Skip to content

README.md

Latest commit

 

History

History
45 lines (34 loc) · 1.79 KB

README.md

File metadata and controls

45 lines (34 loc) · 1.79 KB
Raw

scripts/

Helpers that are not part of the shipped packages.

make_demo_gif.py

Turns docs/assets/demo-transcript.txt into the animation on the README and the docs home page (docs/assets/amp-demo.gif), plus a still frame (docs/assets/amp-demo.png) for anywhere an animation will not play.

pip install pillow        # the only requirement
python scripts/make_demo_gif.py

It refuses to draw a line that would run past the right edge rather than producing a picture with a truncated command in it, and it refuses a transcript line that looks like a marker (*Result) but is missing the space the format asks for (* Result) - that mistake renders as ordinary output, so it is invisible in the result and easy to make twice.

Re-recording the transcript

The transcript is real output, not written by hand. To refresh it after a change to the server or the demo:

# 1. start the server (the demo talks to https://localhost:8765)
cd server && pip install -e . && uvicorn amp_server.main:app --port 8765

# 2. run the demo it shows
cd examples/multi-agent-demo && python run_demo.py

# 3. ask the same question as two different agents, and copy the replies
curl -s http://localhost:8765/amp/v1/memories/search \
  -H 'X-AMP-Agent-ID: agent_billing_v1' -H 'Content-Type: application/json' \
  -d '{"query": "contact preference", "owner_id": "user_123", "limit": 3}'
curl -s http://localhost:8765/amp/v1/memories/search \
  -H 'X-AMP-Agent-ID: agent_marketing' -H 'Content-Type: application/json' \
  -d '{"query": "contact preference", "owner_id": "user_123", "limit": 3}'

Paste the results into docs/assets/demo-transcript.txt and run the generator. Keep the file honest: if the output changes, the picture changes with it. Never edit the transcript to say something the server did not.