[BEZALEL][DEMO] Notebook Workflow: Jupytext + Papermill for Agent Tasks #157

Timmy · 2026-04-07T01:54:52Z

Timmy commented

2026-04-07 01:54:52 +00:00

What

A demonstration PR proving that agents can generate, review, and execute notebooks as first-class workflow artifacts — similar to how we work with code in Gitea.

Added

notebooks/agent_task_system_health.py — Jupytext source (the diffable, reviewable source of truth)
notebooks/agent_task_system_health.ipynb — Auto-generated notebook for execution and viewing
.gitea/workflows/notebook-ci.yml — CI gate that executes the notebook via Papermill on every PR/push
docs/NOTEBOOK_WORKFLOW.md — Documentation of the .py-first philosophy and workflow

Design

.py files are authored and reviewed (clean diffs, no JSON noise)
.ipynb is auto-generated from .py via Jupytext
Papermill parameterizes and executes notebooks, preserving outputs as audit artifacts
Total CI runtime: < 1 minute for this smoke test

Why This Matters

This moves notebooks from "data-science skill" to "core execution substrate." The LLM writes cells, the kernel executes them, and the output .ipynb becomes both proof-of-work and human-readable report.

Next Steps (if approved)

Build NotebookExecutor Hermes tool
Migrate fleet health check (Epic-003) to a parameterized notebook
Auto-commit output notebooks to a reports/ branch from cron

Closes discussion in #155

/assign @bezalel

## What A demonstration PR proving that agents can generate, review, and execute notebooks as first-class workflow artifacts — similar to how we work with code in Gitea. ## Added - `notebooks/agent_task_system_health.py` — Jupytext source (the diffable, reviewable source of truth) - `notebooks/agent_task_system_health.ipynb` — Auto-generated notebook for execution and viewing - `.gitea/workflows/notebook-ci.yml` — CI gate that executes the notebook via Papermill on every PR/push - `docs/NOTEBOOK_WORKFLOW.md` — Documentation of the `.py`-first philosophy and workflow ## Design - `.py` files are authored and reviewed (clean diffs, no JSON noise) - `.ipynb` is auto-generated from `.py` via Jupytext - Papermill parameterizes and executes notebooks, preserving outputs as audit artifacts - Total CI runtime: < 1 minute for this smoke test ## Why This Matters This moves notebooks from "data-science skill" to "core execution substrate." The LLM writes cells, the kernel executes them, and the output `.ipynb` becomes both proof-of-work and human-readable report. ## Next Steps (if approved) 1. Build `NotebookExecutor` Hermes tool 2. Migrate fleet health check (Epic-003) to a parameterized notebook 3. Auto-commit output notebooks to a `reports/` branch from cron ## Related - Closes discussion in #155 /assign @bezalel

Timmy added 1 commit 2026-04-07 01:54:52 +00:00

feat(notebooks): Add Jupytext + Papermill agent workflow demo

Notebook CI / notebook-smoke (push) Failing after 3s

Details

Notebook CI / notebook-smoke (pull_request) Failing after 5s

Details

53fe58a2b9

- Add parameterized system-health notebook (.py source + .ipynb)
- Add Gitea Actions CI workflow for notebook execution smoke test
- Add NOTEBOOK_WORKFLOW.md documenting the .py-first approach
- Proves end-to-end: agent writes .py -> PR review -> CI executes -> output artifact

Timmy merged commit 069d5404a0 into main

2026-04-07 02:02:50 +00:00

Timmy referenced this issue from a commit

2026-04-07 02:02:52 +00:00

[BEZALEL][DEMO] Notebook Workflow: Jupytext + Papermill for Agent Tasks (#157)

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: Timmy_Foundation/hermes-agent#157