Seven open-source experiments extending the Foundation's continuity infrastructure research into testable, buildable tools. Each repository has a clear scope, defined success criteria, and a direct connection to published Foundation work.
The EM Foundation's research produces conceptual frameworks, governance proposals, and mathematical formulations. Turning those into working infrastructure requires engineers, data scientists, and developers willing to build the experimental systems that prove — or disprove — the underlying claims.
Each repository below corresponds to a published proposal on this site. The proposal documents the intellectual foundation. The repository is where the actual work happens. Contributions of any size are welcomed: a working prototype, a benchmark harness, a visual interface, a test dataset, or a challenge to the underlying assumptions.
All EM Foundation open-source work is published under MIT license. Attribution appreciated. Commercial use permitted. Critique essential.
A continuity-aware context management layer for LLM applications. Classifies conversation context into memory card types, assigns Continuity Compression Scores, and retains only continuity-critical context before each prompt. Benchmarked against full-context and semantic-cache baselines.
Acceptance criteria:
Connects to: Continuity Compression proposal · Continuity Receipts · CIIC
A lightweight wrapper for any AI model that generates structured Failure Receipts when confidence thresholds are not met for the requested reliance level. Drop-in addition to existing deployments — no model modification required. Demo at emfoundation.net/cr-lite.html.
Connects to: Failure Receipts Standalone proposal · CR Standard · CR-Lite Demo
A Python simulator comparing three GPU workload scheduling policies — random, least-loaded, and thermal-continuity — across configurable node counts and thermal zone configurations. Produces charts for peak heat, cooling proxy cost, throttling events, and job delay.
Connects to: Thermal-Continuity Routing proposal · SEMA · Grid-Aware Scheduling
A deep space telemetry simulator comparing ordinary compression against continuity-aware SVP-ordered transmission under packet loss scenarios. Uses NASA PDS public datasets. Explicitly grounded in real deep space communication physics — no faster-than-light claims.
Connects to: Deep Space Telemetry proposal · No-Signaling Boundary · CR provenance architecture
A semantic synchronization protocol transmitting only continuity-significant changes — new evidence, changed confidence, resolved contradictions, altered decisions — rather than full-state or structural diffs. JSON schema compatible with OCMS. Includes drift detection.
Connects to: Delta Protocol proposal · OCMS schema · CIIC
A grid-load simulator comparing FIFO scheduling against grid-aware scheduling for flexible AI workloads, using EIA and ElectricityMaps public data. Outputs peak load reduction, renewable alignment, job delay, and carbon proxy metrics. Includes a policy memo for regulatory engagement.
Connects to: Grid-Aware Scheduling proposal · Systems Stability Framework · Thermal Routing
An educational web package explaining why quantum entanglement cannot transmit information faster than light, redirecting toward optical deep-space communication, DTN, and quantum-secure key distribution as the credible research directions. Interactive simulation showing why entanglement correlations do not carry controllable messages.
Connects to: No-Signaling Boundary essay · Deep Space Telemetry
Read the proposal paper linked from the repository card above. Each paper documents the intellectual foundation, the formula, the experiment design, and the acceptance criteria. Understanding the why is as important as implementing the what.
Open a GitHub issue in the relevant repository describing what you plan to build and how it connects to the acceptance criteria. This prevents duplicate work and lets the Foundation provide early feedback before significant effort is invested.
Build and document. Every contribution must include a README explaining what was built, how to run it, and what the results show. If your results challenge the underlying hypothesis, document that too — negative results are valuable.
Submit a pull request. The Foundation reviews all PRs for scientific integrity and connection to the published proposal. Approval does not require agreement with Foundation positions — it requires honest methodology and transparent reporting.
Publish and be cited. Significant contributions will be credited in the Foundation's publications and annual report. If your contribution produces results worth publishing, the Foundation will work with you on a joint publication under open access terms.
Questions: research@emfoundation.net