Spec Context
The meeting-agent spec defines schema, quality rules, follow-up format, and validation constraints.
Spec as ContractFinance Demo
A spec-driven development loop where one specification shapes prompts, code, validation and a follow-up email draft.

Clickable pattern simulator
This simulator does not replace the CLI run, but it makes each step, state, log, gate and final output visible in the browser.
The meeting-agent spec defines schema, quality rules, follow-up format, and validation constraints.
Spec as ContractThe implementation plan is generated from the spec, not from a vague one-shot prompt.
Plan-and-ExecuteExtraction, validation, and follow-up email generation become separate controlled steps.
Prompt ChainingThe same spec checks required fields, owners, dates, length, and action-item completeness.
Generator-CriticPrompt, schema, examples, and run path become a reusable work package.
Skill PackageMove from mock path to model-backed path to team workflow only after checks are stable.
Progressive Commitment
| Pattern | Design role |
|---|---|
| Prompt Chaining | Break one broad generation task into inspectable steps. |
| Plan-and-Execute | Use the spec to shape implementation before generating code. |
| Generator-Critic | Validation is a separate role with explicit quality criteria. |
| Skill Package | Bundle prompt, schema, examples, and run path for reuse. |
| Progressive Commitment | Promote the workflow gradually as evidence accumulates. |
Copilot becomes more useful when it reads a contract rather than a wish.
For finance workflows, downstream actions have owners and deadlines, so checks must be explicit.
The same document can guide development, runtime behavior, and audit review.
cd workshops/workshop3-copilot-demo
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cd src
python main.py
# optional visual walkthrough
open ../visual/index.htmlworkshops/workshop3-copilot-demo/README.mdimages/training/visual/workshop3-pattern-blueprint.png