Cross-checked the deployment against OmniRoute's own docs
(docs/reference/ENVIRONMENT.md, docs/guides/DOCKER_GUIDE.md) and found
three gaps from the original migration:
- OMNIROUTE_WS_BRIDGE_SECRET was entirely missing - ENVIRONMENT.md marks
it REQUIRED (production), for the internal Codex Responses WebSocket
bridge. docker compose config validated fine without it (compose
doesn't know omniroute's own required-var list), so this went
unnoticed until checking the docs directly.
- No mem_limit/OMNIROUTE_MEMORY_MB - the Docker guide is explicit that
the 1024MB default heap is dashboard-only sized; coding-agent workloads
(every client this stack has) need OMNIROUTE_MEMORY_MB=8192 and a
10+ GiB container ceiling. Set both.
- No stop_grace_period - the guide's --stop-timeout 40 equivalent, so
SQLite WAL changes checkpoint back into the main DB file on shutdown
instead of getting killed mid-write.
Redis checked and confirmed correctly absent - OmniRoute uses SQLite
only, no Redis anywhere in its docs.
Still open: whether API_PORT actually isolates /dashboard and /api/*
from the published port, or bridges everything through (see issue #31)
- OmniRoute's own ARCHITECTURE.md doesn't document split-port mode as a
real security boundary, and the live "[API Bridge] ... -> dashboard"
log line is ambiguous. Waiting on a live curl test against
proxy-ai.home before treating that as resolved.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VPZ6TogJiYxG8E4EQBB197
Not omniroute's own internal port (API_PORT stays at its default 20129,
unreconfigured) - just the Docker port mapping, so existing NPM/firewall
config pointed at :4000 keeps working without changes on that end. New
OMNIROUTE_PORT env var is the host side of "OMNIROUTE_PORT:API_PORT" in
docker-compose.yml's ports: entry.
Also corrected docs/proxy-key-onboarding.md's dashboard-access
instructions - DASHBOARD_PORT was never published to the host in the
first place, so "http://<host>:20128" was never actually reachable as
written; documented reaching it via the container's own bridge-network IP
or an SSH port-forward instead.
llama-server remains unexposed (no ports: entry, only expose:) -
unaffected by this change, confirming it stays that way.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VPZ6TogJiYxG8E4EQBB197
Confirmed live on the R9700: omniroute starts up fine ("[API Bridge]
Listening on 0.0.0.0:20129") but docker reported it unhealthy forever -
the healthcheck's python3 -c "..." command can never run (which python3
wget curl node found only node in the image), so it failed every single
check regardless of actual app health.
Switched to a node-based TCP-connect check on the same port instead of an
HTTP GET against /healthz - OmniRoute's own Docker guide already treats a
bare TCP probe as an acceptable liveness check, and this sidesteps needing
to confirm /healthz's exact path/response shape on this image.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VPZ6TogJiYxG8E4EQBB197
No longer needed - every client is a coding CLI behind the OmniRoute
gateway, not a chat UI. Drops the open-webui and qdrant services,
WEBUI_PORT/OPENWEBUI_OMNIROUTE_KEY env vars, and the openwebui-data/
qdrant-data volumes. Qdrant only ever served Open WebUI's own built-in
memory/RAG (unrelated to the gateway-level knowledgebase removed in
472e3a4), so it goes too rather than sit unused.
Docs updated: README, docs/network-access.md (ai.home/ai.haylan.ch
section was entirely about Open WebUI, rewritten around the gateway),
docs/proxy-key-onboarding.md, docs/proxy-request-priority.md.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VPZ6TogJiYxG8E4EQBB197
LiteLLM -> OmniRoute (issue #31, wayfinder map + research tickets #32-37):
replace the litellm/litellm-db services with omniroute, split-port mode
(API_PORT published/reverse-proxied, DASHBOARD_PORT never published -
tighter than litellm's old /ui NPM path-deny rule), 5 new secrets in place
of LITELLM_MASTER_KEY/LITELLM_SALT_KEY, llama-server/searxng registered as
omniroute providers post-boot (no static config.yaml equivalent). No
scripted per-workload key minting yet - omniroute's POST /api/keys needs a
dashboard session, not a static bearer key - so OPENWEBUI_OMNIROUTE_KEY is
a manual step for now (docs/proxy-key-onboarding.md).
Caveat carried into the map and README: OmniRoute's own docs
(docs/security/STEALTH_GUIDE.md, MITM-TPROXY-DECRYPT.md, PUBLIC_CREDS.md
on its release/v3.8.51 branch) describe shipped features for AI-provider
client-detection evasion, system-wide HTTPS interception via a locally
installed root CA, and hiding credentials from secret scanners. Proceeding
anyway was an explicit, informed user decision.
Also drops the gateway-level memory/knowledgebase feature entirely (user:
"I don't need it") - litellm-pgvector, pgvector-db, embedding-server,
scripts/ingest-memory.sh, vendor/litellm-pgvector/, docs/memory-
knowledgebase.md. Open WebUI's own qdrant-backed memory/RAG is unrelated
and untouched. litellm-config.yaml deleted (was kept as a rollback
reference, but there's no rollback path to a feature being deliberately
removed).
Not yet verified against real hardware - see issue #31's open tickets.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VPZ6TogJiYxG8E4EQBB197
tea comments create doesn't exist (add/a is the subcommand); a wayfinder
session hit this live resolving OmniRoute-migration tickets. Also note
that concurrent ticket resolutions racing to edit the same map issue
(a full-body replace, no append) can clobber each other's Decisions-so-far
lines -- observed the same session across 5 parallel research tickets.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VPZ6TogJiYxG8E4EQBB197
LLAMA_CTX_SIZE default 65536 -> 131072: real usage was burning through 64K
fast. ~25.6GB VRAM (17.6GB weights + ~8GB KV cache) on the 32GB R9700,
~6GB headroom — see docs/research/qwen3.8-27b-quant.md for the full table.
Also documents LLAMA_GPU_LAYERS as the RAM-offload knob for this dense
model (no separate RAM-offload flag exists in llama.cpp, and --n-cpu-moe/
--cpu-moe/--override-tensor "exps" are MoE-only, no-ops here), and that
there's no explicit SSD offload tier to enable — llama.cpp's default mmap
already falls back to disk implicitly if GPU+RAM run out.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FCAUsjGNSoJTtK8hyLKg5m
The vector_store_registry block in litellm-config.yaml only seeds the
store into litellm's in-memory registry at boot — the Admin UI's Vector
Stores page (/ui/vector-stores) reads litellm's own DB
(LiteLLM_ManagedVectorStoresTable) instead, via /vector_store/list. A
config-only entry there gets silently deleted from memory the first time
anyone loads that page, since /vector_store/list treats the DB as the
source of truth and removes anything not also present there.
update.sh now registers it in the DB too, via POST /vector_store/new
(idempotent, same pattern as the existing virtual-key minting) — with the
literal resolved key, not the os.environ/... form used in
litellm-config.yaml, since the management API doesn't do config.yaml-style
substitution on request bodies.
Found in the process: /vector_store/update in this litellm version can't
touch litellm_params at all (VectorStoreUpdateRequest has no such field,
so PGVECTOR_API_KEY sent through it is silently dropped) — worth knowing
if this key ever needs rotating; documented in update.sh's comment.
Verified live: registered via the API, confirmed the row appears in
/vector_store/list (what the Admin UI page reads) with the real key, and
re-verified search still works end-to-end afterward.
memory-knowledgebase.md no longer says 'not yet verified' — it's been
smoke-tested end-to-end (direct search + the file_search tool on a chat
completion) and the bugs found are fixed in the preceding commits.
VENDORED.md documents the three local patches on top of the upstream
litellm-pgvector commit so a future re-vendor doesn't silently drop them.
litellm-pgvector's Dockerfile only runs 'prisma generate' (codegen) at
build time — nothing ever ran 'prisma db push' against pgvector-db, so the
vector_stores/embeddings tables plain didn't exist on a fresh deploy. Every
vector store call failed with 'relation "vector_stores" does not exist'.
update.sh now runs the push itself after bringing litellm-pgvector up.
Two bugs, either one fatal:
- litellm-pgvector's embeddings endpoints take a precomputed vector per
chunk (they don't call the embedding model themselves) — the script
posted {content, metadata} with no embedding field, guaranteed 422.
- The batch endpoint expects {"embeddings": [...]}; the script posted a
bare JSON array as the body.
Now embeds each file's chunks via LiteLLM's /v1/embeddings first (using
LITELLM_PGVECTOR_EMBEDDING_KEY, already minted for exactly this) before
batch-inserting. Also made both source files optional — a missing file is
skipped, not a hard failure, since neither exists in this checkout yet.
litellm.aembedding can't infer a provider from a bare model name plus a
custom api_base — litellm-pgvector's embedding_service.py was hitting
'litellm.BadRequestError: LLM Provider NOT provided' on every query-time
embedding (i.e. every search). Same openai/ prefix already used for
qwen3.8-27b-local and local-embedding in litellm-config.yaml.
POST /v1/vector_stores always minted a random UUID for the new store's id,
ignoring the request entirely. litellm-config.yaml's vector_store_registry
addresses this store by a fixed id (memory-and-notes), which could never
match a real row as a result — every search/write 404'd.
Added an optional id field to VectorStoreCreateRequest; create_vector_store
uses it when given, falls back to a random UUID otherwise (unchanged
behavior for callers that don't care).
schema.prisma hardcoded vector(1536), OpenAI ada-002's size. This stack's
actual embedding model is nomic-embed-text-v1.5, confirmed 768-dim live
against /v1/embeddings. Left as-is, the first insert after a schema push
would fail on a pgvector dimension mismatch.
Without it, litellm's pg_vector provider sent Authorization: Bearer None,
which round-tripped to the real api.openai.com and came back with
'Incorrect API key provided: None' — masking the actual problem and making
every vector store call fail. Points at litellm-pgvector's own
LITELLM_PGVECTOR_API_KEY (its SERVER_API_KEY), already present in .env.
Smoke-tested end-to-end against the live deploy for issue #24, alongside
the litellm-pgvector fixes in the following commits.
Today's embedding-server crash-loop (missing nomic-embed-text GGUF) was a
manual step nobody ran. update.sh now runs both downloader profiles
itself, every time, before bringing services up -- no separate command to
remember.
- docker-compose.yml: downloader/downloader-embedding commands gain a
`test -f ... && skip || curl ...` guard, so re-running update.sh never
re-downloads an existing model file.
- scripts/update.sh: runs both profiles after image pull/build, before
service recreation.
- scripts/download-model.sh removed -- folded in, redundant standalone
script.
- README.md / docs/memory-knowledgebase.md updated accordingly.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018WHfjWrSEcGhCoeu6dQfDa
Confirmed in llama-server logs: a real request hit exactly n_gen=4096
(the old floor) and returned no answer -- reasoning_content alone ate
the whole budget before any content was written, exactly the failure
mode this config's own comment predicted. 16384 is the user-chosen
middle ground between 8192 and 32768: ~10 min worst-case at ~26.7 t/s,
well under the 65536-token context window.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018WHfjWrSEcGhCoeu6dQfDa
lazytainer panics on every start with 'Could not determine container ID
of lazytainer': its self-detection (vmorganp/Lazytainer,
configureFromLabels()) matches os.Hostname() against the Docker
container-ID list, but network_mode: host makes the container inherit
the host machine's hostname instead of its own container ID, so the
match never succeeds. Upstream's own example compose file doesn't use
host networking (default bridge + published ports) — nothing about
lazytainer requires it.
It also couldn't have seen the traffic it's meant to watch: llama-server
has no published host port (issue #15) and only exists on the ai-stack
bridge network, which host networking has no visibility into. Joining
ai-stack instead fixes both problems at once.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018WHfjWrSEcGhCoeu6dQfDa
README.md and docs/proxy-key-onboarding.md were edited by later commits
(b996b1f, 24d749b) that the litellm-pgvector revert didn't touch, so they
still named the now-removed MEMORY_RETRIEVAL_EMBEDDING_KEY var. Left
docs/research/langchain-pgvector-vs-litellm-pgvector.md as-is — it's a
dated record of the decision being reversed, not live config.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018WHfjWrSEcGhCoeu6dQfDa
This reverts commit abeadc49c8.
Restores vendor/litellm-pgvector/ and the vector_store_registry wiring
(in-band file_search tool-call support) at the user's request, after
re-confirming against docs.litellm.ai/docs/completion/knowledgebase and
litellm-pgvector's own README that pg_vector is still not an in-process
vector_store_registry backend -- it requires this same standalone
connector service either way, so there is no simpler 'native' path that
was missed. Trading back in: 793 lines of vendored code, the untested
Prisma migration, and the git-context build risk noted in VENDORED.md
(all flagged as unverified against real hardware in issue #24), in
exchange for the file_search in-band tool call memory-retrieval did not
support.
Conflicts resolved on top of later commits (Redis, update.sh key-minting
fold-in):
- .env.example / docs/memory-knowledgebase.md: kept the auto-mint-via-
update.sh language, renamed MEMORY_RETRIEVAL_* back to
LITELLM_PGVECTOR_*.
- scripts/generate-secrets.sh: left deleted -- its job was folded into
update.sh in 24d749b, unrelated to this revert.
- scripts/update.sh: renamed the MEMORY_RETRIEVAL_* secret/mint calls to
LITELLM_PGVECTOR_* to match.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018WHfjWrSEcGhCoeu6dQfDa
New redis service (redis:7-alpine, password-protected, no persistence
volume — everything it holds is cache/coordination state). litellm gets
REDIS_HOST/REDIS_PORT/REDIS_PASSWORD, which is all LiteLLM needs to use it
for router state, rate limits/budgets, and cache invalidation — no
litellm-config.yaml block required (docs.litellm.ai/docs/proxy/caching).
REDIS_PASSWORD added to .env.example and update.sh's auto-generated
secrets.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
set_if_blank/mint_key_if_blank used sed -i "s|^KEY=.*|...|", which only
replaces an existing KEY= line — a no-op if the line isn't there at all.
Any .env created before a var was added to .env.example (SEARXNG_LAN_IP,
PGVECTOR_DB_PASSWORD, MEMORY_RETRIEVAL_*, etc.) silently never got that var
filled in. Both helpers now check for "line missing" as a third case and
append it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
update.sh now creates .env from .env.example if missing, idempotently fills
in every random secret (same logic generate-secrets.sh had, now removed),
resolves SEARXNG_LAN_IP from search.home via the host's own DNS, and mints
OPENWEBUI_LITELLM_KEY / MEMORY_RETRIEVAL_EMBEDDING_KEY through LiteLLM's own
/key/generate API once litellm is up — no more manual Admin UI step for the
stack's own two workload keys. Docs updated to point at update.sh as the
one command; docs/proxy-key-onboarding.md keeps the manual/API steps as the
fallback and for onboarding other workloads.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Per docs/research/langchain-pgvector-vs-litellm-pgvector.md (issue #25):
the vendored litellm-pgvector connector (793 lines, Prisma migrations, a
fragile git-context build) is replaced by a ~90-line FastAPI service
(services/memory-retrieval/) wrapping langchain_postgres.PGVector directly
against pgvector-db. Same gateway boundary — it still calls litellm for
embeddings, nothing talks to Postgres or the model directly except this
service.
- New services/memory-retrieval/ (main.py, Dockerfile, requirements.txt):
POST /ingest, POST /query, GET /health.
- docker-compose.yml: litellm-pgvector service replaced by memory-retrieval;
pgvector-db and embedding-server untouched.
- litellm-config.yaml: vector_store_registry block removed (no
langchain_postgres provider exists to register against; callers query
memory-retrieval directly instead of an in-band file_search tool call —
that mechanism was never confirmed working per issue #24 anyway).
- scripts/ingest-memory.sh rewritten for the new /ingest endpoint (same
per-line chunking, no dedup).
- .env vars renamed: LITELLM_PGVECTOR_API_KEY/LITELLM_PGVECTOR_EMBEDDING_KEY
-> MEMORY_RETRIEVAL_API_KEY/MEMORY_RETRIEVAL_EMBEDDING_KEY.
- vendor/litellm-pgvector/ removed entirely.
- docs/memory-knowledgebase.md updated for the new setup/query flow.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Verdict: switch. See docs/research/langchain-pgvector-vs-litellm-pgvector.md
for full reasoning, LOC comparison, and a concrete replacement service spec.
Research/doc only — no compose/config/vendor changes.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The server's Docker/BuildKit couldn't do a git-context build of a public
github.com repo (fails with "could not read Username ... terminal prompts
disabled" — an auth-shaped error for what should be an anonymous clone).
Rather than debug that, vendor litellm-pgvector's small source tree
directly (vendor/litellm-pgvector/, see VENDORED.md for provenance/update
steps) and build from the local path.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
One script to bring the running stack in sync after any change: git pull,
validate compose config, rebuild the litellm-pgvector local build, re-pull
images, recreate what changed. Documented in README's quickstart, and
CLAUDE.md now tells agents to run it after touching compose/config/scripts
rather than just describing a change as done.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- search_tools block in litellm-config.yaml (SearXNG as a first-class
search_provider, standalone /v1/search endpoint, not a model tool) plus
extra_hosts on the litellm service so it can resolve search.home.
- New embedding-server (nomic-embed-text-v1.5 on a second llama.cpp
instance), pgvector-db, and litellm-pgvector services — LiteLLM's native
knowledgebase feature has no Qdrant backend, so this is the only
self-hosted path (docs/research/litellm-knowledgebase.md).
- vector_store_registry + local-embedding model entry in
litellm-config.yaml, wiring it together.
- scripts/ingest-memory.sh to load data/memory.md and
data/claude-legacy-memory.md into the knowledgebase.
- docs/memory-knowledgebase.md documenting the whole setup; data/
gitignored (personal memory content, not meant to be committed).
- New .env vars (SEARXNG_LAN_IP, PGVECTOR_DB_PASSWORD,
LITELLM_PGVECTOR_API_KEY, LITELLM_PGVECTOR_EMBEDDING_KEY,
EMBEDDING_MODEL_FILE) and generate-secrets.sh support for the
auto-generatable ones.
Resolves#22 and #23 (wayfinder map #21). Not yet verified on real
hardware — see #24.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Answers whether SearXNG is natively supported by LiteLLM's /v1/search
feature, whether it's a model-tool or automatic retrieval, what
docker-compose.yml networking change is needed for the litellm container
to reach the LAN's search.home host, and how it interacts with this
project's known-flaky Qwen3.8-27B tool-calling.
Research only — litellm-config.yaml and docker-compose.yml are unchanged.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Qwen3 spends output tokens on reasoning_content before writing content.
Open WebUI's default chat request doesn't set max_tokens, so it fell
through to llama.cpp's low default and the model ran out mid-thought,
returning finish_reason=length with empty content — no reply shown in
Open WebUI. Confirmed via a manual /v1/chat/completions call: works with
max_tokens=2000, fails without it.
litellm_params.max_tokens is a default, not a cap — any caller (or Open
WebUI's per-model Advanced Params) that sets its own max_tokens still
overrides it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q2CR8yawSf7pwAVYnjwFea
Without these, LiteLLM's /ui falls back to username 'admin' and the
master key as the password, forcing the master key to be typed into
a browser. generate-secrets.sh now fills UI_PASSWORD like the other
secrets.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q2CR8yawSf7pwAVYnjwFea
The curlimages/curl image drops to non-root curl_user (uid 100) by
default, but the models named volume is created root-owned by Docker,
so writes into it failed with 'Permission denied'.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q2CR8yawSf7pwAVYnjwFea
Answers whether the $20/mo Claude Pro subscription can be used as a
LiteLLM backend instead of a pay-per-token Anthropic API key. Finding:
no — Pro/Max OAuth credentials are contractually restricted to Claude
Code / native Anthropic apps (Consumer ToS + Usage Policy, actively
enforced server-side since March 2026). Documents the supported path:
a separate Anthropic Console API key as a model_list entry, same shape
as the existing llama.cpp entry. Same pattern applies to other
providers (ChatGPT Plus, Gemini subscriptions).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Verified against the live GitHub repo/docs (not secondhand): real,
active, functioning AI gateway, ~5 months old, effectively
single-maintainer, BSL 1.1. Fails two of the requirements from
docs/research/proxy-tool-choice.md / issue #9 — no Anthropic Messages
inbound endpoint (VoidLLM's own docs say Claude Code's LLM traffic
can't be routed through it) and no request-priority/queuing feature.
Recommendation: stay on LiteLLM.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Open WebUI now points at litellm instead of llama-server directly, using a
provisioned virtual key. llama-server's host port is dropped (internal-only
on the ai-stack network) since the proxy is the only intended entry point
now. docs/coding-cli-setup.md repointed at the proxy's endpoints/ports with
per-CLI virtual keys instead of the old shared dummy key.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Adds litellm + litellm-db to docker-compose.yml, litellm-config.yaml with
custom shadow-cost pricing (Claude Sonnet 5 reference, per #11) and a
priority-scheduling stub (per #16, needs real-hardware smoke test), and
required LITELLM_MASTER_KEY/SALT_KEY/DB_PASSWORD env vars. Untested on real
hardware — that's #17. Open WebUI/coding CLIs still talk to llama.cpp
directly, migration is #15.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
haylanandClaude-Bot