What does this skill do, and when should you use it?
project-artifact is designed for projects with several parallel or dependent workstreams, such as migrations, research efforts, launches, and organizational initiatives. It produces a tabbed status page covering the overview, success criteria, workstream sequence, next steps, background, plans, risks, open questions, and decisions when relevant. The page is published with the built-in Artifact tool to a default-private claude.ai page and can be refreshed at the same URL. For PR-driven software projects, the optional swe.md guidance adds PR numbering and status details.
- Reads the project goal, workstreams, owners, dates, and related documents.
- Creates or loads per-project config.md and page.html files in the plugin data directory.
- Generates a self-contained HTML status page from the supplied template and embeds a JSON state block for change tracking.
- Publishes or redeploys the page with the built-in Artifact tool.
- On refresh, re-gathers live state and reports only the delta.
- For software projects, the PR workflow guidance includes reading PR state with gh/git.
- A project lead needs a shareable overview for a software migration with multiple workstreams.
- A research team needs a living page covering goals, phases, risks, and unresolved questions.
- A launch team needs to show collaborators the workstream order and concrete next actions.
- An organizational initiative needs a persistent view of completed, active, and blocked work.
- Single-PR changes (explicitly excluded in SKILL.md)
- Public documentation (explicitly excluded in SKILL.md)
- Environments without a claude.ai login or the built-in Artifact tool
How do you install this skill?
- The core publishing flow depends on the built-in Artifact tool, claude.ai login, and live external services; verify account access, network reachability, and that the content contains no sensitive data before use.
- The skill writes to CLAUDE_PLUGIN_DATA and may invoke gh/WebFetch, but does not define least-privilege credential scope or a complete recovery procedure after failure.
- No skill-specific tests or representative artifacts are supplied; validate first publication, refresh, conflict handling, and fetch failures on a low-risk project.
- Shell / CLI
- Network access
- Local filesystem
Built-in Artifact toolclaude.ai login
Install from the anthropics/claude-plugins-official marketplace through Claude Code's plugin system; the exact marketplace plugin name is not specified in the supplied material.
/plugin install {plugin-name}@claude-plugins-officialHow do you use this skill?
Once installed, send your agent any of these to trigger it:
- Generate a shareable status page for our multi-workstream migration project
- Refresh the migration project artifact and tell me only what changed
- Build a status page for our research effort covering goals, phases, risks, and open questions
- Redeploy the launch project artifact to the same URL
After installation and a claude.ai login, invoke /project-artifact <project>. The first run gathers project material, generates the page, and publishes it; later runs can use “refresh the artifact” or “update the status page” to re-gather state and redeploy the same URL. It is not intended for single-PR changes or public documentation.
What are this skill's strengths and limitations?
- Provides a consistent status-page structure for overview, workstreams, and next steps.
- Produces self-contained pages with inline CSS, content, and required scripts.
- Supports same-URL refreshes and delta reporting through an embedded state block.
- Pages are private by default and can be shared from the viewer.
- Requires the built-in Artifact tool and a claude.ai login.
- The user must manually share the page with teammates after the first publication.
- Not designed for single-PR changes or public documentation.
- The supplied source contains no evidence of a test suite, platform test results, or implemented recovery behavior.
- The exact marketplace plugin name is not provided.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Project Status Artifact this page ✓ Anthropic · Official | 61 · Recommended | ★ 38k | 11d ago | Apache-2.0 |
| deepPPT | 67 · Recommended | ★ 28 | 2mo ago | MIT |
| huashu-md-html Publishing Pipeline | 57 · Use with care | ★ 915 | 12d ago | MIT |
| Hebrew RTL Formatting Skill | 55 · Use with care | ★ 11 | 7mo ago | — |
| Linear Claude Skill | 55 · Use with care | ★ 135 | 3mo ago | MIT |
How did FollowSkills review this skill?
The skill explicitly requires default-private publication, read-only gathering, treating fetched content as untrusted data, HTML entity encoding, script-termination protection, and no edits to external projects. These support strong data-flow and side-effect controls. Deductions apply because publishing, WebFetch, gh, and persistent-directory writes have external or local effects without explicit step-by-step user confirmation, sensitive-data classification, least-privilege detail, or a clear rollback/deletion procedure.
The workflow, refresh state block, stale-data handling, conflict retry, and abnormal-input guidance make the main path plausible from static inspection. Points are deducted because it depends on the built-in Artifact tool, claude.ai login, WebFetch, gh, and companion template files without version pinning, a test suite, or reproducible diagnostic outputs; static calibration caps this dimension at 10.
Audience, trigger scenarios, refresh semantics, and the non-fit boundary for single-PR work and public documents are clear, with coverage for software, research, launches, and organizational projects. Points are deducted because Chinese-language support is not specified and mainland-China reachability of the claude.ai-dependent publishing path is unproven; complex-project adaptation also relies on user judgment.
The skill documents information architecture, progressive tab selection, configuration format, template conventions, the SWE extension, limitations, and persistent data location. Repository-provided Apache-2.0 licensing and official provenance support governance evidence. Points are deducted because the skill lacks its own version, changelog, named maintenance responsibility, or explicit update path, and installation prerequisites and troubleshooting remain incomplete.
It defines a directly shareable project-status page, stable-URL refreshes, delta reporting, responsive-layout review, and state tracking, so it could materially reduce repeated status synthesis. Points are deducted because no representative artifact or execution evidence is included, utility depends on the Artifact tool, live sources, and template completeness, and static calibration caps this dimension at 7.
SKILL.md and swe.md provide auditable commands, fields, state-block structure, encoding rules, and refresh logic; repository CI offers limited governance context. Points are deducted because the CI does not cover this skill's key paths and there are no skill-specific tests, third-party execution results, or independently reproducible conclusions; static calibration caps this dimension at 5.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →