Proof Chain

digital-twin is easiest to trust when each claim points to a file you can inspect. This page collects the repo's public proof files and what each one demonstrates.

What the repo proves today

Claim	Public file or folder	What to verify
A personal agent should retrieve durable context before answering.	Steven Workflow Demo	The demo starts from playground/wiki/_index.md, prior outputs, summaries, and writing rules before drafting.
Agent autonomy should be bounded before execution.	Capability Contract Demo	The contract declares intent, required evidence, safe outputs, write-back targets, privacy boundaries, and pass/fail checks before any file edits.
Capabilities should be explicit workflow contracts, not hidden prompt vibes.	content-creation.md	The capability defines inputs, context requirements, outputs, and write-back behavior.
The system must produce files, not only chat responses.	playground/Blog/Published/	A successful run leaves a durable draft in the workspace.
Learning should become reusable system state.	agent-learnings/	A successful run captures a future-facing rule or preference.
A run should leave a simple review trail.	Agent Work Receipt	The receipt records the request, files read or written, evidence, and items that still need human review.
The operating model transfers beyond Steven's own notes.	Elon Musk Demo	The same raw-source → wiki → style/decision framework structure works on a public persona demo.

Verification path

Use this as a reviewer checklist when evaluating the project:

1. Start with the thesis to understand the operating model.
2. Run the Steven Workflow Demo and check its success checklist.
3. Fill out the Agent Work Receipt for the run.
4. Inspect the linked capability file and output folders on GitHub.
5. Compare the Elon Musk Demo to confirm the pattern is not hardcoded to one personal workspace.
6. Read the Capability Contract Demo to see the boundary an agent should declare before execution.
7. Run the Steven Workflow Demo and check its success checklist.
8. Inspect the linked capability file and output folders on GitHub.
9. Compare the public persona demo to confirm the pattern is not hardcoded to one personal workspace.
10. Treat missing write-back as a failed run, even if the chat answer looks polished.

Repository size snapshot

This repo is intentionally small enough to audit manually. Recorded public-source snapshot, generated with pygount --format=summary --folders-to-skip='.git,node_modules,venv,.venv,__pycache__,.cache,dist,build,.next,.tox,vendor,third_party,docs/.vitepress/dist' . on 2026-05-04:

Language / type	Files	Code lines	Comment / docs lines
Markdown	40	0	1,502
YAML	1	45	0
JSON	1	23	0
Text only	4	0	56
Other / generated / binary	9	0	0
Total	55	68	1,558

Notes for reviewers: pygount classifies Markdown as documentation/comment lines rather than code, which is expected for this blueprint-style repository. The build output under docs/.vitepress/dist is excluded so the snapshot reflects source files rather than generated output.

Honest boundary

This repository is a blueprint and demo workspace, not a hosted SaaS runtime. The proof chain is intentionally file-first: it shows the structure, contracts, and observable files that a capable coding agent or AI IDE can execute against.