Design Uniswap Hook
v0.1.0Design a Uniswap V4 hook architecture without code generation. Use when user wants to plan a hook, understand which callbacks to use, or review an architecture before building. Returns a design document, not code.
⭐ 0· 693·0 current·0 all-time
by@wpank
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (design Uniswap V4 hook architecture) align with the actual behavior: it is instruction-only, requests parameters about hook behavior/constraints/integrations, and delegates to a 'hook-builder' subagent for producing a design document. The declared allowed tools (a Task subagent and an mcp__uniswap__get_supported_chains helper) are relevant to the stated purpose.
Instruction Scope
SKILL.md stays within scope: extract parameters, invoke a design-only hook-builder subagent, and return a structured design doc. It explicitly forbids code generation and file writes. Minor note: the allowed-tools list includes Read/Glob/Grep which could be used to read files if the agent were instructed to do so, but the SKILL.md does not instruct arbitrary file reads or access to unrelated system data.
Install Mechanism
No install spec and no code files — instruction-only skill. This is the lowest-risk model and matches the declared behavior (design document only).
Credentials
The skill requests no environment variables, no credentials, and no config paths. That is proportionate for a design/documentation task and matches the SKILL.md claims.
Persistence & Privilege
always is false and the skill does not request persistent presence, system config changes, or other skills' credentials. disable-model-invocation is false (normal), and there is no instruction to modify agent-wide settings or store tokens.
Assessment
This skill appears coherent and low-risk: it only produces a design document and asks for no secrets or installs. Before using it, confirm you are comfortable with the agent delegating to the 'hook-builder' subagent (verify that subagent's behavior/policies if possible), avoid pasting private keys or proprietary secrets into prompts, and if you need absolute assurance, ask the skill to explicitly confirm it will not write files, perform network calls, or generate code before proceeding.Like a lobster shell, security has layers — review code before you run it.
latestvk97b9tpd84ek4jzsyvst6g7mm580xskm
License
MIT-0
Free to use, modify, and redistribute. No attribution required.