Resume a project
Recover the working ramp around a project without starting from a generic summary.
Local continuity for AI-assisted projects
threadVault turns exported AI conversations into a local working vault. Search by project, review the history by authorship, and open a useful result back to the exact message and nearby chronology.
Built by neuroVectoring. RC3 is Windows-first, and the original provider export remains higher-authority evidence than every derived layer.
exact messagehash-linkedWhere it helps
Long-running work is shaped by sequence: where an idea began, which paths were rejected, what the human decided, and what the assistant only proposed.
Recover the working ramp around a project without starting from a generic summary.
Review human decisions and assistant proposals as different kinds of evidence.
Move from a continuity result to the exact message, locator, hash, and neighboring chronology.
The working model
The original export stays above the derived intake. Retrieval remains connected to exact messages.
Keep the original provider file intact in the raw-export area.
Use the supplied prompt to normalize the export into threadvault_intake.jsonl.
Import stable message records with source mappings and integrity checks.
Use a distinctive project anchor and supporting terms to reduce unrelated recall.
Open the exact source message, locator, hash, and surrounding chronology.
What it brings into view
threadVault does not ask one summary to carry the whole history. It surfaces different candidates for review and keeps their relationship to source visible.
Why threadVault exists
threadVault began with a recurring failure while resuming Project Ikigai. The source conversations had not disappeared, but the project no longer returned as a coherent whole. Facts could be searched. The sequence that made them useful had to be reconstructed by hand.
That experience produced a simple hierarchy: preserve the provider export; treat normalized intake as derived; treat retrieval results as candidates for review; and keep an exact path back to source.
Project Ikigai is referenced only as the real origin context. No private project details or conversation content are used on this page.
The finished film will combine the origin story with the complete tutorial. It will use a real RC3 session and synthetic data—not a sales montage.
What was tested
The release moved through three candidates. RC3 received an independent GO within the stated scope after the earlier security, packaging, and Windows restore-path findings were repaired.
Read the complete verification record →EF957F6BB3F3B35297EA52EF8218855EE35FEDB5A6FBDE3322F78221F4A6F5FERelease status
The manual-intake edition is available at no cost. It is the complete tested RC3 package—not a hosted demo—and it continues to run locally after delivery.
Practical questions
No. The provider export remains the higher-authority evidence. threadVault creates a local, normalized working vault while preserving a path back to source.
The tested RC3 application runs locally. Independent testing observed loopback-only listeners and no telemetry, update checks, activation, or outbound connections within the tested scope.
A supplied prompt helps an AI normalize a provider export into threadvault_intake.jsonl. That intake is derived material—a navigation layer, not a replacement for the original export.
threadVault is designed for returning to long-running AI-assisted work: finding where a project began, distinguishing human decisions from assistant proposals, reviewing relevant chronology, and opening a continuity result back to an exact source message.
RC3 is Windows-first and uses standard-library Python. It requires no package installation or administrator privileges within the tested scope.
Yes. The current manual-intake release is available at no cost through approved platform-release links. Each link sends a private, time-limited download by email; the ZIP is not hosted at a permanent public address.