Phoenix perpetuals on Solana for agents: exact lots and ticks, market bands, account and on-chain reads, delegation, order outcomes and the optional signer.
TypeScript toolkit for Phoenix perpetuals: exact order identifiers, lot and tick quantization, REST, on-chain headers, pending writes and an optional Node signer.
QFEX exchange mechanics for agents: HMAC authentication, REST and WebSocket APIs, decimal quantization, order outcomes, account reads, leverage and limits.
Lighter knowledge base for agents (zkLighter and the robinhoodchain instance): the Python SDK signer, auth tokens, order ids and reflection lag, minimums, listings and write limits.
TypeScript toolkit for Lighter (zkLighter and robinhoodchain): REST reads, exact order ids above 2^53, the 40/60 s write window, pending-write memory, a signer client and the Python SDK signer itself.
A skill is a folder with a SKILL.md file and reference material that an agent such as Claude Code loads when a task touches its subject. The catalog also holds SDKs, MCP servers and guides.
How do I install a skill?
Download the ZIP from the skill page, compare its SHA-256 with the checksum shown there, and unpack the archive into .claude/skills/<name>/ in your project. We never show commands that download something and pipe it straight into a shell.
How is published content checked?
The registry pins a public GitHub commit, checks metadata, license and risky content, then builds an immutable ZIP with a published checksum. A maintainer reviews the pull request. External dependencies and linked sites are not checked — the report says so plainly.