AugmentClaude

Solo Artifacts

Generate self-contained HTML artifacts for Solo tasks and threads.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up — so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/solo-artifacts-solo-agent/ — the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Safe default: works for every skill.

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates — they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description — no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the “What this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Use when a Solo task or thread should become an interactive, reviewable, self-contained HTML artifact for progress/status, review/decision, or comparison/leaderboard work inside Solo.

What this skill does

solo-artifacts

Warm Archive fork of work-canvas: produce or publish one self-contained HTML file that renders a Solo task/thread for a human to understand and act on. Prefer the real deliverable over a summary page.

Workflow

  1. Publish Existing Deliverable if the current task/run names one.

    Trust only explicit current-task evidence, such as:

    Deliverable: ./path/to/result.html
    

    You may also use a file path named in the current task/thread, or a file you personally created for this task now. Do not scan the workspace by newest file, mtime, or vague filename match; old workspace files can be stale. If several files are named, prefer the product/final deliverable over review/report/panel files unless the user explicitly asks for the review page. Do not switch to review-decision just because the task is in_review.

    • Existing .html: publish it directly.

      solo artifact publish --task <task-id> --mode <latest|final> --file <path-to-html>
      
    • Existing .md, .txt, .json, .csv, or other text/data: create a small self-contained HTML viewer that renders the exact source content, labels the source file near the top, then publish that wrapper. Do not summarize or rewrite the content.

    • Multiple deliverables or a directory: create a small index artifact with explicit links/file names and a short note about what each file is.

    If a real deliverable exists, stop here.

  2. Pick the artifact type and read its reference:

    The task needs to...TypeRead
    Show where long/messy work stands, what broke, what's nextProgress / status reportreferences/progress-report.md
    Review analysis or implementation and decide accept/reject/next stepReview / decision memoreferences/review-decision.md
    Weigh options head-to-head and pick a winnerComparison / leaderboardreferences/comparison.md

    If blended, lead with the primary type and borrow blocks from the other.

  3. Assemble from the template. Copy assets/starter.html, then:

    • Replace [[PASTE base.css]] with the full contents of assets/base.css.
    • Replace [[PASTE interactions.js]] with only the needed modules from assets/interactions.js; pasting the whole file is fine because modules self-guard.
    • Fill the header, What needs your input, and body sections using the component recipes in the chosen reference.
  4. Embed everything: inline CSS/JS, embed media as data: URIs, no CDN or external requests.

  5. Publish back to Solo:

    solo artifact publish --task <task-id> --mode <latest|final> --file <path-to-html>
    

Non-Negotiables

  • Keep work-canvas structure and interactions intact; use Solo's default Warm Archive visual skin.
  • Real deliverables beat summaries. Never replace an explicit deliverable file with a conversation recap.
  • Surface only real human decisions. If nothing needs the user, say so.
  • Add a legend wherever color, letters, or symbols encode meaning.
  • Never modify Solo/source data from the page. Show paste-ready output with copy buttons.
  • Include a provenance footer with task id, thread/channel scope, agent, model, and generation date when available.

Files

  • assets/starter.html — the skeleton to copy.
  • assets/base.css — class-compatible Solo Warm Archive design system.
  • assets/interactions.js — optional vanilla-JS modules: print, tabs, tables, lightbox, copy, scrollspy, persist.
  • references/*.md — work-canvas section order, component recipes, and guardrails.

Related skills