Empty-state rethink: replace "Start a conversation." with a richer
landing showing recent conversations (top 5 as resume links) and a
voice-mode entry button. Gated to full variant only; widget/briefing
keep the simple message.
Context sidebar: switch from grid column (4-col layout) to absolute
overlay on the right gutter (3-col layout). The reading column is now
always symmetrically centered regardless of whether the sidebar is
visible — eliminates the leftward shift that occurred when context
notes appeared.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Content-based gating (_should_think) was introduced in 87fcaa6 to cut
TTFT on simple prompts, but it has no way to tell that short prompts
like "create a task titled X" are going to trigger a tool call — and
qwen3:14b's tool-call template is unreliable at think=False, producing
intermittent silent generations where output tokens burn but nothing
parses into content or tool_calls.
Reverting to always-on thinking restores the pre-87fcaa6 reliability
of tool emission at the cost of TTFT latency on short conversational
prompts. This also lets us delete the silent-round retry loop (which
can no longer fire) along with its bookkeeping.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Retry attempts were previously conflated with the initial call,
making prompt_tokens and headroom look cumulative and useless for
diagnosing the silent-round behavior. Move start-of-attempt captures
inside the retry loop and emit attempt_start / attempt_end lines so
each attempt's numbers stand alone.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Log num_ctx, message count, prompt/output tokens, headroom, and a
silent flag per round so we can correlate silent generations against
context pressure on the dev instance.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Streaming bubble (Generating response…) was flush against the input
wrapper top border. Give the scroll column 0.75rem of breathing room
so the assistant bubble sits visibly separate from the input.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Context sidebar sections (Auto-included / Suggested / In Context) are
now collapsible with chevrons, and collapsed state persists per
conversation via localStorage. A new Context (N) button in the chat
header toggles the whole sidebar — open by default on wide screens
(>=1200px), closed by default below. Visual differentiation: auto-
included notes get a muted background with an AUTO pill, in-context
notes get an active-chip style with a primary-colored left border,
and each section header shows its item count.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Qwen3's tool-call tokens sometimes fail to parse into either content
or tool_calls, burning output tokens and producing empty bubbles.
Detect the signature within a round (empty content, no tool calls,
eval_count > 0) and re-run the same round once with reasoning mode
enabled, which emits more reliable output. The post-loop fallback
remains as the final catch.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Qwen3:14b sometimes burns output tokens on tool-calling attempts whose
emission doesn't parse into any field we read — eval_count > 0 but no
thinking/content/tool_calls ever stream to the caller. Generation
completes "successfully," the user sees an empty assistant bubble, and
no error is logged. Seen in conv 220 today.
Two safety rails:
- stream_chat_with_tools now tracks whether it yielded anything; when
Ollama's done frame reports eval_count > 0 with zero yields, log a
warning including the last ~5 raw frames so the next occurrence leaves
breadcrumbs for diagnosis.
- run_generation checks the same post-condition after the tool loop
exits and, if content is empty with no tool calls but output_tokens
> 0, substitutes a visible fallback message and streams it as a chunk
so the user gets something readable instead of a blank bubble.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Phase B of chat view refinement. Bumps chat reading column to 1200px,
lifts the RAG scope chip out of the footer and into the ChatView header
(next to the title), and collapses the listen toggle / volume slider /
stop-playback buttons into a single speaker popover on the input bar.
Listen mode + TTS are now available in briefing and workspace chats too,
since the controls live on the shared ChatInputBar.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Header:
- Remove Research button + modal (skill fires reliably from normal chat)
- Move Summarize-as-Note into a header kebab (disabled until the
conversation has messages)
Left sidebar:
- Full-width "+ New Chat" primary button
- Search input filters conversations by title (case-insensitive
substring) with a "no matches" empty state
- Move Select into a sidebar kebab next to search
Shared .btn-kebab + .kebab-menu styles give future conv-level actions
a home. Document mousedown + Escape dismiss both kebabs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The inherited .chat-panel-fill class lands on ChatPanel's new .chat-full
root via attribute inheritance, and its display:flex was overriding the
grid layout — stacking messages, sidebar, and input bar vertically
inside the reading column instead of placing the sidebar in grid col 4.
Layout is ChatPanel's responsibility now; .chat-panel-fill only sizes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replaces the full-variant flex layout with a single CSS grid owning the
messages column, input bar, and context sidebar. Messages and input bar
share a centered ~820px reading track (--chat-reading-width token) so
the mic button can't get pushed off-screen on wide monitors, and the
context sidebar spans grid-row 1/-1 so it runs the full height from the
chat header down to the bottom edge instead of stopping above the input
bar.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The web silence detector previously ran RMS over getByteFrequencyData
bytes (which are already dB-scaled) and re-log'd the result, producing
numbers that didn't line up with real dBFS. Combined with a static
-40 dB threshold that sat right on top of room ambient, silence
detection rarely fired and the user had to click stop manually.
- Rewrite useSilenceDetector to use getFloatTimeDomainData for honest
linear RMS → dBFS.
- Silence threshold is now dynamic: track session peak dBFS and treat
"silent" as 15 dB below peak. Auto-calibrates per mic/room.
- Grace period (1500 ms) at start so the user can begin speaking
before checks arm; static -45 dB fallback until peak clears -25 dB
so dead-silent sessions don't spin forever.
- Silence duration bumped 1500 → 2000 ms for breathing room.
Visual: detached red radial-gradient disc sits behind the mic in a
wrapper; the button no longer scales, so the white mic icon stays
legible on top while the halo grows from 1x to ~2.6x with live
amplitude. Much more obvious "you are live" signal than the prior
subtle box-shadow pulse.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The mic button now scales and glows proportional to the silence
detector's RMS amplitude instead of sitting static. Gives obvious
feedback that audio is actually being picked up — silence still
breathes via a 0.1 floor, loud input caps at ~1.18x scale so it
doesn't punch through the input bar.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Discuss flow was hallucinating unrelated content when article
extraction returned empty or RAG pulled in orphan notes that looked
more relevant than the generic seed prompt.
- seed_article_discussion raises EmptyArticleError on empty body;
briefing and /news routes return 422 instead of staging an empty
synthetic tool result.
- build_context skips RAG auto-injection when user_message matches
ARTICLE_DISCUSS_SEED so the article IS the context on turn one;
follow-up turns keep RAG on.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Both the /news discuss button and the briefing discuss button now call a
shared seed_article_discussion() helper that stages the synthetic
read_article tool exchange and the conversational seed prompt — behavior
stays byte-identical across entry points. /news also auto-starts
generation so the chat screen lands on an in-flight stream.
First assistant reply in a seeded article conversation is persisted as a
Note (tags: article-summary + article topics) and backlinked via
rss_items.discussion_note_id, so the knowledge base stops being amnesiac
about articles the user has engaged with.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Forgejo Actions doesn't consistently honor `on.push.branches` as a
filter — the merge commit landing on main after a release PR was still
triggering the full CI workflow on a SHA that had already passed on
dev, burning ~1 minute of runner time for no reason.
The `build` job already had a ref guard, so no duplicate images were
being pushed, but typecheck/lint/test ran unconditionally. Add the
same guard (`refs/heads/dev` or `refs/tags/v*`) to those three so main
pushes trigger the workflow but every job skips immediately with zero
runner time.
Also tightened the build job's tag filter from `refs/tags/` to
`refs/tags/v` for consistency with the new guards and to avoid ever
building from a non-version tag.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The Discuss button on news cards was producing one-shot replies because
the model got the whole trafilatura blob dropped into history with a
canned "summarize and discuss this article" prompt — no length guard, no
prep, no invitation to converse. Large articles got silently truncated by
Ollama; small articles got a tepid reply.
This reworks discuss_article around a three-layer cache:
context_prepared → content_full → fresh trafilatura fetch
First click on a small article fetches once, writes through to both
caches, and passes the body straight into the synthetic read_article
tool-result. First click on a large article additionally runs a parallel
map step (services/article_context.py) that chunks the body on paragraph
boundaries, summarizes each ~8k chunk to ~300 words of dense factual
prose via the background model, and concatenates the summaries under
section headers — all pinned to num_ctx=16384 so the map step doesn't
itself fall victim to silent truncation. Repeat clicks on either path
skip straight to the chat turn.
The canned summary prompt is replaced with a conversational seed that
invites the user into an actual discussion rather than a one-shot
synopsis, matching the goal of "have a conversation about an article,
not just read it."
discuss_topic is intentionally left untouched — it's the multi-article
aggregation path and needs a separate rework. Follow-up task will decide
whether to retire it or rework it on the cached-context approach.
Closes task #106.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Non-streaming generate_completion was the only LLM entry point that
didn't default num_ctx — stream_chat and stream_chat_with_tools both
fall back to Config.OLLAMA_NUM_CTX (16384). When a caller omitted the
argument, Ollama silently used the model's default window (~4k on
qwen3) and truncated the prompt.
That footgun was masked by fallback paths in the research pipeline:
_generate_outline's prompt carries ~12 sources × 2000 chars (~6k
tokens) of source material plus a system prompt, so the prompt got
chopped, the model never saw the sources, JSON parsing failed twice,
and run_research_pipeline dropped into the single-note "monolith"
fallback (research.py:251). The user reported chat 215 producing such
a monolith note for a multi-source research topic.
Two-layer fix:
- Default num_ctx to Config.OLLAMA_NUM_CTX inside generate_completion,
matching the streaming entry points. Any current or future caller
that forgets the argument stops silently losing input.
- Pin num_ctx=16384 explicitly in _generate_outline and
_generate_executive_summary with comments pointing at the failure
mode, so a refactor of the generate_completion default can't
silently regress the research pipeline.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
_generate_title was receiving the full messages list from build_context,
which prepends RAG snippets, RSS excerpts, URL content, and briefing
article dumps INTO the user-role message string. The role=="user" filter
inside _generate_title then handed that composite blob (capped at 300
chars) to gemma3:4b as "the user's message", so the background model
was titling conversations based on article excerpts instead of what the
user actually typed — producing wildly wrong titles like "Briefing
Profile Preferences & Schedule" for a plain calendar query. See #109.
Pass the raw history + user_content + assistant reply instead.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two related bugs where the server defaulted naive datetimes to UTC instead
of the configured user timezone, causing all-day events to land on the
previous day and briefings to "disappear" at UTC midnight.
- New services/tz.py helpers: get_user_tz, user_today, user_briefing_date
(the briefing day flips at 4am local to align with the compilation slot,
so the 00:00-04:00 local window still shows yesterday's briefing until
the new one is generated).
- calendar create/list/update tools now parse naive datetimes in the
user's TZ before converting to UTC for storage, and tool descriptions
tell the model to pass plain local dates.
- briefing_conversations.get_or_create_today_conversation and the
reset-today route use user_briefing_date so the in-progress briefing
doesn't get replaced at 19:00 NY / UTC midnight.
- _run_profile_closeout targets user-local "yesterday" for consistency.
Regression tests added for the TZ helpers and the calendar tool.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds 38 parametrized tests for the _should_think classifier covering the
explicit-override path, empty/whitespace content, short/medium/long length
boundaries, case-insensitive keyword matching, and a chatty-message negative
set. These pin the content-based semantics so future tweaks to the keyword
list or length thresholds surface regressions immediately instead of going
unnoticed behind subtle latency changes.
Also drops the `think=True` overrides from the briefing /discuss-article
and /discuss-topic entry points. With `"discuss"` added to _THINK_KEYWORDS,
those canned prompts trip the classifier naturally, so the overrides were
redundant — keeping a uniform "classifier is authoritative" rule makes the
code easier to reason about.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The frontend hardcoded think=true on every chat send (ChatPanel full +
widget variants, KnowledgeView minichat), which defeated the _should_think
gate on the backend and made qwen3:14b spend 5-20s on chain-of-thought
reasoning for every turn — even "hi". This was the root cause of the
warm-path TTFT variance tracked in followup_ttft_variance.md: the logged
ttft_ms was really prefill + full thinking phase, bouncing with the depth
of the model's reasoning, not with cache or eviction.
All three frontend callers now pass think=false and let _should_think be
authoritative. The classifier is now a real content-based gate: explicit
think_requested=True still forces on as an override (briefing discuss
actions, future UI toggles, MCP callers), otherwise messages <80 chars
without reasoning keywords skip thinking, messages >=400 chars or
containing keywords like why/explain/analyze/debug/review/etc. get it.
Generation timing now separately records think_requested, the final
think decision, first_token_ms (first any chunk), and thinking_ms
(duration of the thinking phase). ttft_ms keeps its existing semantic
(first content token) so existing log analysis still works. The timing
log line surfaces all four fields so the old "just a big ttft number"
ambiguity is gone.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two small hardening fixes from the mistral-nemo testing round:
1. stream_chat / stream_chat_with_tools now read the Ollama response
body and log it before raising on non-2xx. Previously all we saw
was 'HTTP 400 Bad Request' — the gemma3-no-tools failure would
have been diagnosed in one step if we'd been logging the body,
which says e.g. 'model does not support tools'.
2. backfill_project_summaries() now also targets summaries stamped
before 2026-04-12 (the gemma3:4b cutover). The remaining projects
still carrying the broken qwen2.5:3b output (token repetition,
hallucinated topics) will regenerate on next startup on the
better model.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Ollama's /api/tags returns whatever casing was used at pull time
(e.g. 'gemma3:12B' if the user ran 'ollama pull gemma3:12B'), but
/api/chat rejects mixed-case tags with a 400. The two code paths
are inconsistent, which surfaces the capitalized tag in the model
dropdown and then silently kills every chat request against it.
Lowercase on read (get_installed_models), on settings write
(update_settings_route), and on ensure_model() input so a legacy
mixed-case user setting can't trigger a spurious re-pull at
startup. The dropdown and stored settings are now always in the
form Ollama will actually accept.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
qwen2.5:3b produced broken auto-summaries (misspellings, token repetition,
hallucinated topics) — its synthesis ceiling is too low for free-form
summarization. Gemma 3 4B is stronger on summarization at similar size
and still fits comfortably alongside the main chat model in VRAM, so it
preserves the KV-cache-separation strategy that keeps chat TTFT fast.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Project auto-summaries were using the 3B background model, but the
task — synthesizing a coherent paragraph over ~10 notes — is well past
what 3B can do reliably. Evidence on dev: "doging conversation
hygiene", "MCPview). MCP).", trailing stray quotes, and hallucinated
topics ("AI regulation").
Route through the user's default chat model instead. Project summary
regeneration is rare (only when a project changes) so the KV cache
eviction cost on the main model is negligible.
Title generation, tag suggestions, and RSS classification continue to
use the background model — those tasks are within what 3B handles.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three related fixes uncovered while benchmarking qwen3:14b against 8b:
- pick_num_ctx was only counting message content, missing the ~15K
tokens of tool schemas. num_ctx=8192 was being selected while actual
prompt_tokens hit 14K+, causing silent prompt truncation on every
tool-using request. Now includes json.dumps(tools) in the estimate.
KV cache priming in app.py and routes/settings.py also fetches tools
so the primed num_ctx matches what real chat requests will use.
- _should_think's heuristic classifier was overriding explicit
think=true requests from the frontend toggle and MCP, gating on
message length and regex patterns. Now a pass-through — the caller
is the source of truth. quick_capture hardcodes think=False since
it's a fast classification path that was relying on the old gating.
- delete_note description only mentioned "note or task", so the model
refused to call it for entries created by save_person / save_place /
create_list. Description now explicitly lists all five note_types it
handles.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Update tools.py references to tools/ package, remove stale intent
router section, update research pipeline to multi-note output,
fix create_task references now merged into create_note.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Merge create_task into create_note (set status='todo' for tasks, omit
for notes), merge delete_task into delete_note, consolidate entity
tools (create/update_person → save_person, create/update_place →
save_place), rename get_note → read_note with clearer descriptions,
move calculate out of rag.py into utility.py, and extract shared
duplicate detection into check_duplicate() helper.
Updates all downstream references in generation_task.py, quick_capture.py,
ToolCallCard.vue, and WorkspaceView.vue.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Some exceptions (e.g. connection errors) produce empty str(e),
resulting in "Research failed: " with no explanation. Fall back to
the exception class name when the message is blank.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Research pipeline now produces an index note with:
- Executive summary (2-3 paragraphs synthesized from sections)
- Clickable links to each section note (/notes/{id})
- Section notes have parent_id pointing to the index
Also improves outline resilience: lowered minimum sections from 3
to 2, retries once on failure before falling back to monolithic note.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The .btn-new-conv class has flex:1 for the sidebar row layout, but
when reused inside .no-conversation (a column flex), it stretched
vertically to fill the entire viewport — appearing as a giant
purple rectangle. Override with flex:none in the no-conversation
context.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The Forgejo registry occasionally returns 400 on large cache layer
blob uploads, failing the entire build even though the image itself
pushed successfully. Adding ignore-error=true to cache-to so cache
failures don't block deployments.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
KnowledgeView and WorkspaceView were passing explicit titles
("Knowledge chat", "Project — Workspace") to createConversation(),
which prevented the backend's auto-title generation from firing
(condition: `not conv_title`). Pass no title so the background
title generator names conversations after their first message.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The type-badge template had no v-else-if for task, so tasks fell
through to the v-else which renders 'List'.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Split 2566-line tools.py into a tools/ package with @tool decorator
registration. Each tool's schema, metadata, and implementation live
together. Briefing eligibility is now a briefing=True flag instead of
a separate frozenset allowlist. Conditional inclusion (CalDAV, SearXNG)
uses requires= metadata. Public API (get_tools_for_user, execute_tool)
unchanged.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
uv creates bare venvs without setuptools (unlike pip). http-ece
doesn't declare setuptools as a build dependency but needs it when
built with --no-build-isolation.
The py3.12-node22 runner doesn't ship pipx, so the previous commit's
pipx install ruff failed with command not found. Switched to the same
venv pattern the test job uses.
Task card was the outlier in the Knowledge view: half-width fade-to-
transparent gradient bar while note/list use full-width two-stop
gradients, and the type badge was purple while the border+bar were gold.
- k-card--task::before: full width with gold→amber gradient (#d4a017
→ #fbbf24) matching the two-stop pattern
- badge--task: switched from purple to amber so the badge matches
the card chrome
bvandeusen