llms.txt

# verifiable-agent-demo

Repository name: verifiable-agent-demo

One-line role: reviewer-facing minimal demo for verifiable agent execution evidence.

Mainline category: runnable demo and reviewer walkthrough in the Agent Evidence / Operation Accountability Profile mainline.

Primary entry points:

- `README.md`
- `AGENTS.md`
- `demo/`
- `examples/enterprise_sandbox_demo/`
- `Makefile`

Important files:

- `scripts/run_demo.sh`
- `scripts/run_mvk_aep_bridge_demo.sh`
- `artifacts/demo_output/`
- `artifacts/enterprise_sandbox_demo/`
- `DEMO_RECEIPT_VALIDATION_REPORT.md`

Safe commands:

```bash
python3 -m demo.agent
bash scripts/run_demo.sh
python3 examples/enterprise_sandbox_demo/run.py
make mvk-aep-bridge-dry-run
```

Outputs:

- intent input
- execution trace
- evidence bundle
- replay verdict
- audit receipt
- reviewer-facing local artifacts

Non-goals:

- not the main specification repository
- not a general Artificial Intelligence (AI, 人工智能) agent framework
- not the canonical evidence packaging implementation
- not post-execution audit ownership
- not legal non-repudiation or compliance certification

Related repositories:

- `agent-evidence`: evidence packaging, validation, verification, receipts, and review packs
- `fdo-kernel-mvk`: execution-integrity proof path
- `digital-biosphere-architecture`: architecture and discovery hub
- `aro-audit`: post-execution audit and receipt review layer
- `token-governor`: pre-execution governance layer

Citation notes:

Use this repository as the shortest runnable demo path for reviewers. Use
`agent-evidence` for canonical evidence profile and bundle verification claims.