--- title: "Zed Replaces Its Rules Library With Skills, Cutting AI Context Usage by 90%" subtitle: "Zed v1.4.2 ships Skills, replacing the old Rules Library with a smarter, on-demand instruction system backed by an open cross-editor standard" url: "https://alphasignal.ai/news/zed-replaces-its-rules-library-with-skills-cutting-ai-context-usage-by-90" author: "AlphaSignal Newsroom" author_url: "https://alphasignal.ai/editorial-team" company: "Zed" company_url: "https://alphasignal.ai/company/zeddotdev" published: "2026-05-27T21:19:21.000Z" updated: "2026-06-08T19:27:41.939Z" type: "News" topics: - "Development" - "Agents" subtopics: - "Code Agents" - "Code Agents" - "Agent Frameworks" - "Tool Use" upvotes: 1063 read_time: "6 min" source_url: "https://zed.dev/docs/ai/skills" comments: 0 --- # Zed Replaces Its Rules Library With Skills, Cutting AI Context Usage by 90% > Zed v1.4.2 ships Skills, replacing the old Rules Library with a smarter, on-demand instruction system backed by an open cross-editor standard [Zed](https://zed.dev) just shipped v1.4.2 with a feature that quietly changes how you collaborate with your AI agent: **Skills**. Think of them as reusable instruction packages, each one a folder with a `SKILL.md` file, that teach the agent a specialized workflow. Test-driven development, PDF processing, your team's internal code standards, a deploy procedure you never want triggered by accident. You define it once, and the agent knows when to reach for it. ## Out with Rules, In with Skills The old Rules Library has been removed and replaced with skills support. Upgrading requires swapping the old `@rule` syntax to `@skill` in agent chats, but all other configuration files carry over without forcing a manual rewrite. Existing rules are migrated automatically, so there's no cliff edge here. A global `AGENTS.md` file now sits alongside `settings.json` and feeds instructions into every project automatically, preventing context drift when jumping between unrelated repositories. That's a separate but complementary addition: Skills handle task-specific knowledge, while `AGENTS.md` handles the always-on baseline. ## The Architecture Behind It The key design decision here is **progressive disclosure**, a loading strategy borrowed from the [Agent Skills open specification](https://agentskills.io/specification). Rather than dumping all your instructions into the system prompt at once, the agent only loads what it needs, when it needs it. The system breaks knowledge loading into three levels: L1 metadata (~100 tokens per skill), which is just the skill name and description loaded at startup so the agent knows what's available; L2 instructions (under 5,000 tokens), the full skill body loaded only when the agent explicitly activates a skill; and L3 resources, external reference files like style guides or API specs, loaded only when the skill's instructions require them. An agent with 10 skills starts each call with roughly 1,000 tokens of L1 metadata instead of 10,000 tokens in a monolithic prompt, translating to roughly a 90% reduction in baseline context usage. Each skill costs roughly 30-50 tokens at startup. An agent with 100 skills installed uses approximately 3,000-5,000 tokens at session start for skill metadata, and the full instructions only load for the skill that activates on a given task. ## How to Use It There are several ways to create a skill in Zed: - Use the built-in `/create-skill` command and the agent walks you through the process. - Open the Skill Creator from the Agent Panel using `cmd-alt-l` / `ctrl-alt-l`, or via the command palette with `agent: open skill creator`. It opens a window where you fill in the skill's name, description, scope (global or project-local), body, and optionally toggle `disable-model-invocation`. - Import a skill from an existing GitHub Markdown file via the `agent: create skill from url` action in the command palette. A minimal `SKILL.md` looks like this: ``` --- name: tdd-workflow description: Use when writing new features with test-driven development. Guides the agent through red-green-refactor cycles. --- ## Instructions 1. Write a failing test first 2. Write the minimum code to pass it 3. Refactor without breaking tests ``` You can share a skill with a teammate without hosting it anywhere. Clicking the link icon on a skill row copies a self-contained `zed://skill?data=…` link to your clipboard that embeds the full `SKILL.md` contents (base64url-encoded), so the recipient doesn't need access to your project or any registry. ## Autonomous vs. Manual Invocation By default, the agent picks up skills on its own. It scans the catalog of installed skills and calls the `skill` tool when a task matches a skill's description. But you can also invoke skills manually: - Type `/skill-name` as a slash command in the message editor - Use `@skill` and select from the completion menu For workflows you never want the agent triggering automatically, like deploy or release procedures, set `disable-model-invocation: true` in the skill's frontmatter. The skill stays available as a slash command, but won't fire unless you explicitly call it. Zed also added a user confirmation requirement before letting tools modify files inside `.agents/skills/` or `~/.agents/skills/`, so skills installed by any agent are protected from unsolicited edits. ## Where Skills Live Zed loads skills from two locations, giving you both personal and team-scoped control: | Scope | Path | When it applies | | --- | --- | --- | | Global | `~/.agents/skills/` | Every project | | Project-local | `/.agents/skills/` | Only the current project | If a global and project-local skill share the same name, the project-local one wins. This lets teams define defaults that individuals can override. Note that project-local skills only load from trusted worktrees, preventing a cloned repo from silently injecting instructions into your agent. ## A Growing Open Ecosystem The bigger story here is that Zed is joining a cross-editor standard. The Agent Skills specification at agentskills.io is an open standard. A `SKILL.md` file you build for Claude Code works identically in OpenAI Codex CLI, Gemini CLI, GitHub Copilot, Cursor, VS Code, and any other platform that has adopted the standard, which now includes more than 26 tools. Skills solve the problem of packaging procedural knowledge and company-, team-, and user-specific context into portable, version-controlled folders that agents load on demand. The community registry at [skills.sh](https://skills.sh) already hosts over 91,000 total installs across skills from Anthropic, Vercel, Microsoft, Firebase, Supabase, Sentry, and hundreds of community contributors. Popular ones include `frontend-design` (391K installs), `tdd`, `pdf`, and `webapp-testing`. ## What to Watch The current limitations are worth knowing before you go deep: - **Flat layout only.** Skills must be direct children of the skills root. Nested folders like `~/.agents/skills/group/my-skill/` are not discovered. - **50KB catalog budget.** The total size of all skill names and descriptions is capped at 50KB. Skills that don't fit are dropped from the catalog with a warning. - **Zed Agent only.** Skills apply to the Zed Agent. External agents and terminal threads have their own instruction systems. - **Live reload works.** Adding, removing, or editing a `SKILL.md` takes effect immediately without restarting your session, though changes to a skill's name or description will invalidate the model's prompt cache for the current session. The shift from a bespoke Rules Library to an open, cross-platform standard is the real move here. Skills you write today for Zed will work in Claude Code, Codex, and whatever agent tooling you pick up next year. That portability is what makes this worth investing time in now. ## Sources - [X post by @zeddotdev](https://x.com/zeddotdev/status/2059746303068762493) - [Skills | Agent Skills - Zed (Official Docs)](https://zed.dev/docs/ai/skills) - [Release v1.4.2 · zed-industries/zed](https://github.com/zed-industries/zed/releases/tag/v1.4.2) - [The Agent Skills Directory - skills.sh](https://skills.sh) - [Agent Skills Explained: How SKILL.md Files Work - Firecrawl](https://www.firecrawl.dev/blog/agent-skills) - [Agent Skills Overview - agentskills.io](https://agentskills.io/home)