-
refactor(tools): drop parallel_fanout — rlm is the only RLM tool
Release / parity (push) Has been cancelledRelease / build (deepseek-linux-x64, deepseek, ubuntu-latest, x86_64-unknown-linux-gnu) (push) Has been cancelledRelease / build (deepseek-macos-arm64, deepseek, macos-latest, aarch64-apple-darwin) (push) Has been cancelledRelease / build (deepseek-macos-x64, deepseek, macos-latest, x86_64-apple-darwin) (push) Has been cancelledRelease / build (deepseek-tui-linux-x64, deepseek-tui, ubuntu-latest, x86_64-unknown-linux-gnu) (push) Has been cancelledRelease / build (deepseek-tui-macos-arm64, deepseek-tui, macos-latest, aarch64-apple-darwin) (push) Has been cancelledRelease / build (deepseek-tui-macos-x64, deepseek-tui, macos-latest, x86_64-apple-darwin) (push) Has been cancelledRelease / build (deepseek-tui-windows-x64.exe, deepseek-tui.exe, windows-latest, x86_64-pc-windows-msvc) (push) Has been cancelledRelease / build (deepseek-windows-x64.exe, deepseek.exe, windows-latest, x86_64-pc-windows-msvc) (push) Has been cancelledRelease / release (push) Has been cancelled发布于
2026-04-27 07:14:10 +00:00 Two near-duplicate top-level tools made the surface confusing. With
parallel_fanout (formerly rlm_query) removed, there's exactly one RLM
shape: load a long input ascontextin a Python REPL viarlm, and
let the sub-agent fan out from inside the REPL viallm_query_batched
where it hascontextin scope to chunk against.For non-RLM parallel work the dispatcher already runs multiple tool
calls per turn concurrently — no separate fan-out tool needed. The
GenericToolCell.prompts rendering hook stays (one-row-per-child for any
future fan-out tool), but no tool currently populates it.Also drops two stray test artifacts (rlm_catalog.md, rlm_test_doc.md)
the model wrote to repo root during a previous live test session.Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com
下载附件