OmniClip RAG v0.4.5 Release Notes
Release Focus
v0.4.5 is the extension oversized-routing hardening release.
This version finishes a practical stability problem that real users hit during PDF / Tika builds: some extension files are valid large text carriers and should still enter the search index, while others are really structured noise explosions that should never be allowed to poison the index or pin a build at a fake-progress plateau.
The release turns that distinction into a stable contract.
Highlights
- Oversized extension inputs are no longer handled with one blunt rule:
- text-heavy carriers such as
PDF / DOC / DOCX / EPUB / TXT / RTF / ODT / EML / MSG now regroup oversized parse output before indexing
- structured/noisy carriers such as
HTML / MHTML / XML / XLSX / PPTX / ZIP are skipped before they can drag builds into long vector-write stalls
- Extension builds now leave a durable per-file issue report:
- regrouped oversized text carriers
- structured oversized skips
- parse/runtime/I/O failures
- Activity feedback is more honest without becoming noisy:
- the desktop shell reports grouped summaries
- the file-level reasons go into a dedicated issue log instead of flooding the activity area
- The practical
Tika at 92% stall class is narrowed significantly:
- oversized structured pages now exit before
write_vector
- oversize text carriers are retained through regrouping instead of being silently thrown away
Release Assets
v0.4.5 ships the same three public release assets:
OmniClipRAG-v0.4.5-win64.zip
OmniClipRAG-MCP-v0.4.5-win64.zip
- manual MCP package for direct
stdio setup
omniclip-rag-mcp-win-x64-v0.4.5.mcpb
- MCPB package for the official MCP Registry line
MCP / Registry Note
This release does not add new MCP tools or widen the MCP surface.
What changes is release alignment:
- Registry metadata is refreshed to
v0.4.5
- the
.mcpb asset is rebuilt from the same tagged source
- the MCP ZIP, GUI ZIP, and Registry-facing metadata stay on the same release line
Notes
- This release intentionally does not optimize or replace the underlying parsers.
- It also does not redesign extension query semantics, resume semantics, or index-state contracts.
- The goal is narrower and safer: keep valid large text carriers searchable, keep structured noise out of the indexes, and make the reasons auditable after every build.
