Nexus v2.15.1

Bug Fixes

• deps: patch DOMPurify to 3.4.17 and ip-address to 10.7.4 for current production advisories (manually reviewed supplement to the prepared v2.15.1 notes).
• closeout: merge bounded archive-readiness work (3998b20)
• closeout: qualify archive contract location and final gate (9b4ffec)
• closeout: reconcile historical archive authority and evidence (0cb4fc7)
• deps: raise MCP SDK and hono floors past new production advisories (893a1a3)
• v2.15: finish bounded archive-readiness repairs (c43432f)

Changes

• Image edits approved for an agent now save a recoverable checkpoint, including the redo history present at approval.
• Deleting an Image Studio session retains shared original files and explains that behavior in the interface.
• Saved video jobs restore in studio history, and Chat folder colors persist and can be cleared.
• Experimental document-outline tools and settings are retired under their recorded quality-gate fallback.
• MCP SDK and Hono dependency floors are updated to 1.32.1 and 4.13.13. DOMPurify 3.4.17 and ip-address 10.7.4 address the subsequently reported production advisories.

Recorded Closeout

The four triage batches and active closeout account for all 262 original identities: 245 obligations remain open in v2.16, 14 close with evidence and three prior resolutions are retained. A migration records work still owed; it does not mean the feature is implemented. Two later-discovered release automation defects are recorded separately as new open v2.16 entries, bringing that ledger to 247 open obligations. Their permanent fixes remain owed.

Historical v2.12 and v2.13 work remains tied to the actual shared v2.12.1 release. Archive readiness is checked separately after publication; this release does not archive documentation or delete branches or worktrees.

The historical pinned inventory verifier retains its older ID behavior. Its ambiguous-match diagnostic is preserved alongside the separately qualified verifier's exact copy proof. Hardware, model and installed-application guarantees carried into v2.16 remain subject to their recorded qualification requirements.

Nexus v2.15.0

Maintenance release: the project's record of known gaps is now one countable list. No app, extension or installer behavior changes.

Known gaps

• Every known-gaps file from v1.17 to v2.12, including the v2.12 carry-forward register, now reads cleanly in the release completion checker (204 unreadable rows before, 0 now), with no wording changed. Rows that only repeated an entry recorded elsewhere now point at it (v2.4#BG-2) instead of counting twice.
• The 47 items still open in v2.12 were checked against the current code: 3 are resolved with evidence, and 44 moved into the v2.15 known gaps word for word, each with why it is still owed and a next step. The v2.12 register is finalized.
• The 199 older items that are still open are listed once each in four follow-up batches, each sized for one plan (v2.15.1 to v2.15.4).
• What is still owed: 62 items in v2.15 plus the 199 batched items, 261 in all, with none counted twice.

Tooling

• node scripts/known-gaps/inventory.mjs reports every ledger's items, the open count, drift against a dated baseline, and (--triage, --batches, --verify-migrations) whether the triage record and batch list cover every item once and every migrated copy verifies. It reads ledgers through scripts/known-gaps/checker-adapter.py, which runs the completion checker only when its files match pinned hashes.

Not changed

• The app, the VS Code extension and the installers behave as in v2.14.2. The installers are rebuilt by the release workflow from the same product code.
• Release completion is still blocked by conflicts in the external completion checker, recorded with a handoff for its maintainer (v2.15 QG-11).

Nexus v2.14.2

Planned and implemented as v2.14.0 ("Faster spoken replies and a measured LFM2.5-24B worker trial"); released as v2.14.2 because v2.14.1 shipped first and version numbers must keep increasing. This release includes everything in v2.14.1.

Voice

• Desktop voice now runs on install: the installer provisions Kokoro and Whisper (pinned weight files) into the media runtime, and the sidecar starts the audio runtime with that interpreter.
• Spoken replies start from the first complete sentence while the model is still writing, instead of after the whole reply is written and synthesized. Median time to first audio on the medium test fixture fell from 38.6 s to about 3.9 s. The 2.0 s target was missed and is recorded as v2.15 DF-1.
• Speaking over a reply cancels synthesis (26 ms in the scripted run).
• What is spoken: reasoning and tool-call text are removed, secrets are redacted (including inside underscore emphasis), code and tables become a short spoken cue, and links are spoken by label.
• Voice turns ask the model to skip its hidden reasoning pass, so speech starts in seconds. Typed turns are unchanged.

Voice loop (existing opt-in, behavior changed):

• Activation: in desktop Chat, tick the "Voice loop" checkbox. It stays off by default.
• Validation: ask a three-sentence question by voice; speech should start before the reply text finishes.
• Rollback: untick "Voice loop". The voice models stay installed; remove them in Settings > Models if wanted.
• Authority: turning it on does not send audio or text off this machine (0 non-loopback connections measured during voice turns), and does not change typed turns or the model used.
• Docs: voice latency results

Models

• LFM2.5 24B-A2B (lfm2.5:24b-a2b) joins the catalog as an opt-in 16 GB mixture-of-experts worker. It fits fully on a 16 GB GPU at 32K context and matched the Nemotron offload build's golden-task completion within run-to-run spread, with a 46% better p95 turn time. The admission rests on the owner's reading of that spread (v2.15 DF-3).
• Pulls of commit-pinned hf.co models are checked against the pinned GGUF digest in both install paths; a mismatch removes the model and says so.
• The in-app installer can now install multi-file Hugging Face models (such as the voice models), each file checked against its own pin.

LFM2.5 24B-A2B (new opt-in model):

• Activation: Settings > Models on a GPU with at least 16 GB, install "LFM2.5 24B-A2B". It is never pre-ticked.
• Validation: the model shows as installed and can be selected for coding.
• Rollback: remove it in Settings > Models; no default changes.
• Authority: installing it does not make it a default for any tier, and its LFM Open License use restrictions still apply.
• Docs: model acceptance job map

Developer tooling

• nexus golden run accepts --options, --metrics and --tool-format, and runs live outside VS Code.
• Prompt-opened reasoning is handled in every path for catalog entries that declare it.
• CI runs the audio runtime tests on the Windows init job.

Fixes

• Secrets wrapped in underscore emphasis, or split by a tool call, are never spoken.
• A long reply no longer ends speech early, and a cold first voice turn no longer fails on the model load time.
• A rejected voice turn no longer leaves its voice job running; a half-finished weights install no longer looks installed.
• Skill tests no longer write folders into the source tree; the outline smoke report uses a private temp folder; handbook slugs strip nested tags.
• The Video2X deadline tests no longer race the real clock under load.

Known gaps carried forward: v2.15 known gaps.

Nexus v2.14.1

Development evaluation

• Add a bounded local rewrite evaluator with frozen synthetic English/Chinese drafts, strict evidence scoring, durable attempt budgets, loopback-only transport and exclusive outputs.
• Record the sole installed-model pilot as format HOLD and all sixteen held-out drafts as not-run. Factual preservation and readability improvement remain unmeasured under DF-5.
• Record Humanizer exact-runtime feasibility as HOLD, with no model download, runtime upgrade, catalog admission or production-client change.
• Verify 50 focused boundary regressions, the reproduced ledger race correction, independent review and current handbook receipts.

The owner explicitly selected v2.14.1. The separately isolated v2.14.0 plan is not included.

Nexus v2.12.1

v2.12.1 - release build fix for v2.12.0: durable image edits, approve…

Nexus v2.11.0

Bug Fixes

• deps: patch production advisories in range (eedcd7a)
• v2.11: bind the outline IPC methods and regenerate the module catalog (64ef6df)
• v2.11: keep outline fixture documents byte-exact on Windows checkouts (b2e40ed)
• v2.11: read the feasibility report without a check-then-read race (5e5400c)
• v2.11: store the CRLF fixture byte-exact (389be9b)

Features

• v2.11: document outline tools in both channels (phase 4) (42f436c)
• v2.11: final-phase hardening, CI reconciliation, and handbook (phase 6) (03effd2)
• v2.11: measure document-outline feasibility (phase 2) (2421ecc)
• v2.11: native tool calls and the outline smoke test (phase 5) (7c35a43)
• v2.11: parser seam and document outline core (phase 3) (53c60e1)

Nexus v2.10.0

Documentation

• record hosted Jev, Laya, and Kev as not admitted (f4d126f)

This release changes no opt-in capability, installer flag, or host surface.

Nexus v2.9.0

Documentation

• record CrisperWhisper 2.0 as not admitted (d5cf4bf)

This release changes no opt-in capability, installer flag, or host surface.

Nexus v2.8.0

Documentation

• record the community Qwen-Image-2.1 GGUF and Bespoke Nimble as not admitted (40c86b7)

This release changes no opt-in capability, installer flag, or host surface.

Nexus v2.7.0

Features

• video: refuse talking-head until the official avatar weights are installed (62b7b7f)