Uniswap Design Integration

PassAudited by ClawScan on May 1, 2026.

Overview

This is a coherent instruction-only Uniswap architecture skill with no code, credentials, or persistence, though users should notice that it can delegate project context to a subagent and has file edit tools available.

This skill appears safe to install as an instruction-only design aid. Before using it on a private repository, avoid sharing secrets, supervise any proposed file edits, and prefer a pinned or registry install source if you need reproducible provenance.

Findings (3)

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.

Low

#ASI02: Tool Misuse and Exploitation

What this means

The agent could potentially create or edit local files if it chooses to use the granted tools.

Why it was flagged

The skill grants file mutation tools even though the documented workflow is to present an integration blueprint. This does not show unsafe behavior, but users should supervise any proposed file changes.

Skill content

allowed-tools: >-\n  Read, Write, Edit, Glob, Grep,\n  Task(subagent_type:integration-architect),\n  mcp__uniswap__get_supported_chains

Recommendation

Review and approve any file edits explicitly; if the skill is only needed for advice, consider limiting it to read/search and response-only tools.

Low

#ASI07: Insecure Inter-Agent Communication

What this means

Project details or code snippets included in the request may be shared with the delegated integration-architect agent.

Why it was flagged

The skill delegates to another agent and may pass full project context, including existing codebase details. This is purpose-aligned for architecture analysis, but it is still a data-boundary point users should notice.

Skill content

Invoke `Task(subagent_type:integration-architect)` with the full context.

Recommendation

Avoid including secrets or unrelated private files in the prompt/context, and confirm that subagent delegation is acceptable for the project.

Info

#ASI04: Agentic Supply Chain Vulnerabilities

What this means

Installing from a moving branch may fetch different skill text in the future.

Why it was flagged

The README documents installation from a GitHub branch path rather than a pinned commit or release. For this instruction-only skill the impact is limited, but pinning improves provenance.

Skill content

npx skills add https://github.com/wpank/Agentic-Uniswap/tree/main/.ai/skills/design-integration

Recommendation

Prefer installing from the registry version or a pinned commit/release when reproducibility matters.