Verified data for AI agents, provable in one click.

Agents act on what they're told — so what they're told should be provable. Every Dynamic Feed response carries its source, licence and timestamp, and is Ed25519-signed over its exact bytes. Verify one right now, in your browser.

The call

Keyless. Fetch signed data, then verify it against the published key — in the box above, with scripts/verify_awareness.py, or by hand.

# keyless — a signed, grounded snapshot any agent can verify
curl -X POST https://dynamicfeed.ai/v1/awareness -H "Content-Type: application/json" -d '{"robot":{"class":"ground"},"location":{"lat":-27.47,"lon":153.02}}'

# the public key — verify the signature yourself, no account
curl https://dynamicfeed.ai/.well-known/keys

Sample response

{ "facts": [ { "id": "wx.wind_kmh", "value": 21, "unit": "km/h",
               "source": "Open-Meteo", "age_s": 240, "stale": false } ],
  "signature": { "alg": "Ed25519", "key_id": "df-ed25519-4cb32e72f333",
                 "canonicalization": "json-sorted-compact", "sig": "..." } }

Why live data

To an agent, a hallucinated number, a stale cache and a fresh fact all look identical: text. The fix isn't bigger models, it's a data layer whose answers carry their own evidence — source, licence and timestamp on every datapoint, an Ed25519 signature over the exact bytes, and optional Bitcoin anchoring for proof-of-when. The agent (or its auditor, or its customer) can verify any answer independently, even years later — against us, not just with us. A signature proves we reported this data at this time; it doesn't claim the data is objectively true — no signature can.

Use it for

• Agent pipelines that must cite checkable facts, not vibes
• Regulated copilots — fintech, legal, medical — that need a defensible data trail
• Multi-agent systems passing facts between models without a provenance gap
• Pair with /v1/notarize to witness exactly what your agent acted on

FAQ

Related live feeds