mcpField
A
artifact-server

artifact-server

github.com/plannotatorProductivity24185 installsUpdated 5w ago
Repository

About artifact-server

A skill for routing Artifact Server requests to artifact workflows or server administration. It distinguishes local file work through the `artifactserver` CLI from server-data tasks through MCP, and requires success to be verified from structured results.

Key features

  1. 01Route requests to the relevant workflow
  2. 02Separate artifact work from administration
  3. 03Handle local files with the CLI
  4. 04Use MCP for server-side data
  5. 05Verify results and report artifact version details

Use cases

  1. 01Publish, update, or share artifacts
  2. 02Manage projects, comments, and artifact versions
  3. 03Install, inspect, or repair a server
  4. 04Investigate target, authentication, or compatibility issues

MCP support

For work involving only server data, the skill uses the connected Artifact Server MCP; local file operations use the `artifactserver` CLI.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/plannotator/artifact-server --skill artifact-server

After installing, type/artifact-serverin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

agents/
openai.yaml
evals/
evals.json
references/
artifact-operations.md
failures-and-compatibility.md
server-operations.md
target-selection.md

Details

From the project README

Artifact Server

Use one skill for Artifact Server work. Route the request before loading detailed instructions.

Route the request

  • For publishing, updating, opening, sharing, commenting, tagging, comparing, restoring an artifact version, or managing projects, read artifact operations. Read target selection when the server or project is not already exact.
  • For an explicit request to install, deploy, upgrade, reconfigure, back up, restore, inspect, repair, or delete an Artifact Server installation, read server operations. Do not load or follow administrator instructions for routine artifact work.
  • Read failures and compatibility only after a failure, incompatibility, authentication problem, or ambiguous target.

If a request includes both artifact work and server administration, keep the steps separate and state which installation is affected before changing it.

Keep these boundaries

  • Use the installed artifactserver CLI for files on the user's machine. Use the connected Artifact Server MCP for work that needs only server data.
  • Never replace a real file with inline HTML, base64, invented bytes, or a local path sent to a remote MCP server.
  • Never read, print, request in chat, or write bearer tokens, API keys, cookies, or refresh tokens.
  • Keep public sharing off unless the user explicitly requests it.
  • Verify success from structured CLI, MCP, HTTP, or deployment health results. Never infer success from a command that merely started.

Report the result

After publication, give the full-screen review-and-comment link first. Give the raw immutable artifact link second when it is useful. Include the server, project ID, artifact ID, and exact version ID in technical details rather than making the user reconstruct them.

Comments

Markdown supported