* Cookbook: Engine filter + intelligent hardware-computed serve profiles
Two related Cookbook serving improvements for accurate, hardware-aware model
serving (especially on consumer GPUs that can only run GGUF/llama.cpp).
Engine filter
- New "Engine" dropdown (All / llama.cpp / vLLM / SGLang) beside the quant
picker. Pure client-side view filter over the fetched list via the same
_detectBackend() the serve commands use, so what you filter to is exactly what
would launch. Re-renders from cache (no refetch). Empty-state message + the
instant-cache-paint path account for it too.
Intelligent serve profiles (Quality / Balanced / Speed)
- services/hwfit/profiles.py: compute_serve_profiles() turns detected VRAM +
model size into concrete llama.cpp flags (n_gpu_layers, n_cpu_moe, cache-type,
context). Encodes the by-hand tuning: a too-big MoE offloads experts to CPU
instead of failing; a model that fits stays fully on GPU; quant tracks profile
intent; vision models keep image-encoder headroom. Reuses models.py VRAM math
so filtering and serving agree on what fits. Pure/deterministic (no t/s claims
— partial-offload speed isn't reliably predictable; fit is what's computed).
- /api/hwfit/profiles endpoint returns the profiles + the model's trained
context limit, with loose name matching (strips org/ prefix, -GGUF suffix,
quant tag) so a local GGUF folder name resolves to its catalog entry.
- _buildServeCmd (llama.cpp) now emits --n-cpu-moe / --flash-attn /
--cache-type-k/v when set, with llama-cpp-python fallback equivalents. It
previously only set -ngl/-c, which is why it OOM'd or ran slow.
- Serve panel: profile chips that fill the fields on click, plus CPU-MoE / KV
Cache / Flash Attn fields. Context is clamped to the model's trained limit
(and an absolute 1M sanity ceiling) on type/blur/profile-load and at launch —
fixes a crash where a stale 256k/16M preset + quantized KV cache caused an
amdgpu ErrorDeviceLost.
Tests: tests/test_serve_profiles.py (7) — offload vs full-GPU fit, never exceed
VRAM, context cap, launchable flags, vision headroom, no-GPU empty.
Checks: py_compile + node --check pass; pytest test_serve_profiles + test_hwfit_amd
green; verified live on an RDNA4 box (gfx1200) — Balanced lands ~ncm18 q4 128k,
matching hand-tuning.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Cookbook: make column-header sorting discoverable (incl. Newest)
Sorting in Cookbook is via clickable column headers (pewds' design), but the
headers had no visual cue that they're interactive — so sorting in general, and
the Newest sort on the Model header specifically, was undiscoverable.
- Style sortable headers as interactive: pointer cursor, hover underline, and
the active sort column bolded/highlighted. There was no CSS for
.hwfit-sortable / .hwfit-sort-active at all; this helps every existing sort,
not just Newest.
- The Model column header sorts by release_date (newest first), reusing the
existing header-click sort wiring and the "newest" SORT_KEY.
No new sort control — uses the existing column-header paradigm.
Checks: node --check passes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Cookbook serve profiles: keep the on-disk file's quant fixed (don't propose Q6/Q2)
In the Serve tab the model is a specific GGUF file already on disk, so its quant
can't change — but the profiles were suggesting "Quality · Q6_K" / "Speed · Q2_K"
as if you could re-quantize it. That's meaningless when serving a fixed file.
- compute_serve_profiles gains serve_weights_gb / serve_quant. When set (SERVE
mode), the quant is locked to the file's and profiles differ only in the real
serving knobs — n_cpu_moe, KV-cache type, context. _weights_gb / _cpu_moe_for_budget
use the file's actual size instead of a quant-derived estimate. DOWNLOAD mode
(no override) still varies the quant to show download options.
- /api/hwfit/profiles accepts serve_weights_gb & serve_quant.
- The Serve panel parses the file's size (from m.size "20.6 GB") and quant (from
the repo/file name) and passes them, so profiles match what's actually served.
Result for a 20.6 GB Q4_K_M file: all three profiles stay Q4_K_M and differ by
KV/ctx/offload (Quality q8 KV 128k ncm21, Balanced q4 128k ncm17, Speed q4 32k
ncm15) — no nonsensical quant changes.
Tests: test_serve_mode_keeps_fixed_quant. Full serve-profile suite green (9).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Cookbook serve: Vision toggle (auto-find mmproj) + live VRAM/RAM-spillover monitor
Two serve-panel additions:
1. **Vision toggle.** A "Vision" checkbox that serves the model with its
multimodal projector so it can read images. The mmproj path is resolved at
runtime (find mmproj-*.gguf next to the model), so dropping an mmproj file in
the model folder makes the toggle just work; `--mmproj … --image-max-tokens
1024` (native) / `--clip_model_path` (llama-cpp-python) only when on + found.
2. **Live GPU-memory monitor.** A readout that polls /api/cookbook/gpus every 4s
while the panel is open and shows VRAM used/total/%, free, and — crucially on
a discrete card — **RAM spillover** (AMD gtt_used_mb), with a plain-language
health hint: green/healthy, amber/tight, red/"spilled to RAM — slow (raise
CPU MoE or lower context)". Surfaces gtt_used_mb from the gpus endpoint
(previously read for total only and discarded for 'used').
Lets you see at a glance whether a config fits VRAM (fast) or is paging to system
RAM over PCIe (slow) instead of guessing.
Checks: node --check + py_compile pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Improve Docker GPU setup diagnostics
Add a Docker GPU preflight script for NVIDIA users. The script is
read-only by default, checks host NVIDIA drivers, Docker availability,
and container GPU passthrough, and prints actionable next steps.
Add explicit opt-in modes to print install commands, install NVIDIA
Container Toolkit on Ubuntu/Debian, and enable the NVIDIA Compose overlay
in .env after passthrough is verified.
Document common NVIDIA Docker failure modes, ignore generated .env
backups, and clarify that Cookbook can only detect GPUs exposed to the
Odysseus container.
* Clarify Docker GPU diagnostic limits
Two small polish items in the Cookbook Serve panel.
Saved-config badge
The little count badge next to the Save button ("3 ▾" etc.) had a
generic "Saved launch configs" tooltip, so the number reads like a
notification dot. Make it spell out what it is and what clicking does:
"3 saved launch configs for <model> — click ▾ to load or delete"
(and "No saved launch configs for <model> yet — click Save to add
one" when empty). Tooltip stays in sync via _updateSavedToggleLabel
so save/delete updates both the count and the hint.
GPU chip on mixed-GPU boxes (#711)
The chip label was `${gpuCount}x ${gpu_name}`, where gpu_name is
just gpus[0].name — so a 4090 + 3060 reads as "2x RTX 4090". The
backend already emits gpu_groups (identical cards grouped, used by
the serve flow to pin CUDA_VISIBLE_DEVICES) and a per-card gpus[]
array, so use them:
- Label renders each homogeneous pool: "1× RTX 4090 + 1× RTX 3060".
Homogeneous setups keep the existing "2× RTX 4090" form.
- Tooltip lists each GPU with its index + VRAM, useful for picking
the right device when launching.
Refs #711.
Adds a diagnosis pattern for the 'Failed to infer device type' error
vLLM raises when no CUDA or ROCm GPU is found (e.g. systems with only
integrated or Intel Xe graphics). The existing pattern only caught
'No CUDA GPUs are available' which fires later in startup; this new
entry catches the earlier device-probe failure and the NVML/amdsmi
library-not-found messages that precede it.
Surfaces in the Cookbook serve card as: "vLLM could not find a supported
GPU — switch to llama.cpp or Ollama" instead of a raw Python traceback.
Co-authored-by: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Windows has 'App Execution Aliases' that can make shutil.which('python3')
and shutil.which('python') resolve to a Microsoft Store stub instead of
real Python -- even when Python is properly installed. The stub outputs:
'Python was not found; run without arguments to install from the
Microsoft Store, or disable this shortcut from Settings > Apps >
Advanced app settings > App execution aliases.'
and exits 9009, producing empty stdout. The JSON parse of the local
model cache scan then fails with 'Expecting value: line 1 column 1
(char 0)', and the Cookbook model list shows nothing.
Fix: prefer sys.executable as the interpreter for the local scan.
Odysseus already runs inside its own venv, so sys.executable always
points to the real venv Python and bypasses PATH / Store alias lookup
entirely. which_tool() is kept as a fallback.
Cross-platform: sys.executable works identically on Linux and macOS
(returns the real interpreter path), so this change is safe everywhere.
start-macos.sh now skips Homebrew formulae that are already installed, so re-runs no longer re-hit Homebrew. tmux and llama.cpp are treated as optional: a failed install warns and continues instead of aborting the launch under set -e. Python stays required (it builds the venv).
When the operator sets AUTH_ENABLED=false, three owner-scoped endpoints still
returned 401 (api/models, api/research/*, api/email/*), so the front-end
redirected the browser to /login and the app was unusable despite auth being
turned off. require_user() in src/auth_helpers.py already documents and honors
this contract (issue #622) via 'if _auth_disabled(): return ""', but these
endpoints did their own get_current_user/is_configured check without it.
Make _require_user (research), the /api/models anti-leak guard, and
email_helpers._require_auth consult _auth_disabled() and let anonymous through
(owner='') only when the operator explicitly disabled auth. The 401 protection
is fully intact when AUTH_ENABLED=true. Verified end-to-end: with
AUTH_ENABLED=false the SPA now loads instead of bouncing to /login.
The spinoff endpoint authenticated the caller (_require_user) but never
verified the research session belonged to them before reading the
persisted report and seeding it into a new chat session owned by the
caller. Any authenticated user who knew or guessed another user's
research session ID could exfiltrate that user's full report into their
own session — a cross-user data disclosure (IDOR).
Every other endpoint in this router gates on _owns_in_memory /
_assert_owns_research right after validating the session ID; spinoff was
the lone exception. Add the same _owns_in_memory check (covers both the
in-memory task and the on-disk JSON) so a non-owner gets a 404 before any
data is read or a session is created.
Add regression tests pinning the anonymous (401) and wrong-owner (404)
cases.
_resolve_ddg_redirect (the DuckDuckGo /l/?uddg= redirect resolver used on every
HTML-fallback result href) gated on `"duckduckgo.com" in parsed.hostname`. That
substring test also matches look-alike hosts like `duckduckgo.com.evil.com` and
`notduckduckgo.com`, so a result link on such a host would be silently rewritten
to its embedded `uddg` target. Same substring-vs-hostname pitfall fixed for
provider detection in 54ecfa3.
Match the host properly: exactly `duckduckgo.com` or a `.duckduckgo.com`
subdomain. Genuine redirects (`//duckduckgo.com/l/...`, and relative `/l/...`
hrefs resolved against `html.duckduckgo.com`) keep working.
The resolver was a closure inside duckduckgo_search; lifted it (plus the new
_is_duckduckgo_host helper) to module scope so it can be unit-tested directly.
Adds tests/test_ddg_redirect_resolution.py (red on the look-alike case before
this change, green after).
The /api/vault/unlock handler ran `bw` as
`_run_bw(["unlock", req.master_password, "--raw"])`. _run_bw launches it with
`asyncio.create_subprocess_exec(bw_path, *args)`, so the master password became
a process argument — readable by any local user through `ps` and
`/proc/<pid>/cmdline` for the lifetime of the unlock subprocess. The Bitwarden
master password decrypts the entire vault, so this is a serious credential
exposure on any multi-user / shared host (CWE-214).
The sibling /login handler already avoids this by feeding the password on
stdin; unlock was the outlier. Hand the password to `bw` through the
environment instead (`--passwordenv BW_PASSWORD`), mirroring how BW_SESSION is
already passed — `/proc/<pid>/environ` is readable only by the process owner,
not other local users. Add regression tests pinning that the secret reaches
the subprocess env and never appears in argv.
The dependency-install fallback chain unconditionally ran
'pip install --user', which fails inside a virtualenv (and as root in
LXC/containers) with 'Can not perform a --user install. User site-packages
are not visible in this virtualenv.' — even though the function's docstring
already noted --user is invalid in venvs.
Guard the --user fallback with a venv check so it only runs outside a venv
(where --user is actually valid for PEP-668 system Pythons). Derive the venv
probe interpreter from the install command (python for 'pip', python3 for
'pip3'/'python3 -m pip') so the check runs in pip's own environment. System
PEP-668 installs keep the --user fallback; venv/LXC-root installs no longer
hit the --user error. Updated the unit test for the new chain.
Closes#388
Add a hashchange handler for #document-<id> so refresh / URL-bar nav opens the document, and replace the silent console.error in loadDocument with a user-facing toast.
Closes#560
* fix(stream): read 'reasoning' SSE field for vLLM 0.20.2 / NIM
vLLM 0.20.2 / NVIDIA NIM emit reasoning-parser output in the `reasoning` delta field; older builds use `reasoning_content`. stream_llm() read only the latter, so reasoning from models like Nemotron-3-Nano (--reasoning-parser) was silently dropped and never rendered. Accept either field.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(agent): keep reasoning_content only on the latest assistant turn
The agent loop echoed each round's reasoning back as `reasoning_content` on every assistant turn, assuming vendors ignore it. Nemotron's chat template re-injects ALL prior reasoning_content as <think> blocks, and the loop is trimmed only once (before it starts) — so reasoning accumulated unbounded across rounds, bloating context and feeding the model its own prior reasoning, which reinforced repetition/looping. Strip reasoning_content from earlier assistant turns so only the most recent round carries it (still satisfies DeepSeek's thinking-mode follow-up requirement).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(agent-ui): wrap each round's reasoning in its own <think> block
The streamed think-tag wrapper gated on whole-message substring checks (accumulated.includes('<think>')), which only ever wrapped ONE reasoning block per message. A multi-round agent response has a reasoning phase per round, so once round 1 closed its <think>...</think>, rounds 2+ reasoning was emitted unwrapped and leaked into the visible answer. Replace the substring checks with a stateful open/close flag that toggles per think/answer cycle, so each round's reasoning gets its own collapsible block. Single-turn chat is unchanged (one open, one close).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* test(stream): reasoning/reasoning_content delta surfaces as thinking chunk
Covers @pewdiepie-archdaemon's requested regression: a streamed {reasoning: ...} delta emits a thinking chunk while {content: ...} streams as normal content; plus the older reasoning_content field for backward compat. Mirrors the #591 scenario.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(cost): treat dotless container hostnames as local (free)
getModelCost() substring-matches model names against a cloud price table, so a self-hosted 'nemotron'/'llama' model was billed at cloud rates. isLocalEndpoint() only recognized IPs / localhost / .local, not bare Docker service names (nim-nano, llamaswap), so the local-is-free guard missed them. A single-label hostname (no dot) can never be a public API -> treat as local.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* test(cost): isLocalEndpoint classifies service names local, cloud FQDNs billable
Covers @pewdiepie-archdaemon's requested cases: llamaswap/nim-nano + localhost/private-IPs/.local => local (free); api.openai.com/openrouter.ai/etc => not local. Drives the real function via node --input-type=module (same approach as test_reply_recipients_js.py), skips when node is absent.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Update Styles.css
Small update to the styles that bothered me, i noticed in the window/modal for calendar when editing a day the time icons had a mask that overlapped the icon. I simply added 'background-image: none' prop to it/
* Importing files bug
I found a bug that wouldn't let me upload files in the library window during the documents tab, when a user selected a file, the code grabbed a reference to fileInput.files and immediately cleared the input value (fileInput.value = '') to allow for re-uploading the same file later. However, because fileInput.files is a live FileList tied directly to the DOM element, clearing the input inherently emptied our saved variable as well, resulting in lost file data.
Note this error might be browser specific as it worked fine on Zen/Firefox but failed on Edge and chrome
Fix use Array.From which copies the value into files instead of using refrences
The model-name detector treated every Qwen model as a Qwen3, falling
into the qwen3_xml parser:
if (n.includes('qwen3') && n.includes('coder')) return 'qwen3_coder';
if (n.includes('qwen')) return 'qwen3_xml'; // catches qwen2.5 too
qwen3_xml is the parser for Qwen3 reasoning/instruct models. Qwen2.5
(and Qwen2, Qwen1.5) ship with hermes-style tool calling, so the
qwen3_xml parser never recognises their tool calls — they leak through
as plain text in the assistant reply and the agent silently fails to
execute anything.
Reproduces with:
vllm serve Qwen/Qwen2.5-Coder-14B-Instruct-AWQ ... \
--enable-auto-tool-choice --tool-call-parser qwen3_xml
→ ask the agent to call any tool → JSON shows up in chat, no call runs.
Fix the ordering:
qwen3 + coder → qwen3_coder
qwen3 → qwen3_xml
qwen → hermes (Qwen2.5 / Qwen2 / Qwen1.5)
Verified against the model matrix:
Qwen2.5-Coder-14B-Instruct-AWQ → hermes
Qwen2.5-7B-Instruct → hermes
Qwen3-8B → qwen3_xml
Qwen3-32B → qwen3_xml
Qwen3-Coder-30B-A3B → qwen3_coder
Qwen2-72B-Instruct → hermes
Qwen1.5-7B-Chat → hermes
The fallback memory extractor (used by routes/memory_routes.py when the LLM
extractor fails) matched list items with `r'^[-*•]|\d+\.\s*(.*)'`. Operator
precedence makes that `(^[-*•]) | (\d+\.\s*(.*))`, so the capture group only
exists on the numbered-list branch.
A bullet line ("- foo") matches the first branch, so `group(1)` is None and
`text_match.group(1).strip()` raises AttributeError — crashing extraction for
any assistant message that contains a bullet list (i.e. most of them). Numbered
lists happened to work.
Group both markers — `r'^(?:[-*•]|\d+\.)\s*(.*)'` — so the capture applies to
bullets and numbers alike.
Adds tests/test_memory_bullet_extraction.py (red before, green after).
read_file/write_file passed the raw path to open(), so a tilde path like
~/notes.txt failed ("not found") — the shell's ~ expansion never happened
because there's no shell. Agents then fell back to bash to reach home-dir
files. Expand ~ (and ~user) with os.path.expanduser before opening.
Checks: python -m py_compile src/tool_execution.py.
TaskScheduler.start() aborts stale TaskRun rows but never advanced
ScheduledTask.next_run. Across a restart the in-process _executing set
is empty, so the first post-restart _check_due_tasks() call dispatches
every task whose next_run is still in the past — and so does every
subsequent poll, until the task's regular _execute_task path finally
runs compute_next_run and pushes it forward.
start() now queries active tasks with next_run < now and pushes each
one to now + 60s. The first poll after restart sees them as not-yet-due,
the task runs once normally, and compute_next_run puts the schedule
back on its real cadence. Paused and not-yet-due tasks are left alone.
The validator test was rewritten as a regression test asserting the
opposite of the bug it originally demonstrated, plus two narrower cases
to lock down the filter (only active+overdue is touched).
* feat: publish all configured email addresses for reply-all exclusion
* fix: exclude all of the user's own addresses from reply-all, not just the active one
* test: reply-all excludes all of the user's configured addresses
* fix: match topic keywords on word boundaries, not substrings
* fix: apply word-boundary matching to topic example snippets too
* test: topic keywords match whole words, not substrings
* macOS/Apple Silicon: detect Metal backend, surface MLX models, brew tmux hint
- hardware.py: add _detect_macos() via sysctl/system_profiler; report
backend=metal + unified_memory on Apple Silicon instead of cpu_arm
- fit.py: add Apple Silicon (M1-M5) unified-memory bandwidths + metal
FALLBACK_K so throughput estimates use the real bandwidth formula
- setup.py: Mac-specific 'brew install tmux' hint
Verified on M5 Pro 48GB: backend=metal, 273GB/s matched, 6 MLX models now
visible (were hidden), cuda still hides MLX, no new test failures.
* Fix native macOS tailnet launch and Metal GPU probe
---------
Co-authored-by: Elijah (Hermes) <hermes@local>
The agent's multi-round (tool-result) follow-up request was rejected with
HTTP 400 on two providers, so tools ran but the agent never produced an answer:
- OpenAI-compatible streaming (Gemini 3) dropped the per-call thought_signature
and collided parallel tool calls, which arrive with index=None: they all
landed in slot 0, overwriting the first call's name and corrupting its
arguments by concatenation, so the follow-up request 400'd. Capture and replay
each call's extra_content (thought_signature), and give every parallel call
its own accumulator slot (allocated above the max key, so sparse or mixed
indices can't collide).
- Native Ollama /api/chat expects object tool-call arguments, but Odysseus
carries them as a JSON string, which Ollama rejected ("Value looks like
object, but can't find closing '}' symbol"). Convert them to objects in the
Ollama payload builder.
Both compose with the no-prose null-content sanitize fix from #862.
Tested: python -m pytest tests/test_llm_core_streaming.py
tests/test_llm_core_ollama.py tests/test_agent_loop.py (53 pass), and
python -m py_compile src/llm_core.py src/agent_loop.py.
Split 2/4 of the companion bridge (#863 was 1/4). A paired bearer-token caller
runs as the sandboxed 'api' pseudo-user, so its sessions were stranded in a
separate 'api'-owned silo, invisible to the owner's desktop UI.
Add effective_user(): for a bearer token it resolves to the token's real owner
(request.state.api_token_owner); for cookie sessions it is identical to
get_current_user, so the swap is a no-op for browser users. Route session
ownership/attribution in routes/session_routes.py through it.
Tests (tests/test_session_owner_attribution.py):
- cookie/browser users are unchanged
- a bearer token attributes to its owner; with no owner it does NOT escalate
- _verify_session_owner: a bearer token for owner A cannot verify owner B's
session (404); owner verifies their own; missing -> 404; unauth -> 403
POST /api/v1/chat (the n8n/Make/Activepieces sync-chat endpoint) verified
session ownership with `_tok_user and _sess_owner and _sess_owner != _tok_user`.
The `_sess_owner and` clause skipped the check entirely whenever the session's
owner was null — so any chat-scoped API token (e.g. a token minted for a paired
mobile device) could pass a legacy/migrated null-owner session id, inject a
message into that session, and read back its conversation history plus reuse
the owner's endpoint credentials.
This is the same `if owner and owner != user` null-owner-bypass pattern that
was already hardened in the gallery, calendar, and notes routes (see
test_null_owner_gates.py) and in session_routes._verify_session_owner. Make
this gate strict and fail closed too: require a resolvable caller and an exact
owner match, mirroring _verify_session_owner. Extract the decision into
_caller_owns_session() and pin it with regression tests.
When the selected model fails before producing output, stream_llm_with_fallback
quietly switches to the next candidate and the reply is shown under the
originally selected model's name, so a misconfigured provider looks like it
works. (Concretely: a Bedrock gateway that 400s every Anthropic/Claude request
appears fine because another model silently answers under the Claude label.)
Emit a `fallback` SSE event ({selected_model, answered_by, reason}) the first
time a non-primary candidate produces output, forward it through the agent loop
and both chat-route paths, stamp the response metrics with the model that
actually answered, and show a notice + relabel the reply in the UI.
Tested: python -m pytest tests/test_llm_core_fallback.py (3 pass);
python -m py_compile src/llm_core.py src/agent_loop.py routes/chat_routes.py;
node --check static/js/chat.js.
When serving with the llama.cpp backend and no .gguf file exists on the host,
the GGUF launcher prelude exits with 'ERROR: No GGUF found on this host', but
_diagnose_serve_output had no matching pattern, so the UI showed a generic
crash instead of explaining the cause. Add a diagnosis pattern for the
no-GGUF case so users are told a .gguf is required and pointed at downloading
a GGUF build, instead of an opaque crash.
Closes#811
Two changes close the cross-tenant topic leak in /api/conversations/topics.
The route at routes/history_routes.py:478 used get_current_user, which
returns None when no auth middleware has set request.state.current_user
(loopback-bypass, AUTH_ENABLED=false, or any path that short-circuits the
middleware). It then forwarded owner=None to analyze_topics.
The helper at src/topic_analyzer.py:21 used an 'if owner:' short-circuit
in its owner filter, so the None owner took the no-filter path and the
helper silently aggregated topic frequencies and per-snippet session_id,
session_name, role, and snippet text across every user's sessions.
analyze_topics now returns an empty result when owner is falsy. The
inner short-circuit is removed because the filter is now strict by
construction. The route is switched to require_user, which raises 401
when auth_manager.is_configured is True and the caller is anonymous,
matching the pattern used by calendar_routes, skills_routes, and other
authenticated routes.
The test test_history_topics_owner_scope.py was rewritten to drive the
real route through FastAPI's TestClient with a stub AuthMiddleware that
mirrors the loopback-bypass branch, and now asserts a strict 401 from
the route and an empty result from the helper. The previous version of
the test accepted either a 200-with-empty-topics or a 401; the strict
assertion means a future regression that drops the require_user wrapper
or re-adds the inner short-circuit is caught immediately.
#718 reported Deep Research drifting into adult / spam URLs several
rounds into a benign session ("research about https://bhagathgoud.com/
and what he doing currently"). The reporter's log showed Japanese
adult sites being crawled even though the model was emitting normal
queries like "Bhagath Goud LinkedIn" and "site:bhagathgoud.com".
The model wasn't generating those URLs. Every provider call site
constructed its params dict without a SafeSearch parameter, so the
underlying HTTP backend (the duckduckgo-search library / DDG's HTML
endpoint in this case) was free to surface "related search" /
trending / spam recommendations that have nothing to do with the
user's query. Per provider:
- SearXNG: instance-dependent; many self-hosted instances default
to safesearch=0.
- Brave API: defaults to "off" for new API keys.
- duckduckgo-search lib: defaults to "moderate", which still lets
related-search recommendations and HTTP-backend fallback URLs
surface trending non-English spam topics.
- DDG HTML fallback (html.duckduckgo.com): no `kp` param, treated
as off.
- Google PSE: omitted `safe` is equivalent to off.
- Serper: omitted `safe` proxies to Google with safe off.
Since the bad URLs entered through the provider layer, not the
model, the provider params are the right place to gate this.
Changes:
- src/settings.py: new `search_safesearch` setting with default
"strict". Documented values ("strict" | "moderate" | "off") plus
a few aliases ("on", "high", "0/1/2", "disabled", ...) so a
hand-edited config doesn't silently fall through to off.
- src/search/providers.py:
- Add `_get_safesearch_level()` (canonical, normalizing) and
`_safesearch_for(provider)` (per-provider param translation).
- Thread the per-provider value into every params dict:
SearXNG JSON, SearXNG language/engines fallbacks, SearXNG HTML,
Brave, DDG library, DDG HTML fallback, Google PSE, Serper.
- Tavily is left untouched — its API has no SafeSearch knob and
its index already filters explicit content at ingest time.
Behavior change for existing installs: default is now "strict", so
explicit results get filtered across every supported provider
without any user action. Users who deliberately want unfiltered
results can set `search_safesearch` to "off" in Settings. No new
dependencies, no schema migrations.
Closes#718.
The agent's RAG tool selector retrieves manage_notes as relevant for
note / todo / reminder requests, but two gaps stopped it from actually
firing on local llama.cpp / vLLM endpoints:
1. FUNCTION_TOOL_SCHEMAS had no entry for manage_notes. Even when the
tool was marked relevant, no JSON schema was sent on the function
tools list, so native-function-calling models had nothing to call.
In practice the model would describe creating the note in prose
while the actual note stayed blank — the symptom reported in #713
("checklist hallucinated as blank").
2. _API_HOSTS only listed hosted providers (OpenAI, Anthropic, etc.).
For local endpoints like http://localhost:8080 or
http://host.docker.internal:8000, _is_api_model fell back to
keyword-sniffing the model name, so any model whose slug didn't
happen to match the keyword list silently lost native tool
schemas entirely.
Fixes:
- src/tool_schemas.py: add a manage_notes function schema covering
list/add/update/delete/toggle_item with the full Keep-style field
set. note_type is exposed as an enum ("note" | "checklist") so the
model picks the mode explicitly instead of inferring it from
content shape. Items are named checklist_items in the schema —
consistent with the description's wording and avoiding the
Python-built-in name clash that #713 calls out.
- src/tool_implementations.py: do_manage_notes accepts both
checklist_items (new, schema-exposed) and items (legacy /
internal). Direct API callers and existing code paths keep
working unchanged; native function calls following the new
schema route through the same path.
- src/agent_loop.py: add localhost, 127.0.0.1, and
host.docker.internal to _API_HOSTS so the function-tool path is
not gated behind model-name guessing for local servers.
Closes#174.
Closes#713.
Deep research asks 2-3 clarifying questions first. When the user answers
with a bare affirmation ('yes', 'ok', 'go ahead'), that short message
becomes latest_message and the query-synthesis fallback returned it
verbatim, so research ran on the literal word 'yes'.
In ResearchHandler.synthesize_query, when synthesis can't run (history
too short) or fails, fall back to the earliest substantive user message
(the original ask) only when the latest message is an explicit
affirmation/continuation phrase or is empty/punctuation-only. There is
deliberately no length heuristic: a short answer like 'UK', 'C++', or
'Rust' in a clarification flow is a real topic and is left untouched.
Tests cover query/topic selection: bare 'yes' -> original ask, short
answers (UK, C++) kept, short-only-substantive message kept, and a
multi-word follow-up still flows through synthesis.
* Fix test suite: ESM loading and stub isolation (refs #605)
Three targeted fixes to reduce suite failures from 9 → 1:
1. package.json: add "type": "module" so Node loads static/js/**
as ES modules. Fixes 7 tests in test_compare_js.py and
test_reply_recipients_js.py that fail with
"SyntaxError: Unexpected token 'export'".
2. test_null_owner_gates.py: add Base and ChatMessage to the
core.database stub. Without Base the scheduler test cannot
import at collection time; without ChatMessage core/__init__.py
fails mid-load when session_manager.py tries to import it,
leaving core partially initialised in sys.modules and poisoning
the auth manager migration test that runs later in the same file.
3. test_task_scheduler_session_delivery.py: skip gracefully when
core.database is stubbed (Base is a MagicMock) rather than
crashing. The test passes correctly when run in isolation.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* Scope ESM declaration to static/js/ and document isolation workaround
Per review feedback on #844:
1. Move "type": "module" from root package.json to static/js/package.json.
The root package.json had no type field (defaulted to CJS) and should
stay that way — vendored UMD bundles in static/lib/ use require() internally
and would break if Node ever tried to load them as ES modules. Node resolves
the nearest package.json, so adding it in static/js/ scopes the ESM
declaration to just the files the JS unit tests actually load
(compare/state.js, emailLibrary/replyRecipients.js).
2. Expand the module-level skip comment in test_task_scheduler_session_delivery
to document that it is a temporary isolation workaround, explain root cause
(test_null_owner_gates installs a module-level sys.modules stub with no
cleanup), record before/after suite numbers, and note the clean path
(refactor to fixture-scoped stub).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Office documents were dropped server-side: .docx fell through to
"[Attached document file]", .xlsx/.pptx weren't recognized at all, and
the personal-docs RAG index only covered txt/md/json/pdf.
Wire the optional markitdown dependency (MIT, Microsoft) into both the
chat-attachment path (build_user_content) and the RAG indexer
(personal_docs), converting .docx/.xlsx/.pptx/.xls/.epub to Markdown.
It is lazy-imported with graceful fallback (mirrors src/pdf_runtime.py):
without it those formats show an "install to extract" banner and the
MIT core is unaffected. pypdf stays the default PDF path.
- src/markitdown_runtime.py: optional-dep loader + convert_to_markdown
- upload_handler: recognize Office/EPUB extensions + MIME types
- document_processor: extract Office docs in the chat else-branch
- personal_docs: index Office docs (DEFAULT_EXTENSIONS + dispatch)
- requirements-optional.txt + ACKNOWLEDGMENTS.md: pinned markitdown 0.1.5
- tests: markitdown_runtime + office index coverage
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Two independent data-integrity bugs:
- services/research/service.py: ResearchService.research() (the public deep-research
API, re-exported from services/__init__) treated the handler return value as a
dict (result.get("sources"/"summary"/...)), but call_research_service() returns a
formatted markdown STRING -> AttributeError: str has no attribute get on EVERY
successful call, making the API unusable for any non-error result. Now uses the
string report as the summary and parses sources from the "### Sources" markdown
section (section-bounded, URL-deduped), with a defensive dict branch for back-compat.
- services/memory/memory_extractor.py: extract_and_store guarded the vector-store
find_similar/add calls only with the .healthy flag set ONCE at init. If the
embedding/ChromaDB backend degraded LATER (OOM, evicted model, remote endpoint
down), those calls raised, the exception escaped the dedup loop, skipped
memory_manager.save(), and was swallowed by the outer try/except -> EVERY
validated fact from the session was silently lost (the function docstring
promises "never raised"). Now falls back to the existing text/fuzzy dedup so
facts are still saved when the vector index is unavailable at runtime.
Tests: test_research_service.py, test_memory_extractor_vector_degraded.py.
The "don't wipe endpoint_url/model on endpoint delete" half of #587 landed
in 6a78b02 (Fix endpoint model preservation for tasks). The three remaining
follow-up pieces from the original PR — flagged in the review on #786 —
are:
- routes/model_routes.py: toggle_model_endpoint (PATCH) now accepts
api_key and base_url, so the admin UI can rotate a key or fix a typo'd
URL without going through delete+recreate. base_url is normalized the
same way the POST handler does (strip /models, /chat/completions,
/completions, /v1/messages, then _normalize_base). Cache invalidation
matches the POST/DELETE paths and the response includes base_url so the
frontend can confirm what was saved.
- routes/chat_routes.py: new _recover_empty_session_model picks
cached_models[0] from the endpoint that matches sess.endpoint_url and
persists it onto the Session row before the LLM call goes out. Wired
into both /api/chat and /api/chat_stream after the existing
_clear_orphaned_session_endpoint guard, so the order is: drop
truly-orphaned sessions first, then heal the "picker showed it, session
never knew" case.
- routes/chat_routes.py: when recovery fails (no endpoint, no cached
models) raise HTTP 400 with a clear message instead of letting
model="" reach the upstream as 401/503.
Closes#587.
