Client-Side Registries, SRC-D & SRC-D2, Enforce Data Integrity for AI Context


I'm MΛVΞRICK (Evanscira Labs). Built SRC-D & SRC-D2: two local, secure data registries to stop losing complex Gemini/LLM context and code snippets.

Hey everyone. I'm MΛVΞRICK, and I built these two dashboards specifically because managing large, detailed outputs from LLMs was getting messy. SRC-D and SRC-D2 are not cloud apps; they are Platform-Secure Persistence (PSP) tools running 100% locally in your browser, using Local Storage and IndexedDB. This makes them perfect for handling sensitive or proprietary context from Gemini or other models.

The core feature for AI workflows is the Omni-Code Scratchpad. It's a multi-tab, auto-saving staging area where you can securely paste massive Gemini responses. You can then use the built-in previewer to draft code integrations or format notes before committing them to a tracked Project or Incident.

Crucially, the architecture is dual-registry and built for safe data integrity:

Dual-Registry Architecture:

These are two distinct registries (not themes) that can run concurrently, managing separate workflows, but their .srcd save files are fully interchangeable.

Enforced Integrity for Teams:

The MERGE protocol is ID-based. If a collaborator creates a project independently (giving it a new ID), the system ensures safety by adding it as a new, separate entry on import, preventing accidental overwrites. For successful master-file collaboration, teams must start from a single shared base .srcd file. This ensures ID consistency, allowing the MERGE function to cleanly update the master project across all users.
If you need a robust, local, and secure way to manage the output and code surrounding your LLM development or data-heavy applications, check these out.

SRC-D: https://evansciralabs.github.io/SRC-D_WEYU/

SRC-D2: https://evansciralabs.github.io/SRC-D2_WEYU/

Leave a Reply