anyshift-io/annie-skills is a public Claude Code plugin from Anyshift. It packages ready-made skills that teach an AI agent how to install, authenticate, and drive every documented Anyshift surface: the Annie CLI for human users, and the remote MCP server for agent harnesses.
The plugin is the agent-facing companion to this documentation. The CLI and MCP pages describe what each surface does for a human or a harness; the plugin teaches a Claude Code agent when to reach for each one, how to chain the calls, and what to do when a call fails.
Install
Add the Anyshift marketplace
From a Claude Code session:/plugin marketplace add anyshift-io/claude-plugins
Install the annie-skills plugin
/plugin install annie-skills@anyshift
The plugin’s skill descriptions self-register inside Claude Code, so the agent picks them up automatically the next time you start a session. No config edits required.
You can also install the plugin directly from the GitHub source, see the repo’s README for the manual path.
What ships in the plugin
| Skill | Status | What it teaches |
|---|
annie | Shipping by end of May 2026 | Drive the remote MCP server end-to-end: discover resource types, search the graph, track changes, audit a single resource’s timeline, inspect resources in detail, recover from auth or staleness failures, and point human users at the Annie CLI when they want to chat with Annie themselves. |
The plugin grows as the surface of Annie and the Anyshift MCP grows. Every skill ships with a self-contained SKILL.md runnable end-to-end in Claude Code, two worked examples against realistic scenarios, and an explicit failure-modes section.
Use it from Claude Code
Once installed, describe the task in natural language and Claude Code loads the annie skill automatically:
Investigate the impact of yesterday's deploy on the payments stack.
Walk me through the timeline of vpc-prod-main this week.
What changed in our AWS account between yesterday at 9am and now?
The skill picks the right tool, chains follow-up calls, cites resources by their stable id, and synthesises the answer. The user sees a clean answer, not the JSON.
The skill is read-only by design. If the user asks the agent to apply a change, the skill hands back to Terraform or your change-management flow rather than inventing write tools.
Adjacent skills
| Repository | What it covers |
|---|
anyshift-io/sre-skills | Vendor-neutral SRE methodology skills (oncall handover, incident investigation, postmortem authoring). Annie-optional. |
anyshift-io/awesome-sre-skills | Curated index of SRE skills, MCP servers, and reading material from the broader ecosystem. |
annie-skills is the vendor-specific layer that assumes Annie is available. sre-skills is the methodology layer that runs without Annie and pulls Annie in as optional context. They compose.
Source and contributions
The plugin source lives at anyshift-io/annie-skills under Apache 2.0. Issues and pull requests are welcome, see CONTRIBUTING.md for the contribution guide and the per-skill quality bar (self-contained SKILL.md, two worked examples, explicit failure-modes section, copy-pasteable install snippet).