Hamanaka Software Factory
Mike Hamanaka
An open invitation into the workshop.
A living ecosystem of tools, experiments and a private integration hub.
Most of my systems work comes together in ai-dashboard. These repositories are the entry points: some useful on their own, some experimental, some private. This atlas connects the dots and shows where other people can help the system grow.
32 repositories · 17 public · 15 private · inventory checked 28 October 2026
Public and private describe source access. They do not decide whether a project belongs in the ecosystem. Private links require permission; their names and roles can still be discussed here. Experiments get a place on the map, with their maturity visible.
The integration hub
Agents, inference, work boards, spend, machine health, evidence and review come together in one local operator surface. The surrounding repositories expose focused tools and seams that can evolve independently, then be considered for integration.
Explore all repository roles →The map groups responsibilities and contribution opportunities. The owner identifies AI Dashboard as the integrated hub; individual cross-repository adapters and deployments still need their own evidence. A place on the map is not a claim that every experiment is already connected.
The private workshop where most integration happens.
The local command center where I bring agents, inference, tickets, spend, evidence and machine health together.
Integration hub
Where you could help: Discuss operator experience, adapter contracts, observability and useful experiments through the public atlas.
Packaged supervisor, job dispatch, local execution, readiness and audited controls.
Private engine
Where you could help: Discuss portable supervisor contracts and reproducible integration examples.
Local AI stack workspace and durable agent handoffs.
Private workspace
Where you could help: Discuss source-bound handoffs and recovery contracts.
Public tools for turning an idea into reviewable work.
Capture browser intent into reviewed proposals and portable context.
Browser extension
Where you could help: Try browser compatibility, accessible capture controls and export/import edge cases.
Explain selected evidence in the context of a current goal.
Desktop companion
Where you could help: Improve evidence explanations, keyboard workflows and intent-bundle handoffs.
Persistent agent terminals, bounded task context and reviewable results.
Desktop alpha
Where you could help: Test terminal recovery, session continuity, accessibility and bounded context adapters.
A Mac supervisor for selected Taskrivet output and reviewed work records.
Prerelease
Where you could help: Exercise output redaction, checkpoint capture and macOS build/recovery flows.
Expand a ticket into bounded context and reviewable execution preparation.
Staging experiment
Where you could help: Build synthetic context fixtures and test dependency ordering and cancellation.
Contracts, independent checks, retained records and lessons.
A shared language for bounded context, evidence and visible failures.
Ticket standard
Where you could help: Contribute small interoperable examples, schema edge cases and adapter fixtures.
Independent intent checks and proof records for reviewable coding work.
Verification tooling
Where you could help: Add falsifiable checks, adversarial examples and understandable failure reports.
Turn local failure evidence into reviewed lessons and reusable heuristics.
Memory tooling
Where you could help: Improve provenance, retrieval evaluation and reviewable memory updates.
Deterministic append-only state verification for edge nodes.
Private verifier
Where you could help: Discuss portable verification formats using synthetic examples.
Local verification and evidence workspace.
Private workspace
Where you could help: Discuss evidence exchange and review ergonomics in the public atlas.
Autonomy evidence and records ledger.
Private ledger
Where you could help: Propose interoperable receipts without requiring ledger access.
Discover resources, understand constraints and explore distributed work.
Hardware discovery, topology snapshots and change reports.
Read-only discovery
Where you could help: Add hardware fixtures, discovery adapters and topology-diff examples.
Thermal, power and memory observations with advisory dispatch planning.
Prototype
Where you could help: Improve sensor fixtures, stale-data handling and resource-limit explanations.
Asynchronous inference queues with retries and failover; the real ledger adapter remains unfinished.
Alpha
Where you could help: Explore idempotency, cancellation, failover and the unfinished ledger seam.
Go event envelopes and an in-process concurrent pub/sub mesh.
Learning experiment
Where you could help: Explore transport, offline durability, discovery and partition tests as proposed next slices.
Fictional fixtures and experimental product surfaces.
Reproducible fictional business stories and offline CRM adapters.
Synthetic fixtures
Where you could help: Add generic fixture scenarios, adapter validation and deterministic examples.
A fictional company dataset for local business-workflow rehearsals.
Private simulation
Where you could help: Use generic public Mockmill fixtures to explore similar workflows.
Synthetic hardware nodes and protocol fixtures for platform testing.
Private simulation
Where you could help: Discuss generic protocol fixtures through the public atlas.
Standalone extension concepts and a proposed browser-local SDK.
Private prototype
Where you could help: Discuss generic extension lifecycles, permissions and fictional fixtures.
A fictional client/talent gesture-planning demo derived from Extension Studio.
Private experiment
Where you could help: Discuss accessible planning and consent models with synthetic examples.
A fictional demo for rehearsing changed project circumstances.
Private experiment
Where you could help: Discuss context continuity and human-reviewed next-step proposals.
A fictional hardware-to-handoff rehearsal with local permissions and reproducible bundles.
Private local demo
Where you could help: Discuss reproducible demo bundles and tamper/failure fixtures.
The front doors, documentation and communication layers.
A private personal clone used for client-platform work.
Private client workspace
Where you could help: Discuss portable client interfaces without copying client-specific source or data.
A multi-tenant web shell for the evidence workspace.
Private web shell
Where you could help: Discuss accessible operator/client surfaces using generic examples.
Publish reviewed ledger receipts through Joomla Web Services.
Private publishing adapter
Where you could help: Discuss portable receipt-to-article formats and preview behavior.
The short introduction to the entire software-factory ecosystem.
Public profile
Where you could help: Improve orientation, project explanations and contributor entry points.
This public map of the workshop, its experiments and ways to contribute.
Public atlas
Where you could help: Improve the map, accessible navigation and project/contribution descriptions.
Forks that connect the workshop to other communities.
An agent skill for interactive technical diagrams.
Upstream fork
Where you could help: Check upstream guidance and contribute diagram usability or schema examples to the right project.
Docker Sandboxes learning material and host-side testing connectors.
Workshop fork
Where you could help: Improve workshop reproduction, platform notes and synthetic sandbox examples.
Build a useful piece of the bigger system
You do not need the private factory to help. Start with a public tool, a reproducible problem or a small synthetic example. Improvements can stand on their own; I can evaluate how they fit the larger system later.
Run a public project using its own setup guide. Report your platform, the steps, what you expected and what happened. Accessibility, recovery and confusing first-run experiences are valuable starting points.
Suggest a small contract first: inputs, outputs, failure behavior and an offline example. Good places include ticket formats, evidence interchange, hardware discovery, inference queues and synthetic CRM data.
Prototype ideas in a bounded branch. Label simulations and proposals, keep the example reproducible, and show what would be needed before integration. Learning projects are welcome here too.
Contribution ideas are invitations to discuss scope, not pre-approved implementation requests. Check each repository’s license and contribution guidance. For private components, use the public atlas issue tracker to propose a generic interface or ask about collaboration.
Propose an ecosystem contribution ↗