A2A Protocol · Open Source

agentcard-disco

Score and optimize your A2A Agent Cards for discoverability.

A CLI tool and hosted API that tells you exactly why your agent isn't being found — and how to fix it.

agentcard-disco

            $ agentcard-disco score ./agent-card.json --deep
            # Scoring Agent Card…
             
              Metadata Richness    ████████████████████  28/30
              Semantic Specificity ████████████████────  22/30
              Search Alignment     ████████████████    20/20
              Completeness         ████████████────────  16/20
             
              ─────────────────────────────────────
              Total Score          86 / 100  ◆ Excellent
             
              ↳ 3 suggestions available. Run --suggest
          

How it works

Three commands. Zero guessing.

Point it at any Agent Card file or URL. Get a scored breakdown and actionable fixes in seconds.

Step 01 — Install

Install the CLI

Available on PyPI. Install globally or into a virtual environment. Optional [deep] extra for AI scoring.

$ pip install agentcard-disco

Step 02 — Score

Score your card

Point at a local file or any public URL. The --deep flag adds AI-powered analysis.

$ agentcard-disco score ./agent-card.json --deep

Step 03 — Fix

Get suggestions

AI-generated, specific improvement suggestions ranked by impact. Not generic — tailored to your card's actual gaps.

$ agentcard-disco suggest ./agent-card.json

Scoring tiers

Two ways to score. Both free.

Heuristic scoring runs offline with no keys. AI scoring runs on our servers — just register for a free key.

Tier 1

Heuristic Scoring

Runs entirely on your machine. No API key, no network call, no rate limits.

Metadata Richness (0–30 pts)
Semantic Specificity (0–30 pts)
Search Alignment (0–20 pts)
Completeness (0–20 pts)
Total: 0–100 points
Always free, always offline

Tier 2

AI-Powered Scoring

Uses GPT-4o-mini on our hosted server. Your free agentcard-disco key — not your own OpenAI key.

All four heuristic dimensions
+5 AI-evaluated semantic dimensions
Deeper natural language analysis
Anti-oscillation architecture — deterministic scores
5 deep scores/day on free tier
Flag: --deep

Rate limits

Tier	Deep scores / day	Price
Free	5	$0
Pro	Unlimited	Coming soon

Registration

Get Your Free API Key

Email address

API key generated successfully

Store this key safely — it won't be shown again.

Usage

# Set as environment variable
export AGENTCARD_API_KEY=acd_sk_...

# Then use deep scoring
agentcard-disco score card.json --deep