Local AI that survives memory loss
Deterministic Fragment Graphs β’ Continuity Engine β’ Self-Healing Runtime
Failure is inevitable. Collapse is optional.
Interactive continuity-runtime simulation exploring self-healing AI memory, deterministic fragment graphs, and post-failure runtime reconstruction.
| Surface | Purpose | Link |
|---|---|---|
| XPADI-SGDS | Canonical survivability ecosystem root | https://github.com/raajmandale/XPADI-SGDS |
| XPADI Proof Engine | Runtime continuity reactor | https://raajmandale.github.io/XPADI_Proof_Engine_V1/ |
| Digital Lifeline | Survivability architecture narrative | https://github.com/raajmandale/digital-lifeline |
| XLifelineAI | AI-native continuity & self-healing runtime | https://github.com/raajmandale/XLifelineAI |
| XPADI-ProofCheck | Recovery intelligence surface | https://xpadi.com/proofcheck/ |
| Research Paper | SGDS architecture & theory | https://zenodo.org/records/19500143 |
XLifelineAI now also operates as the continuity-runtime branch of the broader Mandale-OS ecosystem.
XLifelineAI explores continuity-native AI execution where memory, fragments, graph intelligence, and runtime survivability become part of the execution lifecycle itself.
Within the Mandale-OS ecosystem, XLifelineAI represents:
- continuity-runtime research
- self-healing execution systems
- post-failure runtime recovery
- graph-aware memory continuity
- AI survivability infrastructure
RUN β FAIL β DETECT β REBUILD β CONTINUE
A failure-native AI runtime using:
π Deterministic Fragment Graphs (DFG)
RUN β FAIL β DETECT β REBUILD β CONTINUE
git clone https://github.com/raajmandale/XLifelineAI.git
cd XLifelineAI
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
pip install -e .
python examples/resurrection_demo.py
Fragments created: 8
Fragments destroyed: 3
Integrity score: 0.625
Continuity mode: patched
- partial context
- connected memory
- survivable structure
- detect loss
- analyze graph
- rebuild missing
Graph β Scan β Repair β Continue
Simulator
docs/simulator/index.html
Report
docs/demo/resurrection_report.html
xlifeline/
docs/
examples/
Traditional
β reset
XLifelineAI
β recover
- AI continuity systems
- long-running agents
- failure-resilient runtimes
- memory corruption simulation
v0 β DFG runtime core
v1 β semantic repair
v2 β distributed fragments
v3 β agent-native runtime
Research prototype
DFG continuity model validated
Raaj Mandale
Systems Architect β’ Runtime Intelligence β’ Mandale-OS β’ QBAIX β’ XPADI-SGDS
π https://raajmandale.in
π¬ https://orcid.org/0009-0005-9810-1655
π https://openalex.org/A5127026877
MIT License
If this idea resonates:
β Star the repo
π΄ Fork it
π§ͺ Break it
π§ Build on top of it
AI shouldnβt restart.
It should recover and continue.