Public roadmap after v0.1.0

[leonardsellem]

This is the public roadmap thread for Hypermnesic after v0.1.0.

The current release focus is not adding a large feature surface. It is making the existing git-first memory loop easier to install, inspect, trust, and use from real clients.

Near term

• Finish repository presentation polish, including the custom GitHub Social preview.
• Submit the engine and companion to the relevant MCP, awesome-list, and Obsidian directories.
• Decide whether PyPI publication is worth the release-automation cost.
• Keep launch-week response times under 24 hours for new issues, discussions, and PRs.
• Draft and sequence launch posts only from evidence that is already green.

Contribution funnel

• Keep good first issue tasks small and verifiable.
• Merge or explicitly triage the first external PR quickly.
• Prefer fixes that make setup, diagnostics, docs, or client compatibility clearer.

Product direction

• Preserve the prime invariant: files are truth; the index is a rebuildable projection.
• Keep memory writes reviewable as commits.
• Keep the Obsidian companion read-only.
• Make degraded retrieval states visible rather than silent.
• Treat benchmarks as retrieval-quality evidence, not as a substitute for product readiness evidence.

Not goals right now

• Hosted multi-tenant service.
• Private data in public examples.
• More client-specific adapters when native MCP/OAuth primitives are enough.
• Benchmark spending unless a specific release decision needs fresh numbers.

If you want to help, reply with the workflow you are trying to run, the client you use, and the exact friction you hit.