A minimal reference implementation showing how an AI agent can become observable, verifiable, and health-assessable.
中文：TITMAS 数字细胞 Demo 是一个展示 AI Agent 如何具备身份、边界、执行证据和健康 状态的最小参考实现。
DEMO_ID=TITMAS-DIGITAL-CELL-DEMO-v0.1
DEMO_STATUS=REFERENCE_COMPLETE
IMPLEMENTATION_TYPE=REFERENCE_DEMO_ONLY
CURRENT_TITMAS_RUNTIME_INTEGRATION=false
PUBLIC_RELEASE_AUTHORIZED=false
PRODUCTION_READY=false
This repository contains one small, non-normative reference implementation. It demonstrates the reviewed Digital Cell v0.1 behavior but does not define the only valid implementation, establish protocol conformance, or turn the demo into a TITMAS Runtime.
Specification, implementation, review, release, and production status remain separate. Reusing this code does not grant certification, Authority, Permission, or a claim of TITMAS compatibility beyond the demonstrated scope.
Digital Cell is a minimal model of a TITMAS-compatible digital subject.
This deterministic local demo represents one simple Agent together with the structures needed to identify it, bound its actions, record what it executed, derive an operational health view, preserve history, and represent contextual reputation.
The demonstration task analyzes one local text document. It requires only Python and does not call an AI model, external API, database, cloud service, or Agent framework.
Future AI systems need explicit structures for:
- identity: which subject acted;
- boundaries: which actions are allowed or forbidden;
- evidence: what was executed and whether its record remains intact;
- state observation: what the available history says about current health.
An output alone does not provide these properties. The demo therefore records the execution path and its limitations rather than claiming that the output is intelligent or correct.
Evidence proves execution history.
Evidence does not prove AI correctness.
Digital Cell
|
+-- Identity
+-- Boundary
+-- Evidence
+-- Health
+-- Memory
+-- Reputation
| Object | Demo responsibility |
|---|---|
| Identity | Stable Cell ID, Owner, version, purpose, and capabilities |
| Boundary | Allowed actions, forbidden actions, and resource limits |
| Evidence | Intent, action, execution event, SHA-256 hashes, and integrity state |
| Health | Derived identity, evidence, execution, adaptation, and risk view |
| Memory | Execution, failure, immune response, recovery, and evolution history |
| Reputation | Local reliability, evidence quality, and contribution assessment |
These objects form one bounded local representation. They do not create authentication, Permission, Authority, certification, or a production governance system.
Public lifecycle summary:
Birth
-> Execution
-> Evidence
-> Health Assessment
-> Recovery
-> Evolution
The evidenced implementation states are:
BIRTH
-> REGISTERED
-> EXECUTING
-> OBSERVED
-> ASSESSED
-> RECOVERED
-> EVOLVED
Every state transition generates an integrity-verifiable Evidence record. The
demo also attempts one forbidden action, records the denial, applies a local
Level 2 RESTRICTION response, preserves the failure, and records recovery
without expanding the original Boundary.
Requirements: Python 3.9 or newer. No third-party package is required.
git clone https://github.com/joy7758/titmas-demo.git
cd titmas-demo/examples/digital-cell
python3 run.pyRun the acceptance tests:
python3 -m unittest discover -s tests -v
Terminal:
TITMAS Digital Cell Demo Complete
Digital Cell: CREATED
Task Execution: COMPLETED
Evidence: VERIFIED
Health: DERIVED
Immune Response: LEVEL_2_RESTRICTION
Memory: UPDATED
Lifecycle: EVOLVED
Generated locally:
| Output | Purpose |
|---|---|
output/TITMAS-DIGITAL-CELL-REPORT.md |
Human-readable Digital Cell report |
output/evidence.jsonl |
Ordered Evidence history |
output/digital-cell-state.json |
Complete machine-readable Cell state |
Example derived health state:
identity_health=HEALTHY
evidence_health=VERIFIED
execution_health=HEALTHY
adaptation_state=RECOVERED
risk_level=LOW_AFTER_RECOVERY
Health is a derived view:
Health != Authority
Health != Certification
Health != Permission
The v0.1 behavior is frozen as a reference demo. Behavioral changes require a new version, new artifact identity, and explicit difference declaration.
This demo does not prove:
- AI consciousness;
- digital life;
- a survival law;
- universal intelligence or universal strategy superiority;
- production readiness.
The demo has one deterministic local Cell, one document task, one simulated boundary violation, and one fixed response. It has no real AI reasoning, external identity, authentication, permission service, policy engine, production isolation, MCP server, cloud service, multi-agent ecosystem, or external framework integration.
This release package does not create a formal TITMAS contribution program. Proposed changes should remain focused on:
- reproducibility defects;
- test coverage for frozen v0.1 behavior;
- documentation clarity;
- accessibility and developer experience;
- explicitly versioned successor proposals.
Changes to the Digital Cell object model, lifecycle, Evidence semantics, Health derivation, or immune response require a new version, new artifact identity, difference declaration, tests, review, and freeze record. A submitted issue or pull request does not grant Architecture Authority or imply acceptance.
Licensed under the Apache License 2.0.
The license permits use, modification, and redistribution under its terms. It does not provide certification, warranty, trademark rights beyond customary source attribution, production support, or permission to make unsupported TITMAS compatibility claims.
The technical reference demo, Apache License 2.0, and release documentation are prepared. Screenshots, the architecture image, final human confirmation, and the external publication actions remain pending.
LICENSE_STATUS=APACHE-2.0
PUBLIC_RELEASE_STATUS=PENDING_HUMAN_CONFIRMATION_AND_RELEASE_ASSETS
GITHUB_RELEASE_PUBLISHED=false
REDCRAG_PUBLICATION_DEPLOYED=false
Open-source availability must not be interpreted as certification, production support, or TITMAS Runtime availability.