-
Notifications
You must be signed in to change notification settings - Fork 2
Create demo recording assets for both launch GIFs #35
Copy link
Copy link
Open
Labels
documentationImprovements or additions to documentationImprovements or additions to documentationhook-1-gradeGIF 1: Grade your agent (scan → audit)GIF 1: Grade your agent (scan → audit)hook-2-yaml-to-agentGIF 2: YAML to running agent (init → generate → run)GIF 2: YAML to running agent (init → generate → run)launch-infraNon-code launch infrastructure (community, marketing, docs)Non-code launch infrastructure (community, marketing, docs)
Description
Problem
Both launch GIFs need reproducible demo environments so the recording is clean and deterministic.
GIF 1: "Grade your agent" (scan → audit)
Create demo/grade-your-agent/ with:
-
app.py— A deliberately flawed Python agent to produce a satisfying D/F grade:- Hardcoded API key (not
$env:) - No guardrails
- No fallback model
- No memory hygiene
- No evaluation
- Hardcoded API key (not
-
README.md— Recording instructions:export AGENTSPEC_LLM_API_KEY=... export AGENTSPEC_LLM_BASE_URL=https://openrouter.ai/api/v1 agentspec scan --dir ./demo/grade-your-agent/ --out agent.yaml agentspec audit agent.yaml # Expected: Score D/F with 3-5 colorful violations
GIF 2: "YAML to running agent" (init → generate → run → curl)
Create demo/yaml-to-agent/ with:
README.md— Recording instructions:export AGENTSPEC_LLM_API_KEY=... export AGENTSPEC_LLM_BASE_URL=https://openrouter.ai/api/v1 agentspec init --yes agentspec validate agent.yaml agentspec audit agent.yaml agentspec generate agent.yaml --framework langgraph --output ./agent/ cd agent && pip install -r requirements.txt python -m uvicorn server:app --port 8000 & curl -s localhost:8000/v1/chat -H 'Content-Type: application/json' -d '{"message":"hello"}'
Optional: vhs tape files
If using vhs for deterministic GIF recording, include .tape files for both demos.
Acceptance criteria
-
demo/grade-your-agent/directory with flawed app.py + README -
demo/yaml-to-agent/directory with README - Both flows produce clean, impressive terminal output when followed
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
documentationImprovements or additions to documentationImprovements or additions to documentationhook-1-gradeGIF 1: Grade your agent (scan → audit)GIF 1: Grade your agent (scan → audit)hook-2-yaml-to-agentGIF 2: YAML to running agent (init → generate → run)GIF 2: YAML to running agent (init → generate → run)launch-infraNon-code launch infrastructure (community, marketing, docs)Non-code launch infrastructure (community, marketing, docs)