v9.0.2

What's Changed

• Settings: persist the Theory Source File path by @devin-ai-integration[bot] in #40
• Cloud chat: 15 min reply limit instead of a 20 s total cap; errors name the cause by @devin-ai-integration[bot] in #41

Full Changelog: v9.0.1...v9.0.2

v9.0.1

What's Changed

• Local AI: recalibrate the fit estimator and never disable the Install button by @devin-ai-integration[bot] in #36
• View menu: CPU, Memory and System monitor windows by @devin-ai-integration[bot] in #37
• Local AI: stream replies with no total time cap, live progress, Stop cancels, clearer errors by @devin-ai-integration[bot] in #38
• Release v9.0.1: Task mode DONE gate (+ #37 monitors, #38 long local runs) by @devin-ai-integration[bot] in #39

Full Changelog: v9.0.0...v9.0.1

v9.0.0

What's Changed

• Preserve extracted equations and term definitions in Gemini master axiom output by @devin-ai-integration[bot] in #1
• Let the AI create and write files inside the project root by @devin-ai-integration[bot] in #2
• Rank chapter files whose title names the requested topic first in retrieval by @devin-ai-integration[bot] in #3
• Attach the local retrieval evidence packet to every chat message by @devin-ai-integration[bot] in #4
• Compile the master axiom as a deterministic mathematical primer and inject it whole at startup by @devin-ai-integration[bot] in #5
• Add Math Fixer to Manuscript Tools and prefer Unicode-capable PDF engines by @devin-ai-integration[bot] in #6
• Quick wins: manual file-tree refresh, copyable code blocks, Add to notes, token-limit thread freeze by @devin-ai-integration[bot] in #7
• Autosave threads to SQLite and add a per-pane "Recall session" viewer by @devin-ai-integration[bot] in #8
• Unified document actions, built-in editor, Session Docs menu, and equation-safe app buffer by @devin-ai-integration[bot] in #9
• Add Local AI (Ollama) as a third provider with hardware check, fit estimates, and terminal install by @devin-ai-integration[bot] in #10
• macOS support for the Local AI hardware probe, estimator, and install launcher by @devin-ai-integration[bot] in #11
• README: add October 2026 progress brief and refresh current focus by @devin-ai-integration[bot] in #12
• AI file ops: READ/APPEND/EDIT blocks and allowlisted RUN with output fed back to the model by @devin-ai-integration[bot] in #13
• AI retrieval SEARCH blocks and durable project_memory.md injected into the primer by @devin-ai-integration[bot] in #14
• Local git checkpoints before AI file edits, with a Checkpoints viewer to inspect and revert by @devin-ai-integration[bot] in #15
• Task mode (plan/budgets/Pause/Approve/Stop, DONE signal, cancellable RUN) and CALC scratch computation with inline plots by @devin-ai-integration[bot] in #16
• Local AI: user-selectable model storage location via native folder picker (OLLAMA_MODELS) by @devin-ai-integration[bot] in #17
• Settings: full-screen System Settings window with three-column AI Models tab by @devin-ai-integration[bot] in #18
• Settings: AI Routing Summary first in AI Models tab; rename Customize menu to Settings by @devin-ai-integration[bot] in #19
• Ignore protocol-example and code-formatted op markers; keep PLAN/STEP and REMEMBER out of ordinary chat by @devin-ai-integration[bot] in #20
• Tolerate >>>> markers, strip fences around FILE/APPEND bodies, forbid fabricated RUN output by @devin-ai-integration[bot] in #21
• Run briefing compile and master-axiom generation off the main thread (fixes "not responding" on pane reset/startup) by @devin-ai-integration[bot] in #22
• Run file-op, RUN/CALC, checkpoint, autosave and memory commands off the main thread by @devin-ai-integration[bot] in #23
• Per-pane "Thread tools" menu: reset thread with confirm dialog, and a thread logs viewer by @devin-ai-integration[bot] in #24
• RUN guardrails: refuse inline-code flags, checkpoint before RUN/CALC, forbid rerouting denied commands by @devin-ai-integration[bot] in #25
• Smoke-test fixes A: hypothesis preamble, project_memory.md overwrite guard, task stop reasons, Settings hints by @devin-ai-integration[bot] in #26
• Smoke-test fixes B: remaining Tauri commands off the main thread; Math Fixer scan progress bar by @devin-ai-integration[bot] in #27
• Settings ▸ System & Apps: word-processor picker used by "Open in word processor" by @devin-ai-integration[bot] in #28
• Retrieval hygiene: skip app-generated reports; require topical overlap for the mechanism boost by @devin-ai-integration[bot] in #29
• Manuscript Tools: open full-screen; Master Axiom mode with custom source, output directory and format by @devin-ai-integration[bot] in #30
• Local AI: quantization selector for catalog installs with per-quant size and fit estimates by @devin-ai-integration[bot] in #31
• Release v9.0.0: version bump, tag-driven .deb release workflow, release notes by @devin-ai-integration[bot] in #32
• Release workflow: fix apt dependency conflict on ubuntu-22.04 by @devin-ai-integration[bot] in #33
• Pin @tauri-apps npm packages to the Rust crate minors so tauri build passes its version check by @devin-ai-integration[bot] in #34
• Release workflow: build on ubuntu-24.04 so the prebuilt onnxruntime links by @devin-ai-integration[bot] in #35

New Contributors

• @devin-ai-integration[bot] made their first contribution in #1

Full Changelog: v7.5.0...v9.0.0