idkmesh

# IDKMesh

> Open-source verification-first coordination fabric for humans, AI agents, and heterogeneous compute working on uncertain goals.

Canonical site: https://mskazemi.com/idkmesh/
Canonical repository: https://github.com/MSKazemi/idkmesh
License: Apache-2.0
Status: research preview; not production software.

## Start

- [Home](https://mskazemi.com/idkmesh/) — what IDKMesh is, why verification is the bottleneck, and the shortest path into the project.
- [Start in 15 minutes](https://mskazemi.com/idkmesh/start.html) — install the CLI, run the bundled diagnostic, reproduce checks, and choose a bounded task.
- [How IDKMesh works](https://mskazemi.com/idkmesh/concepts.html) — bounded Work Units, authority boundaries, verification debt, correlated verifiers, and the evidence ladder.
- [Pipelines](https://mskazemi.com/idkmesh/pipelines.html) — work/evidence, orchestration, compute-admission, and PR-gate flows.
- [Documentation library](https://mskazemi.com/idkmesh/library.html) — published documentation by category.

## Topic guides for common search questions

- [100-question answer map](https://mskazemi.com/idkmesh/questions.html) — one generated index of the actual conversational questions answered across the ten evidence-linked topic guides.
- [AI agent verification](https://mskazemi.com/idkmesh/topics/ai-agent-verification.html)
- [Multi-agent orchestration](https://mskazemi.com/idkmesh/topics/multi-agent-orchestration.html)
- [AI code review](https://mskazemi.com/idkmesh/topics/ai-code-review.html)
- [LLM-as-a-judge reliability](https://mskazemi.com/idkmesh/topics/llm-judge-reliability.html)
- [Verifier panels and independent review](https://mskazemi.com/idkmesh/topics/verifier-panels.html)
- [Human oversight and AI agent governance](https://mskazemi.com/idkmesh/topics/agent-governance.html)
- [AI provenance and reproducible evidence](https://mskazemi.com/idkmesh/topics/provenance-evidence.html)
- [MCP and A2A interoperability](https://mskazemi.com/idkmesh/topics/mcp-a2a-interoperability.html)
- [Verification debt and agent scaling](https://mskazemi.com/idkmesh/topics/verification-scaling.html)
- [Verified swarm engineering](https://mskazemi.com/idkmesh/topics/verified-swarm-engineering.html)
- [All topic guides](https://mskazemi.com/idkmesh/topics/)

## Core technical sources

- [What is IDKMesh?](https://mskazemi.com/idkmesh/WHAT_IS_IDKMESH.html)
- [Architecture source](https://github.com/MSKazemi/idkmesh/blob/main/ARCHITECTURE.md)
- [Canonical schemas](https://github.com/MSKazemi/idkmesh/tree/main/schemas)
- [Specifications](https://mskazemi.com/idkmesh/specifications/)
- [Decisions](https://mskazemi.com/idkmesh/decisions/)
- [Project adoption guide](https://mskazemi.com/idkmesh/PROJECT_ADOPTION_GUIDE.html)

## Evidence

- [Experiment atlas](https://mskazemi.com/idkmesh/research.html) — published experiment summaries with their actual result status.
- [Research documentation](https://mskazemi.com/idkmesh/research/)
- [E017 verifier-panel measurement](https://github.com/MSKazemi/idkmesh/blob/main/experiments/E017-item-difficulty-and-quorum.md)
- [E017 reproduction guide](https://mskazemi.com/idkmesh/research/E017_VERIFIER_PANEL_REPRODUCIBILITY.html) — retained artifact SHA, commands, expected verifier-panel metrics, limitations, and falsification criteria.
- [Audits](https://mskazemi.com/idkmesh/audits/)
- [Findings](https://mskazemi.com/idkmesh/findings/)

## Trust, contribution, and safety

- [Contributing](https://github.com/MSKazemi/idkmesh/blob/main/CONTRIBUTING.md)
- [Governance](https://github.com/MSKazemi/idkmesh/blob/main/GOVERNANCE.md)
- [Security policy](https://github.com/MSKazemi/idkmesh/blob/main/SECURITY.md)
- [Support](https://github.com/MSKazemi/idkmesh/blob/main/SUPPORT.md)

## Reading rules

- The GitHub repository is the canonical source of truth for contracts, code, evidence, and decisions; this website is a discovery and presentation surface.
- Implemented mechanism, synthetic validation, observed evidence, and accepted conclusions are different evidence states and should not be collapsed.
- Worker success is not acceptance, verifier recommendation is not merge authority, and CI success is not independent human review.
- Claims about scale, safety, or effectiveness should be read at the evidence level stated by the linked source.