Repository navigation
Conversation
Contributor
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughAdds LM Studio as a selectable local AI provider: new config fields and RPC mappings, LM Studio base-URL normalization and OpenAI-compatible DTOs, provider-selection helpers, service wiring (availability, models, chat completions, diagnostics, assets/bootstrap), routing factory changes, docs updates, and comprehensive tests including an e2e JSON-RPC scenario. ChangesLM Studio Local AI Integration
Sequence Diagram sequenceDiagram
participant Client
participant LocalAiService
participant ConfigStore
participant LMStudioServer
Client->>LocalAiService: request chat (provider=lm_studio, messages)
LocalAiService->>ConfigStore: read local_ai config (base_url, model_id, api_key)
LocalAiService->>LMStudioServer: GET {base}/models (health)
LMStudioServer-->>LocalAiService: models list
LocalAiService->>LMStudioServer: POST {base}/chat/completions (payload)
LMStudioServer-->>LocalAiService: chat completion response
LocalAiService-->>Client: assistant reply (tokens/metadata)
Estimated code review effort 🎯 4 (Complex) | ⏱️ ~45 minutes Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (2)
src/openhuman/config/ops.rs (1)
586-595: ⚡ Quick winAllow explicit clearing for local model IDs.
Line 586 and Line 591 currently ignore empty/whitespace values, so clients cannot clear previously persisted
model_id/chat_model_idvia RPC.💡 Proposed fix
if let Some(model_id) = update.model_id { - if !model_id.trim().is_empty() { - config.local_ai.model_id = model_id.trim().to_string(); - } + config.local_ai.model_id = model_id.trim().to_string(); } if let Some(chat_model_id) = update.chat_model_id { - if !chat_model_id.trim().is_empty() { - config.local_ai.chat_model_id = chat_model_id.trim().to_string(); - } + config.local_ai.chat_model_id = chat_model_id.trim().to_string(); }🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/openhuman/config/ops.rs` around lines 586 - 595, The current code ignores Some("")/whitespace for update.model_id and update.chat_model_id so clients can't clear persisted values; remove the !trim().is_empty() guard and always assign the trimmed string when the Option is Some. Specifically, change the blocks handling update.model_id and update.chat_model_id so that if let Some(model_id) = update.model_id { config.local_ai.model_id = model_id.trim().to_string(); } and similarly for config.local_ai.chat_model_id, allowing explicit clearing by sending an empty/whitespace value.src/openhuman/local_ai/service/bootstrap.rs (1)
306-309: ⚡ Quick winPrefer the new
model_path_for_confighelper for consistency.The new
model_path_for_confighelper (lines 395–401) exists to compute provider-specific URIs, but this code still hardcodes theollama://format. While functionally correct (this path only runs for Ollama after the LM Studio early return), using the helper ensures a single source of truth for model path formatting.♻️ Proposed refactor to use the helper
- status.model_path = Some(format!( - "ollama://{}", - model_ids::effective_chat_model_id(&effective_config) - )); + status.model_path = Some(model_path_for_config(&effective_config));🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/openhuman/local_ai/service/bootstrap.rs` around lines 306 - 309, Replace the hardcoded Ollama URI assignment to status.model_path with a call to the new model_path_for_config helper so formatting is consistent: instead of constructing "ollama://{}" with model_ids::effective_chat_model_id(&effective_config), call model_path_for_config(&effective_config, /* pass effective chat model id or the config that function expects */) and assign its Some(...) result to status.model_path; update references to model_ids::effective_chat_model_id(&effective_config) only if model_path_for_config already uses the effective_config internally.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/api/config.rs`:
- Around line 148-159: The URL-classification function
looks_like_openhuman_backend_endpoint currently returns boolean without emitting
diagnostics; add grep-friendly debug/trace logging inside the function showing
the incoming URL, parse success/failure, extracted host (if any), and final
match decision (e.g., "classified as openhuman backend" vs "not classified"),
using the project's logging macros (debug! or trace!); also add similar
debug/trace messages at the calling routing branch (the fallback/warn path
referenced around the other routing logic) so branch selection and misroutes are
easily traceable.
In `@src/openhuman/local_ai/lm_studio_api.rs`:
- Around line 12-66: Add verbose diagnostics to LM Studio URL/auth resolution:
instrument lm_studio_base_url_from_local_ai to trace each candidate source
inspected (env vars and local_ai.base_url), which candidate passed to
normalize_lm_studio_base_url, and log the final chosen base URL or that
DEFAULT_LM_STUDIO_BASE_URL was used, using a stable grep-friendly prefix like
"[lm-studio]". In normalize_lm_studio_base_url add trace logs for the raw input,
the trimmed/with_scheme result, the without_known_endpoint result, and whether
"/v1" was appended. In apply_lm_studio_auth log whether an API key was present
and applied (e.g. "[lm-studio] auth applied: yes/no") but never log the full key
(only indicate presence/length or masked form). Use the project's standard
logging/tracing macros (trace!/debug!) consistent with other core code.
---
Nitpick comments:
In `@src/openhuman/config/ops.rs`:
- Around line 586-595: The current code ignores Some("")/whitespace for
update.model_id and update.chat_model_id so clients can't clear persisted
values; remove the !trim().is_empty() guard and always assign the trimmed string
when the Option is Some. Specifically, change the blocks handling
update.model_id and update.chat_model_id so that if let Some(model_id) =
update.model_id { config.local_ai.model_id = model_id.trim().to_string(); } and
similarly for config.local_ai.chat_model_id, allowing explicit clearing by
sending an empty/whitespace value.
In `@src/openhuman/local_ai/service/bootstrap.rs`:
- Around line 306-309: Replace the hardcoded Ollama URI assignment to
status.model_path with a call to the new model_path_for_config helper so
formatting is consistent: instead of constructing "ollama://{}" with
model_ids::effective_chat_model_id(&effective_config), call
model_path_for_config(&effective_config, /* pass effective chat model id or the
config that function expects */) and assign its Some(...) result to
status.model_path; update references to
model_ids::effective_chat_model_id(&effective_config) only if
model_path_for_config already uses the effective_config internally.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: f228fb69-59c4-4110-9b73-f3f8c0f6e14b
📒 Files selected for processing (34)
.env.exampleapp/src/utils/tauriCommands/config.test.tsapp/src/utils/tauriCommands/config.tsdocs/TEST-COVERAGE-MATRIX.mddocs/agent-workflows/codex-pr-checklist.mdgitbooks/README.mdgitbooks/features/model-routing/local-ai.mdscripts/codex-pr-preflight.mjsscripts/test-codex-pr-preflight.mjssrc/api/config.rssrc/openhuman/about_app/catalog.rssrc/openhuman/about_app/catalog_tests.rssrc/openhuman/config/ops.rssrc/openhuman/config/ops_tests.rssrc/openhuman/config/schema/local_ai.rssrc/openhuman/config/schemas.rssrc/openhuman/config/schemas_tests.rssrc/openhuman/local_ai/README.mdsrc/openhuman/local_ai/lm_studio_api.rssrc/openhuman/local_ai/mod.rssrc/openhuman/local_ai/model_ids.rssrc/openhuman/local_ai/provider.rssrc/openhuman/local_ai/service/assets.rssrc/openhuman/local_ai/service/bootstrap.rssrc/openhuman/local_ai/service/lm_studio.rssrc/openhuman/local_ai/service/mod.rssrc/openhuman/local_ai/service/ollama_admin.rssrc/openhuman/local_ai/service/ollama_admin_tests.rssrc/openhuman/local_ai/service/public_infer.rssrc/openhuman/local_ai/service/public_infer_tests.rssrc/openhuman/local_ai/types.rssrc/openhuman/routing/factory.rssrc/openhuman/routing/provider.rstests/json_rpc_e2e.rs
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
src/openhuman/local_ai/model_ids.rs (1)
3-6:⚠️ Potential issue | 🟡 Minor | ⚡ Quick winUpdate stale module docs to reflect LM Studio allowlist bypass.
The rustdoc says all
effective_*functions enforce MVP allowlists, buteffective_chat_model_idnow bypasses that logic for LM Studio. Please align the comment with current behavior.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/openhuman/local_ai/model_ids.rs` around lines 3 - 6, Update the module doc to reflect that not all effective_* helpers enforce the MVP allowlist: specifically note that effective_chat_model_id bypasses the MVP allowlist for LM Studio model IDs (while other effective_* functions still fall back to the default MVP model and log a warning when an ID is not allowed). Mention the rationale succinctly and reference effective_chat_model_id and the generic effective_* helpers so reviewers can find and reconcile the documentation with the implemented behavior.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/openhuman/local_ai/model_ids.rs`:
- Around line 77-79: Add grep-friendly debug/trace diagnostics around the new LM
Studio branch and the empty-model return in model_ids.rs: when evaluating
provider_from_config(config) and matching LocalAiProvider::LmStudio, emit a
trace/debug log with a stable prefix like "[local_ai]" indicating the chosen
branch and config details; likewise, in raw_chat_model_id (and the empty-model
return path referenced around lines 100-111) emit a trace/debug log with
"[local_ai]" explaining that an empty model ID is being returned and any
relevant inputs. Use the crate’s tracing/log macros (trace! or debug!) for both
spots and keep messages concise and grep-friendly.
---
Outside diff comments:
In `@src/openhuman/local_ai/model_ids.rs`:
- Around line 3-6: Update the module doc to reflect that not all effective_*
helpers enforce the MVP allowlist: specifically note that
effective_chat_model_id bypasses the MVP allowlist for LM Studio model IDs
(while other effective_* functions still fall back to the default MVP model and
log a warning when an ID is not allowed). Mention the rationale succinctly and
reference effective_chat_model_id and the generic effective_* helpers so
reviewers can find and reconcile the documentation with the implemented
behavior.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 90607537-5ad1-4abe-a898-05998f4381b7
📒 Files selected for processing (8)
app/scripts/e2e-run-session.shsrc/openhuman/about_app/catalog.rssrc/openhuman/config/schema/local_ai.rssrc/openhuman/local_ai/mod.rssrc/openhuman/local_ai/model_ids.rssrc/openhuman/local_ai/service/lm_studio.rssrc/openhuman/routing/factory.rssrc/openhuman/routing/provider.rs
✅ Files skipped from review due to trivial changes (2)
- src/openhuman/routing/provider.rs
- src/openhuman/config/schema/local_ai.rs
🚧 Files skipped from review as they are similar to previous changes (5)
- src/openhuman/local_ai/mod.rs
- src/openhuman/about_app/catalog.rs
- app/scripts/e2e-run-session.sh
- src/openhuman/routing/factory.rs
- src/openhuman/local_ai/service/lm_studio.rs
Member
|
fuck yeah.. LM studio FTW |
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/openhuman/local_ai/service/lm_studio.rs`:
- Around line 44-51: The LM Studio HTTP call paths return Err(String) but lack
structured tracing—add tracing::debug/trace/span logs on each failure branch in
lm_studio.rs (around the HTTP send and subsequent status/body/parse handling).
Specifically, instrument the request/send that uses apply_lm_studio_auth(...)
and the response handling (the branches that map_err like "lm studio models
request failed" and the status/body/parse error branches referenced at ranges
53-74, 133-136, 138-157) to emit contextual fields (url, status, error = %e,
maybe body snippet) so callers can triage; keep existing Err(String) returns but
also call tracing::debug! or tracing::trace! with structured key-value fields
immediately before returning the error.
In `@src/openhuman/local_ai/service/ollama_admin.rs`:
- Around line 1048-1052: The code sets healthy = models_result.is_ok(), which
makes the models_error branch unreachable when healthy is true and collapses
parse/auth/status errors into a generic "server not running" case; modify the
health computation so it reflects the actual upstream status separate from
models_result success (e.g., compute healthy from a distinct status check or
include status_result), and preserve the original models_result error into
models_error when Err(err) so detailed errors are surfaced even if healthy is
true; update the matching logic around models_result (and the analogous block at
1093-1097) to set healthy and models_error independently while keeping the
existing variable names models_result, models_error, and healthy so callers
(e.g., response construction) can report specific parse/auth/status diagnostics.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: a6306e20-fcfa-4afc-bc74-3e827ba58e0f
📒 Files selected for processing (3)
src/api/config.rssrc/openhuman/local_ai/service/lm_studio.rssrc/openhuman/local_ai/service/ollama_admin.rs
🚧 Files skipped from review as they are similar to previous changes (1)
- src/api/config.rs
…routing) - Resolved merge conflict in app/scripts/e2e-run-session.sh (kept PR retry-loop cleanup) - Resolved merge conflict in src/openhuman/local_ai/mod.rs (include install_piper, install_whisper, lm_studio_api, provider, voice_install_common) - src/openhuman/local_ai/service/lm_studio.rs: add #[allow(dead_code)] + doc comment to lm_studio_healthy - src/openhuman/routing/factory.rs: replace inline provider alias match with normalize_provider call - src/openhuman/local_ai/model_ids.rs: raw_chat_model_id no longer falls back to DEFAULT_OLLAMA_MODEL for LM Studio; returns empty string + new test lm_studio_chat_model_returns_empty_when_no_model_configured
…nostics - Add 120 s timeout to LM Studio chat completion requests so long inference runs don't block indefinitely (lm_studio.rs) - Relax ensure_lm_studio_available to probe connectivity only; remove the dead lm_studio_healthy fn that carried an #[allow(dead_code)] attr — the "no models loaded" check was a hard error that prevented bootstrap; the UI surfaces an actionable CTA instead (lm_studio.rs) - Make redact_url_for_log handle bare-host URLs (e.g. "localhost:1234") by retrying with an "http://" prefix so they are redacted consistently rather than returned verbatim (src/api/config.rs) - Guard the duplicate models_error issue push behind `if healthy` in ollama_admin so the low-level error is only surfaced when the server appeared healthy but model-listing failed; the existing "not running" issue is already the canonical message for unreachable servers (ollama_admin.rs)
EvanWangZ
force-pushed
the
codex/OH-1673-lm-studio-local-provider
branch
from
May 15, 2026 16:45
5cf65fc to
c67fc55
Compare
Contributor
Author
|
@coderabbitai review |
Contributor
✅ Actions performedReview triggered.
|
Contributor
There was a problem hiding this comment.
Actionable comments posted: 3
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
src/openhuman/local_ai/service/assets.rs (1)
28-56: 🛠️ Refactor suggestion | 🟠 Major | ⚡ Quick winAdd provider-branch diagnostics for the new LM Studio/Ollama routing paths.
The new provider decision branches should emit debug/trace logs (entry/branch/exit) with stable prefixes and correlation fields so runtime routing is observable.
As per coding guidelines: "
src/**/*.rs: In Rust, uselog/tracingatdebugortracelevel for development-oriented diagnostics on new/changed flows, including logs at entry/exit points, branch decisions, external calls, retries/timeouts, state transitions, and error handling paths."Also applies to: 308-318, 401-411
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/openhuman/local_ai/service/assets.rs` around lines 28 - 56, Add trace/debug diagnostics around the provider routing flow: at entry of the function that calls provider_from_config, emit a trace with a stable prefix (e.g., "assets:provider_routing") and a correlation_id/context; log the selected provider (provider_from_config result) and the branch taken (LocalAiProvider::LmStudio vs LocalAiProvider::Ollama vs default) before calling ollama_binary_present, has_lm_studio_model, or has_model; on each external check (ollama_binary_present, has_lm_studio_model, has_model) log a debug with the target model name and result/Err (unwrap_or paths too), and add an exit trace summarizing chat_ready/vision_ready/embedding_ready so routing decisions are observable (reference functions: provider_from_config, ollama_binary_present, has_lm_studio_model, has_model and enum LocalAiProvider).
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/openhuman/agent/triage/events.rs`:
- Around line 14-48: TEST_EVENTS is a global test mutex that accumulates events
across tests causing cross-test leakage; add a test-only reset function (e.g.,
pub(crate) fn clear_test_events() or reset_test_events()) that locks TEST_EVENTS
(using the same lock/unwrap_or_else pattern as record_test_event) and calls
clear() on the inner Vec, and export it so tests can call it in their
setup/teardown; reference TEST_EVENTS, record_test_event, and
test_events_for_external_id when adding this new reset function.
In `@src/openhuman/local_ai/service/assets.rs`:
- Around line 155-164: The LM Studio branch currently sets provider to "ollama"
but leaves path as None and sets a warning, which contradicts later logic that
bypasses Ollama model bootstrap and blocks embedding downloads; update the
LocalAiProvider::LmStudio handling so that when provider ==
LocalAiProvider::LmStudio you also set path to the same Ollama embedding path
used for Ollama models (instead of None), and adjust the bootstrap/download
logic (the condition that currently bypasses Ollama model bootstrap and the
block that prevents `embedding` downloads) to treat LmStudio the same as Ollama
so embeddings can be downloaded and initialized; keep or update the warning text
to reflect that embeddings use the Ollama path.
In `@src/openhuman/local_ai/service/ollama_admin.rs`:
- Around line 1083-1101: The current logic can push "load_lm_studio_model"
repairs even when models listing failed (models_error is Some), which is
misleading; update the conditional guards around the blocks that inspect models
and chat presence so they only run when models_error.is_none(): change the first
branch that checks healthy && models.is_empty() to require healthy &&
models_error.is_none() && models.is_empty(), and similarly require
models_error.is_none() for the branch that checks chat_found/expected_chat
before pushing a load action; keep the existing separate block that, when
healthy and models_error.is_some(), pushes the "Failed to list LM Studio models:
{err}" issue without suggesting load actions.
---
Outside diff comments:
In `@src/openhuman/local_ai/service/assets.rs`:
- Around line 28-56: Add trace/debug diagnostics around the provider routing
flow: at entry of the function that calls provider_from_config, emit a trace
with a stable prefix (e.g., "assets:provider_routing") and a
correlation_id/context; log the selected provider (provider_from_config result)
and the branch taken (LocalAiProvider::LmStudio vs LocalAiProvider::Ollama vs
default) before calling ollama_binary_present, has_lm_studio_model, or
has_model; on each external check (ollama_binary_present, has_lm_studio_model,
has_model) log a debug with the target model name and result/Err (unwrap_or
paths too), and add an exit trace summarizing
chat_ready/vision_ready/embedding_ready so routing decisions are observable
(reference functions: provider_from_config, ollama_binary_present,
has_lm_studio_model, has_model and enum LocalAiProvider).
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: ea4f97ea-685f-473c-a613-deabeb9eb611
📒 Files selected for processing (45)
.env.exampleapp/.prettierignoreapp/scripts/e2e-run-session.shapp/src/providers/__tests__/CoreStateProvider.test.tsxapp/src/services/api/__tests__/authApi.test.tsapp/src/utils/tauriCommands/config.test.tsapp/src/utils/tauriCommands/config.tsdocs/TEST-COVERAGE-MATRIX.mddocs/agent-workflows/codex-pr-checklist.mdgitbooks/README.mdgitbooks/features/model-routing/local-ai.mdscripts/codex-pr-preflight.mjsscripts/test-codex-pr-preflight.mjssrc/api/config.rssrc/openhuman/about_app/catalog.rssrc/openhuman/about_app/catalog_tests.rssrc/openhuman/accessibility/automation_state.rssrc/openhuman/accessibility/permissions_tests.rssrc/openhuman/agent/triage/escalation.rssrc/openhuman/agent/triage/events.rssrc/openhuman/config/ops.rssrc/openhuman/config/ops_tests.rssrc/openhuman/config/schema/load.rssrc/openhuman/config/schema/load_tests.rssrc/openhuman/config/schema/local_ai.rssrc/openhuman/config/schemas.rssrc/openhuman/config/schemas_tests.rssrc/openhuman/local_ai/README.mdsrc/openhuman/local_ai/lm_studio_api.rssrc/openhuman/local_ai/mod.rssrc/openhuman/local_ai/model_ids.rssrc/openhuman/local_ai/provider.rssrc/openhuman/local_ai/service/assets.rssrc/openhuman/local_ai/service/bootstrap.rssrc/openhuman/local_ai/service/lm_studio.rssrc/openhuman/local_ai/service/mod.rssrc/openhuman/local_ai/service/ollama_admin.rssrc/openhuman/local_ai/service/ollama_admin_tests.rssrc/openhuman/local_ai/service/public_infer.rssrc/openhuman/local_ai/service/public_infer_tests.rssrc/openhuman/local_ai/types.rssrc/openhuman/routing/factory.rssrc/openhuman/routing/provider.rssrc/openhuman/screen_intelligence/tests.rstests/json_rpc_e2e.rs
✅ Files skipped from review due to trivial changes (9)
- app/src/services/api/tests/authApi.test.ts
- app/.prettierignore
- src/openhuman/about_app/catalog_tests.rs
- gitbooks/README.md
- src/openhuman/routing/provider.rs
- src/openhuman/local_ai/provider.rs
- src/openhuman/local_ai/README.md
- src/openhuman/accessibility/automation_state.rs
- src/openhuman/config/schema/local_ai.rs
🚧 Files skipped from review as they are similar to previous changes (24)
- src/openhuman/local_ai/service/mod.rs
- docs/agent-workflows/codex-pr-checklist.md
- app/src/utils/tauriCommands/config.ts
- scripts/test-codex-pr-preflight.mjs
- src/openhuman/local_ai/model_ids.rs
- app/src/utils/tauriCommands/config.test.ts
- src/openhuman/local_ai/service/ollama_admin_tests.rs
- docs/TEST-COVERAGE-MATRIX.md
- src/openhuman/local_ai/service/public_infer_tests.rs
- src/openhuman/local_ai/service/public_infer.rs
- src/openhuman/local_ai/mod.rs
- src/openhuman/config/ops_tests.rs
- tests/json_rpc_e2e.rs
- src/openhuman/routing/factory.rs
- src/openhuman/config/schemas.rs
- src/openhuman/local_ai/types.rs
- src/openhuman/local_ai/lm_studio_api.rs
- gitbooks/features/model-routing/local-ai.md
- src/openhuman/config/schemas_tests.rs
- scripts/codex-pr-preflight.mjs
- src/openhuman/local_ai/service/lm_studio.rs
- src/openhuman/local_ai/service/bootstrap.rs
- src/openhuman/config/ops.rs
- src/api/config.rs
Contributor
Author
|
@coderabbitai review |
Contributor
✅ Actions performedReview triggered.
|
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/openhuman/local_ai/service/assets.rs`:
- Around line 485-516: The early-return branch in download_all_models sets
status.state/embedding_state/warning before calling ensure_ollama_server,
ensure_ollama_model_available, ensure_stt_asset_available and
ensure_tts_asset_available, but any of those functions can return an error and
leave status stuck in "downloading"; update download_all_models to
normalize/restore status on both success and error paths: wrap the preload
sequence (when config.local_ai.preload_embedding_model / preload_stt_model /
preload_tts_voice) in a scoped guard or try/finally-like pattern that ensures
self.status.lock() is reset (state="ready" or previous state,
embedding_state="ready"/"idle" as appropriate, vision_state="disabled", clear
warning and download telemetry) before every return or propagated error from
ensure_ollama_server, ensure_ollama_model_available, ensure_stt_asset_available,
ensure_tts_asset_available; alternatively extract a helper (e.g.,
finalize_lm_studio_status) and call it in both the success path and in the error
handling branch so no error leaves embedding_state="downloading" or stale
warnings.
In `@src/openhuman/local_ai/service/bootstrap.rs`:
- Around line 170-239: The LM Studio bootstrap branch in bootstrap.rs lacks
debug/trace diagnostics; add log::debug or log::trace statements at entry into
the provider_from_config(...) == LocalAiProvider::LmStudio branch, before/after
calling ensure_lm_studio_available(&effective_config), before each preload
decision (checking effective_config.local_ai.preload_embedding_model /
preload_stt_model / preload_tts_voice), around external calls
ensure_ollama_server, ensure_ollama_model_available(&embedding_model,
"embedding"), ensure_stt_asset_available, and ensure_tts_asset_available
(including success/failure outcomes and any returned errors), and log the final
status state transitions when setting status.state to "degraded" or "ready" and
when updating embedding_state/stt_state/tts_state; use consistent “[local_ai]”
prefixes and debug/trace level so these flows are diagnosable.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 34182ee1-c3dc-458a-99e0-e14fbbe77a00
📒 Files selected for processing (6)
src/openhuman/agent/triage/escalation.rssrc/openhuman/agent/triage/events.rssrc/openhuman/local_ai/service/assets.rssrc/openhuman/local_ai/service/bootstrap.rssrc/openhuman/local_ai/service/ollama_admin.rssrc/openhuman/local_ai/service/ollama_admin_tests.rs
🚧 Files skipped from review as they are similar to previous changes (3)
- src/openhuman/agent/triage/events.rs
- src/openhuman/local_ai/service/ollama_admin.rs
- src/openhuman/local_ai/service/ollama_admin_tests.rs
Contributor
Author
|
@coderabbitai resolve |
Contributor
✅ Actions performedComments resolved and changes approved. |
Member
|
LFG |
12 tasks done
7 tasks
25 tasks done
AusAgentSmith
pushed a commit
to AusAgentSmith/openhuman
that referenced
this pull request
May 23, 2026
Co-authored-by: Steven Enamakel <enamakel@tinyhumans.ai>
senamakel
pushed a commit
to nocstah/openhuman
that referenced
this pull request
Sep 11, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
/workspace/openhumansymlinked checkouts pass strict path checks.Problem
Solution
ollamaandlm_studio, includinglm-studioinput compatibility./v1/modelsand/v1/chat/completions, then uses them from local AI admin, diagnostics, prompt, and chat flows.Submission Checklist
diff-cover) meet the gate enforced by.github/workflows/coverage.yml. Runpnpm test:coverageandpnpm test:rustlocally; PRs below 80% on changed lines will not merge.docs/TEST-COVERAGE-MATRIX.mdreflect this change (orN/A: behaviour-only change)## RelatedCloses #NNNin the## RelatedsectionImpact
Related
3.1.4,3.2.4,3.3.3.4AI Authored PR Metadata (required for Codex/Linear PRs)
Linear Issue
Commit & Branch
codex/OH-1673-lm-studio-local-provider8cbd2f63476a3a246439081e25893fa064edd453Validation Run
pnpm --filter openhuman-app format:checkpnpm typecheckbash scripts/test-rust-with-mock.sh api::config::tests -- --nocapture;pnpm --dir app exec vitest run src/utils/tauriCommands/config.test.ts --config test/vitest.config.tscargo fmt --manifest-path Cargo.toml --all --check;cargo check --manifest-path Cargo.toml;pnpm test:rustcargo fmt --manifest-path app/src-tauri/Cargo.toml --all --check;cargo check --manifest-path app/src-tauri/Cargo.tomlAdditional validation:
node scripts/test-codex-pr-preflight.mjscd /workspace/openhuman && node scripts/codex-pr-preflight.mjs --strict-path --lightweightpnpm lintpnpm test:coveragegit diff --checkLM Studio,local model provider, orlmstudio.Validation Blocked
command:N/A — no validation command is blocked.error:N/Aimpact:N/ABehavior Changes
Parity Contract
Duplicate / Superseded PR Handling
codex/OH-1673-lm-studio-local-provider.Summary by CodeRabbit
New Features
Behavior
Documentation
Config
Tests