v1.0.0

First public release, under the MIT License.

• 22 skills for Clarity and Stacks development: clarity:start (router) plus scholar, security, auditor, testing, fuzzing, optimizer, deployment, upgrades, governance, epochs, assets, sip-compliance, protocols, ecosystem, oracles, staking, bitcoin-proofs, wallets, environments, apis and system-integration.
• One repository for every host. The plugin lives in plugin/: Claude Code and the Claude apps (.claude-plugin/marketplace.json, install clarity@clarity-skills) and Codex and ChatGPT (.agents/plugins/marketplace.json, per-skill agents/openai.yaml) list its skills under the plugin namespace (clarity:start, clarity:scholar, clarity:fuzzing). Its SessionStart hook points sessions inside Clarinet projects at clarity:start, the skills for common tasks and the router's pre-finish checklist.
• Hosts that list skills by folder name read the top-level skills/, a copy generated from plugin/skills/ whose skills carry a clarity- prefix (clarity-start, clarity-security), so they cannot clash with other skill sets: Gemini CLI (gemini-extension.json, with GEMINI.md loading the router as context), Google Antigravity, GitHub Copilot and VS Code, Factory Droid, Devin, OpenCode and the skills CLI. The install guide gives the steps per host; llms.txt indexes the skills.
• Skills carry rules, decisions, exact call shapes and short fragments; the must-follow content of each skill is in its SKILL.md, and references keep lookup material. Each SKILL.md body stays within 24,000 bytes, the most Antigravity loads.
• Short skill descriptions: every session carries about 1,940 tokens of skill names and descriptions (measured with claude plugin details), under the 2,000 at which Claude Code's official marketplace flags a plugin.
• Skills that run bundled offline scripts pre-approve them for Claude Code (allowed-tools: check_registry.py, lookup_function.py, compare_costs.py, inspect_costs.py); the two scripts that reach the network still ask. Eight skills state their toolchain in compatibility.
• Directory listing metadata: logo and composer icon in plugin/assets/, category, brand color and links for the OpenAI plugin directory (extensions.com.openai.interface) and the Anthropic directory (.claude-plugin/plugin.json).
• Icons in black and gold: the plugin shows the three Clarity brackets, and each skill has its own glyph (assets/icon-small.svg, assets/icon-large.png), which Codex shows in its skill list (agents/openai.yaml) and the README shows in the skills table.
• SECURITY.md (private reports, and what the skills, the startup hook and the bundled scripts do on your machine), CONTRIBUTING.md, a code of conduct, an AGENTS.md guide for agents working in the repository (imported by CLAUDE.md), bug report and addition request forms, a pull request template, and a CI workflow that checks the skills, the generated copy and every manifest on each push and pull request.
• Versioned releases: main holds the latest release, tagged v<version> with a GitHub Release, and work lands on develop first. Pin a tag to stay on one release.
• Baseline: Stacks epoch 4.0 (mainnet), Clarity 1-6, Clarinet 3.26.0, Clarinet SDK 3.26.0, Vitest 5.0.5, Clarigen 4.1.7, @stacks/transactions 7.6.0, Node 22.18 or later. Protocol (195 contracts) and system (18 records) address registries verified on chain, oldest check 2026-10-20 (dates in each owning skill).

Full Changelog: https://github.com/cybasecurity/clarity-skills/commits/v1.0.0