verify the exact Membrane actions and permissions before approving any management or mutation operation.