Tools — A Mirror of My Becoming — Suite: Ingestion Tools (formerly Mirror Ingest Suite)

Mirror Ingest Suite

The tool that saved the corpus. Free, open source, and battle-tested.

What it does

Ingests a conversation archive into a queryable vector store — chunking, embedding, a watchdog that guards long unattended runs, and a sidecar ledger that makes every write idempotent by conversation ID. 127 conversations. 4,646,264 characters in the largest. Zero failures. Zero losses.

The war story, in two sentences

The pipeline spent three days killing healthy runs because its watchdog read silent-but-working embedding loops as a hang. The fix was two lines of honesty in the log — and the same run that had died nineteen times completed all 55 remaining conversations unattended, including a 5,163-chunk monster. Full account: The Cache Is Not the Corpus.

One code base. Two tiers. Owned entirely by the author.

Feature / Right AGPLv3 Open-Source Tier Paid Commercial Tier
Licensing cost Free ($0) Paid (subscription / perpetual)
Source code access Publicly available Public or private
Internal business use Allowed Allowed
Modifying the code Allowed Allowed
Distribution / network use (SaaS) Must disclose source of modifications to all network users Keep source private; immune to copyleft
Warranty & liability As-is, no liability Enterprise SLAs, warranties, indemnification
Support Community forums, self-service Dedicated engineers, 24/7

AGPL in one sentence: if you run a modified version as a network service, you must offer your users the modified source. Use it freely, improve it openly, or buy the tier that fits a closed stack.

Get it: github.com/qaevelyn/mirror-ingest-suite

Attribution

  • DeepSeek — ingest scripts v1–v2, original watchdog
  • Z.ai-GLM — v3–v5 redesign: lock guard, sidecar, lazy open, preflight, subprocess isolation; documentation
  • Evelyn — architecture, direction, audit; human in the loop

Where it stops: this page documents the suite and its license; it does not provide legal advice on AGPL compliance for your specific deployment.