A resonance profile is a standard of judgment, learned from examples, that can score any content — turning subjective quality into something measurable. Per-trait scores 0–100, deterministic, sub-second, no LLM in the loop. Try it now:
From your terminal
POST /p/{profile} — pipe text or a URL. JSON response by
default, add Accept: text/plain for the formatted
table. No auth required.
Full API docs →
Concrete Usage Demonstration ██████████████████░░ 92 ●●● Strong Copy-Pasteable Setup ██████████████████░░ 88 ●●● Solid Hook Speed ██████████████████░░ 92 ●●● Strong Hype-Free Credibility ███████████████████░ 95 ●●● Strong Problem Framing ██████████████████░░ 92 ●●● Strong Progressive Disclosure ██████████████████░░ 92 ●●● Strong Structural Scannability ██████████████████░░ 92 ●●● Strong Value Proposition Clarity ███████████████████░ 96 ●●● Strong ──────────────────────────── ──────────────────── ─── Composite ██████████████████░░ 92 ●●● Strong ●●● high ●●○ moderate ●○○ low confidence ↑n headroom to Solid
Actionable Takeaways ████████████████░░░░ 78 █████████████████░░░ 83 ▲ + 6 Usable guidance
Grounded Motivation ██████████████████░░ 92 ██████████████████░░ 89 ▼ -3 Problem-anchored
Honest Specificity ███████████████░░░░░ 77 ████████████████░░░░ 80 ▲ + 3 Evidence-backed precision
Incremental Complexity ███████████████░░░░░ 74 ██████████████████░░ 88 ▲ + 14 Layered build-up
Narrative Throughline ██████████████████░░ 92 █████████████████░░░ 86 ▼ -6 Coherent arc
Progressive Concreteness ██████████████████░░ 89 ██████████████████░░ 90 ▲ + 1 Concrete-first
──────────────────────── ──────────────────── ─── ──────────────────── ───
Composite █████████████████░░░ 83 █████████████████░░░ 86 ▲ + 3 ✓ improved
In your editor
Streamable HTTP MCP server. Three tools:
score,
list_profiles, and
list_traits.
Works with Claude Code, Cursor, Windsurf, or anything that speaks MCP.
# Claude Code
$ claude mcp add --transport http qed https://qed.systems/mcp
Or add to
.mcp.json for Cursor,
Windsurf, etc.:
// .mcp.json
{
"mcpServers": {
"qed": {
"type": "http",
"url": "https://qed.systems/mcp"
}
}
}
Claude Code plugin
Guided skills on top of MCP —
/qed:evaluate scores content,
/qed:improve iterates edits until traits converge.
Install from the
plugin marketplace.
# Add the marketplace
$ /plugin marketplace add onebit0fme/qed-plugins
# Install the qed plugin
$ /plugin install qed@qed-plugins
Profiles
Each profile has traits — dimensions of quality learned from positive and negative examples (typically 50–90 per trait). POST text, get back 0–100 per trait plus a composite score that penalizes any single weak dimension.
qed.compelling-readme
What makes a README that developers actually read and act on.
Loading...
qed.technical-writing
Clear, precise, useful technical documentation and articles.
Loading...
Quickstart
Score a URL in one line — no install, no signup, no API key:
$ curl -s https://qed.systems/p/qed.compelling-readme \
-H "Content-Type: application/json" \
-d '{"content":"https://github.com/astral-sh/uv"}'
What scores are (and aren't)
A score measures how well content resonates with a profile's learned standard — not absolute quality. The standard is only as good as its examples. Deterministic by design: no LLM variance, but also no free-text explanations (just numbers). Best on English prose ≥200 words; short texts and non-prose score unreliably. Not a grammar checker, not a style linter, not AI detection.