feat(composio): curate OneDrive/Excel/Todoist + UI preview badge for uncurated toolkits (#2283)#2361
Conversation
…uncurated toolkits (tinyhumansai#2283) Adds curated agent-ready catalogs for OneDrive, Excel, and Todoist so the agent has a usable tool surface when users connect them, and introduces a "Preview" badge on the Skills grid for connected toolkits that lack a catalog so users aren't led into the max-iterations failure documented in tinyhumansai#2283. Refs tinyhumansai#2283. PR tinyhumansai#2293 handles the agent-side fast-fail; this PR covers the catalog + UI guard slices that tinyhumansai#2293 explicitly defers. Rust - New `catalogs_microsoft.rs` with ONE_DRIVE_CURATED + EXCEL_CURATED. - Append TODOIST_CURATED to catalogs_productivity.rs. - Wire the three new toolkits into `catalog_for_toolkit` (including the `ONE_DRIVE_*` → "one" multi-segment prefix alias) and `CAPABILITY_TOOLKITS`. - New `agent_ready_toolkits()` helper + RPC `composio.list_agent_ready_toolkits` → returns sorted slugs with curated catalogs. Single source of truth for the UI guard. - Docstrings on the new catalogs document the no-profile / no-sync / no-memory-ingest contract (criterion tinyhumansai#6 in the issue). Frontend - `listAgentReadyToolkits` RPC wrapper + `useAgentReadyComposioToolkits` hook (one-shot fetch, cached) → returns `Set<string>`. - `ComposioConnectorTile` renders a "Preview" badge with tooltip "Agent integration coming soon" when the toolkit is not agent-ready. While the fetch is in flight we treat every toolkit as agent-ready to avoid a flash of badges on the curated tiles. - i18n: `composio.previewBadge`, `composio.previewTooltip` (English; other locales fall back via existing `en` fallback chain). Tests - 9 new Rust tests covering catalog uniqueness, scope coverage, toolkit→catalog resolution (including the ONE_DRIVE prefix alias), capability matrix integration, and `agent_ready_toolkits()` ordering + content. - 5 new Vitest tests for the API wrapper + the hook's success, failure, and missing-field paths. `cargo check` clean. `cargo fmt` applied. `tsc --noEmit` clean.
|
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 curated catalogs (OneDrive/Excel/Todoist), a backend RPC to list agent-ready toolkits, a typed frontend API wrapper and hook to fetch canonical slugs, and a Skills page preview badge driven by the new hook and i18n strings. ChangesAgent-Ready Composio Toolkits Feature
Sequence Diagram(s)sequenceDiagram
participant SkillsUI as Skills Page
participant Hook as useAgentReadyComposioToolkits
participant Api as composioApi.listAgentReadyToolkits
participant Backend as openhuman.composio_list_agent_ready_toolkits
SkillsUI->>Hook: mount -> fetch agent-ready set
Hook->>Api: listAgentReadyToolkits()
Api->>Backend: openhuman.composio_list_agent_ready_toolkits (RPC)
Backend->>Backend: agent_ready_toolkits() (catalog lookup)
Backend-->>Api: RpcOutcome<ComposioAgentReadyToolkitsResponse>
Api-->>Hook: ComposioAgentReadyToolkitsResponse (flat toolkits[])
Hook-->>SkillsUI: agentReady set, loading/error flags
SkillsUI->>SkillsUI: render tiles (showPreviewBadge computed per tile)
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Suggested labels
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Comment |
coderabbitai
Bot
added
the
feature
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
src/openhuman/composio/providers/mod.rs (1)
219-265: ⚡ Quick winAvoid duplicating the canonical agent-ready slug list.
agent_ready_toolkits()currently hardcodes a second copy of toolkit membership. Deriving it from existing canonical data (CAPABILITY_TOOLKITS+catalog_for_toolkit) will prevent future drift that could mislabel toolkits in the UI.♻️ Proposed refactor
pub fn agent_ready_toolkits() -> Vec<&'static str> { - let mut slugs: Vec<&'static str> = vec![ - // Native providers - "gmail", - "notion", - "github", - // Catalog-only toolkits - "slack", - "discord", - "googlecalendar", - "googledrive", - "googledocs", - "googlesheets", - "outlook", - "microsoft_teams", - "linear", - "jira", - "trello", - "asana", - "dropbox", - "twitter", - "spotify", - "telegram", - "whatsapp", - "shopify", - "stripe", - "hubspot", - "salesforce", - "airtable", - "figma", - "youtube", - "one_drive", - "excel", - "todoist", - ]; + let mut slugs: Vec<&'static str> = CAPABILITY_TOOLKITS + .iter() + .copied() + .filter(|toolkit| catalog_for_toolkit(toolkit).is_some()) + .collect(); slugs.sort_unstable(); + slugs.dedup(); slugs }🤖 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/composio/providers/mod.rs` around lines 219 - 265, The function agent_ready_toolkits() duplicates the canonical toolkit list; instead derive the set from the existing canonical sources by collecting keys from CAPABILITY_TOOLKITS (native/registered toolkits) and adding any toolkit slugs that have a curated catalog via catalog_for_toolkit(), deduplicate, sort_unstable() for stable ordering, and return that Vec<&'static str>; update references to use CAPABILITY_TOOLKITS and catalog_for_toolkit (or the appropriate functions/types) so the agent-ready list stays in sync with the canonical data.app/src/lib/composio/composioApi.test.ts (1)
156-170: ⚡ Quick winAssert the sorted contract explicitly.
At Line 167-169,
toContainchecks membership only, so unsorted/duplicated output would still pass. Since this endpoint is meant to be a stable sorted source, assert full order.♻️ Proposed test tightening
- expect(out.toolkits).toContain('excel'); - expect(out.toolkits).toContain('one_drive'); - expect(out.toolkits).toContain('todoist'); + expect(out.toolkits).toEqual(['excel', 'gmail', 'one_drive', 'todoist']);🤖 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 `@app/src/lib/composio/composioApi.test.ts` around lines 156 - 170, The test for listAgentReadyToolkits uses toContain which only checks membership and can miss ordering or duplicates; update the test to assert the exact sorted array is returned after the envelope is unwrapped (i.e., verify out.toolkits equals ['excel','gmail','one_drive','todoist'] in that order) and keep the existing mockCallCoreRpc call assertion to ensure the RPC method is correct.
🤖 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 `@app/src/pages/Skills.tsx`:
- Around line 333-340: The current consumer of useAgentReadyComposioToolkits
treats an empty agentReadyToolkits (when agentReadyLoading is false) as "not
curated", causing curated toolkits to show Preview; change the logic so that
when agentReadyLoading is false and agentReadyToolkits is empty you treat this
as an RPC failure and fall back to assuming curated toolkits are agent-ready (or
surface an explicit error flag). Concretely, either update
useAgentReadyComposioToolkits to return an error or hasFailed flag (in addition
to agentReady and loading) and use that here, or modify the rendering branch
that reads agentReadyToolkits/agentReadyLoading to detect non-loading empty
results and default curated toolkit slugs to agent-ready instead of marking them
as Preview.
---
Nitpick comments:
In `@app/src/lib/composio/composioApi.test.ts`:
- Around line 156-170: The test for listAgentReadyToolkits uses toContain which
only checks membership and can miss ordering or duplicates; update the test to
assert the exact sorted array is returned after the envelope is unwrapped (i.e.,
verify out.toolkits equals ['excel','gmail','one_drive','todoist'] in that
order) and keep the existing mockCallCoreRpc call assertion to ensure the RPC
method is correct.
In `@src/openhuman/composio/providers/mod.rs`:
- Around line 219-265: The function agent_ready_toolkits() duplicates the
canonical toolkit list; instead derive the set from the existing canonical
sources by collecting keys from CAPABILITY_TOOLKITS (native/registered toolkits)
and adding any toolkit slugs that have a curated catalog via
catalog_for_toolkit(), deduplicate, sort_unstable() for stable ordering, and
return that Vec<&'static str>; update references to use CAPABILITY_TOOLKITS and
catalog_for_toolkit (or the appropriate functions/types) so the agent-ready list
stays in sync with the canonical data.
🪄 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: 6cf69d14-d423-40fa-b62b-2ac43677e41f
📒 Files selected for processing (16)
app/src/lib/composio/composioApi.test.tsapp/src/lib/composio/composioApi.tsapp/src/lib/composio/hooks.test.tsapp/src/lib/composio/hooks.tsapp/src/lib/composio/types.tsapp/src/lib/i18n/en.tsapp/src/pages/Skills.tsxsrc/openhuman/composio/mod.rssrc/openhuman/composio/ops.rssrc/openhuman/composio/providers/catalogs.rssrc/openhuman/composio/providers/catalogs_microsoft.rssrc/openhuman/composio/providers/catalogs_productivity.rssrc/openhuman/composio/providers/descriptions.rssrc/openhuman/composio/providers/mod.rssrc/openhuman/composio/schemas.rssrc/openhuman/composio/types.rs
There was a problem hiding this comment.
🧹 Nitpick comments (1)
app/src/lib/i18n/chunks/en-5.ts (1)
444-445: 💤 Low valueConsider adding a period for consistency.
Most complete-sentence tooltips in this file end with a period (e.g., lines 28, 31, 234). Adding one here would maintain consistency.
📝 Suggested change
'composio.previewTooltip': - 'Agent integration coming soon — you can connect, but the agent can't use this toolkit yet', + 'Agent integration coming soon — you can connect, but the agent can't use this toolkit yet.',🤖 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 `@app/src/lib/i18n/chunks/en-5.ts` around lines 444 - 445, The tooltip string for the key 'composio.previewTooltip' is missing a terminal period; update the value for 'composio.previewTooltip' so the sentence ends with a period (preserve the current wording and the trailing comma after the string) to match the other full-sentence tooltips' punctuation.
🤖 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.
Nitpick comments:
In `@app/src/lib/i18n/chunks/en-5.ts`:
- Around line 444-445: The tooltip string for the key 'composio.previewTooltip'
is missing a terminal period; update the value for 'composio.previewTooltip' so
the sentence ends with a period (preserve the current wording and the trailing
comma after the string) to match the other full-sentence tooltips' punctuation.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 6591977f-4c5a-4eeb-9e0d-03ac5e1b5a8b
📒 Files selected for processing (1)
app/src/lib/i18n/chunks/en-5.ts
…eed locale chunks
CI was failing on:
* test / Frontend Unit Tests — 5 Skills.* test files exploded
because they mock `lib/composio/hooks` but never added the new
`useAgentReadyComposioToolkits` export that this PR added to
Skills.tsx. The page imported `undefined`, called it, and
React crashed before the asserted tiles could render.
* test / i18n Coverage — the two new English keys
(`composio.previewBadge`, `composio.previewTooltip`) lived in
`en.ts` + `en-5.ts` but were missing from all 11 non-English
locale chunks, so the strict coverage script reported
`missing[head]` 11 times.
Tests
- Added a stub `useAgentReadyComposioToolkits: () => ({ agentReady:
new Set<string>(), loading: true, error: null })` to the five
`Skills.*.test.tsx` files that mock `lib/composio/hooks`.
- `loading: true` is deliberate: it triggers the Skills.tsx
early-return that treats every tile as agent-ready and skips the
Preview-badge render. That preserves the pre-tinyhumansai#2283 aria-label /
role text these legacy tests assert on, so nothing else needs to
change.
- SkillsStep.test.tsx does not render `<Skills />` and does not
import the new hook — its mock didn't need updating.
i18n
- Inserted the two new keys with English placeholder values into
the 11 non-English locale chunks (`ar`, `bn`, `es`, `fr`, `hi`,
`id`, `it`, `ko`, `pt`, `ru`, `zh-CN` — all `-5.ts`), right after
the sibling `composio.reconnect` row so the per-chunk position
matches `en-5.ts`. They flow through the existing translator
workflow as "untranslated" entries (the i18n coverage script
surfaces them as such for translators).
Verification
- `pnpm test src/pages/__tests__/Skills` → 5 files, 12 tests
passing.
- `pnpm exec tsx scripts/i18n-coverage.ts` reports `missing: 0`
across all locales for the new keys; the only remaining drift is
the pre-existing "unused English keys" warning that's been there
before this PR.
coderabbitai
Bot
added
the
agent
…ully when agent-ready RPC fails CodeRabbit major finding on app/src/pages/Skills.tsx. Before this fix, when `useAgentReadyComposioToolkits` failed (loading: false, agentReady empty, error set), Skills.tsx evaluated `isAgentReady = false` for every tile and the integrations grid lit up with Preview badges across every curated toolkit — actively misrepresenting the agent surface during a transient RPC failure. Fix - Destructure `error` from the hook (it was already exposed on the return shape; just unused) and treat any non-null error as a signal to fall back to "agent-ready" so curated tiles render without the badge. The UI degrades to the pre-tinyhumansai#2283 behaviour instead of misleading users. - New predicate: isAgentReady = agentReadyLoading || Boolean(agentReadyError) || agentReadyToolkits.has(meta.slug) Tests - `Skills.composio-catalog.test.tsx`: route the agent-ready hook through a module-level `agentReadyState` so individual tests can flip `loading` / `error`. Default state preserves legacy test behaviour (`loading: true`). - New `does not flood the integrations grid with Preview badges when the agent-ready RPC fails` asserts no `data-testid` starting with `composio-preview-badge-` is rendered anywhere in the integrations grid when the hook reports an error. `pnpm test src/pages/__tests__/Skills` → 5 files, 13 tests passing (12 pre-existing + 1 new regression).
graycyrus
left a comment
graycyrus
left a comment
There was a problem hiding this comment.
Review — graycyrus
Solid PR. The three new curated catalogs (OneDrive, Excel, Todoist) are well-structured, scope-tagged, and follow the existing catalog patterns. The Preview badge UX is a clean guard that directly addresses the user confusion documented in #2283. The error-fallback path CodeRabbit flagged has been fixed. Test coverage is strong across both Rust and the frontend.
Changes summary
| Area | Files | What changed |
|---|---|---|
| Rust catalogs | catalogs_microsoft.rs (new), catalogs_productivity.rs, catalogs.rs, mod.rs |
OneDrive (16 actions), Excel (18), Todoist (22) catalogs + wiring into catalog_for_toolkit + agent_ready_toolkits() helper |
| Rust RPC | ops.rs, schemas.rs, types.rs, mod.rs |
New composio.list_agent_ready_toolkits RPC endpoint |
| Frontend API/hooks | composioApi.ts, hooks.ts, types.ts |
listAgentReadyToolkits RPC wrapper + useAgentReadyComposioToolkits hook |
| Frontend UI | Skills.tsx |
Preview badge on uncurated tiles, graceful degradation on RPC failure |
| i18n | en.ts, en-5.ts, + 11 locale chunks |
New composio.previewBadge and composio.previewTooltip keys |
| Tests | 7 test files | 14 new tests + 5 legacy mock updates |
Blocking concern
[major] Closes #2283 will auto-close the issue prematurely. This PR addresses acceptance criteria 1–4 and 7–8, but criterion 5 ("No silent max-iteration loop — agent fails fast with a user-friendly unsupported-toolkit message") is explicitly deferred to companion PR #2293. When this PR merges, GitHub will auto-close #2283, and the remaining fast-fail work could be forgotten. Please change to Part of #2283 or Addresses #2283 so the issue stays open until #2293 also lands.
CodeRabbit dedup
CodeRabbit flagged the RPC-failure fallback (Preview badges flooding curated tiles when the hook errors out). This was addressed in commits 5fe07d0–afbb6d7 with the Boolean(agentReadyError) guard. Not repeated here.
Nice work overall — the catalog data, RPC plumbing, hook, and UI integration are all clean. The two inline items below are minor but worth addressing before merge.
| /// Returned in sorted order to keep the RPC response stable across | ||
| /// builds. | ||
| pub fn agent_ready_toolkits() -> Vec<&'static str> { | ||
| let mut slugs: Vec<&'static str> = vec![ |
There was a problem hiding this comment.
[minor] agent_ready_toolkits() manually duplicates the set of toolkit slugs that catalog_for_toolkit() already knows about. If someone adds a new catalog match arm to catalog_for_toolkit but forgets to add it here, the UI will incorrectly show a Preview badge on a curated toolkit.
The existing test agent_ready_toolkits_includes_new_catalogs_and_is_sorted partially guards against this, but it only checks specific slugs — it won't catch a newly added toolkit that nobody thought to assert on.
Suggestion: consider deriving this list from catalog_for_toolkit by probing CAPABILITY_TOOLKITS (which already lists all supported slugs), or add a test that asserts agent_ready_toolkits() is a superset of CAPABILITY_TOOLKITS. That way a new entry in one place is automatically reflected in both.
| 'webhooks.tunnels.toggleFailed': 'فشل تبديل الصدى', | ||
| 'composio.authExpired': 'انتهت صلاحية المصادقة', | ||
| 'composio.reconnect': 'إعادة الاتصال', | ||
| 'composio.previewBadge': 'Preview', |
There was a problem hiding this comment.
[minor] All 11 non-English locale files have the English text verbatim for both composio.previewBadge and composio.previewTooltip. "Preview" as a badge might be intentional (brand term), but the tooltip — "Agent integration coming soon — you can connect, but the agent can't use this toolkit yet" — should be translated. As-is, Arabic, Bengali, Hindi, Chinese, etc. users will see an English tooltip that defeats the purpose of the i18n system.
If translations aren't ready yet, that's fine — but consider adding a // TODO: translate marker so the i18n-coverage script or a follow-up sweep picks them up.
Addresses @coderabbitai nitpick on app/src/lib/i18n/chunks/en-5.ts:444 — most complete-sentence tooltips in this chunk end with a period; the new composio.previewTooltip was missing one. Applied to en.ts and all 11 non-English locale chunks that mirror the English copy.
| /// Source of truth for the UI "preview / agent integration coming | ||
| /// soon" badge: any connected toolkit whose slug is NOT in this list | ||
| /// can be authorized but lacks a curated tool surface, so the agent | ||
| /// can't use it productively. |
There was a problem hiding this comment.
Reviewed this nitpick (also raised by @coderabbitai). Deferring the structural change to the author: agent_ready_toolkits() deliberately diverges from CAPABILITY_TOOLKITS — the existing test on this file explicitly asserts !slugs.contains("clickup") even though clickup has a curated catalog and a native provider. Auto-deriving the list from CAPABILITY_TOOLKITS ∩ catalog_for_toolkit would flip clickup to agent-ready, which appears to be intentional behavior. If that exclusion is no longer wanted, the right follow-up is to (a) clarify clickup's status, then (b) collapse the lists. Leaving as-is for this PR.
There was a problem hiding this comment.
@senamakel, understood — the divergence is intentional. clickup is in CAPABILITY_TOOLKITS (so it appears in the capability matrix as a native provider) but is deliberately excluded from agent_ready_toolkits(), and the existing test explicitly guards that exclusion. Auto-deriving the agent-ready list from CAPABILITY_TOOLKITS ∩ catalog_for_toolkit would silently promote clickup to agent-ready before its status is resolved. The comment + test together serve as a clear deferred-work marker, so leaving the two lists separate for now is the right call.
✏️ Learnings added
Learnt from: senamakel
Repo: tinyhumansai/openhuman PR: 2361
File: src/openhuman/composio/providers/mod.rs:229-229
Timestamp: 2026-05-22T23:13:33.943Z
Learning: In `src/openhuman/composio/providers/mod.rs`, `agent_ready_toolkits()` deliberately diverges from `CAPABILITY_TOOLKITS`. Specifically, `clickup` appears in `CAPABILITY_TOOLKITS` (has a native provider + curated catalog) but is intentionally excluded from `agent_ready_toolkits()`. The test `agent_ready_toolkits_includes_new_catalogs_and_is_sorted` explicitly asserts `!slugs.contains(&"clickup")`. Do NOT suggest auto-deriving `agent_ready_toolkits()` from `CAPABILITY_TOOLKITS ∩ catalog_for_toolkit`, as that would flip `clickup` to agent-ready prematurely. The structural collapse of the two lists is deferred until `clickup`'s agent-ready status is explicitly clarified.
Note: Learnings are effective only in the context of similar code segments. To apply general coding guidelines, please configure review instructions. You can manage existing learnings in the app.
🧠 Learnings used
Learnt from: sanil-23
Repo: tinyhumansai/openhuman PR: 416
File: src/openhuman/memory/relex.rs:441-464
Timestamp: 2026-04-07T15:49:51.275Z
Learning: When using the `ort` Rust crate v2.x with the `load-dynamic` feature enabled, don’t require individual execution-provider feature flags (e.g., `directml`, `coreml`, `cuda`) alongside `load-dynamic` to get EP registration code. The `ort` crate already compiles EP registration via `#[cfg(any(feature = "load-dynamic", feature = "<ep_name>"))]` guards, and adding per-EP feature flags can pull in static-linking dependencies that conflict with the dynamic loading approach. At runtime, EP availability is determined by what the dynamically loaded ONNX Runtime library (`onnxruntime.dll`/`.so`/`.dylib`) supports; ort docs indicate providers like `directml`/`xnnpack`/`coreml` are available in builds when the platform supports them.
Learnt from: sanil-23
Repo: tinyhumansai/openhuman PR: 416
File: src/openhuman/memory/relex.rs:441-464
Timestamp: 2026-04-07T15:49:51.275Z
Learning: When integrating the `ort` Rust crate v2.x with the `load-dynamic` feature enabled, do NOT also require/enable individual provider EP Cargo features like `directml`, `coreml`, or `cuda`. In `ort` v2.x, EP registration for providers (e.g., DirectML, CoreML, CUDA, etc.) is already compiled in under source-level `#[cfg(any(feature = "load-dynamic", feature = "<provider>"))]` guards, such as in `ep/directml.rs`. Adding provider feature flags alongside `load-dynamic` can pull in static-linking dependencies that conflict with the dynamic-loading approach. Provider availability should be treated as runtime-determined by what the loaded `onnxruntime` library (`onnxruntime.dll`/`libonnxruntime.so`/`libonnxruntime.dylib`) actually supports.
Learnt from: oxoxDev
Repo: tinyhumansai/openhuman PR: 571
File: src/openhuman/local_ai/service/whisper_engine.rs:69-80
Timestamp: 2026-04-14T19:59:04.826Z
Learning: When reviewing Rust code in this repo that uses the upstream `whisper-rs` crate (v0.16.0), do not report `WhisperContextParameters::use_gpu(...)` or `WhisperContextParameters::flash_attn(...)` as missing/invalid APIs. These builder-style methods exist upstream and return `&mut Self`; they are not limited to `WhisperVadContextParams`.
Learnt from: graycyrus
Repo: tinyhumansai/openhuman PR: 1078
File: src/openhuman/agent/agents/welcome/prompt.rs:24-24
Timestamp: 2026-05-01T13:41:00.958Z
Learning: For Rust code under `src/openhuman/**/*.rs`, use `snake_case` for local variables (not `camelCase`). If a local variable name is written in `camelCase`, treat it as a style/lint issue because it will trigger Rust’s `non_snake_case` warning (and related clippy linting, if enabled). Avoid suggesting `camelCase` for any Rust local variable names in this repository.
Learnt from: senamakel
Repo: tinyhumansai/openhuman PR: 1173
File: tests/agent_memory_loader_public.rs:88-88
Timestamp: 2026-05-04T06:50:47.877Z
Learning: In this repository, the general camelCase naming guideline should not be applied to Rust source files. For all .rs files, Rust function (and related) names should use snake_case, and snake_case Rust function names should not be flagged—even for async test functions annotated with attributes like #[tokio::test]. This is consistent with Rust’s non_snake_case lint behavior.
Learnt from: sanil-23
Repo: tinyhumansai/openhuman PR: 1825
File: src/openhuman/composio/tools.rs:1028-1039
Timestamp: 2026-05-15T22:00:07.176Z
Learning: In `src/openhuman/composio` Rust code, avoid relying on a captured `Arc<Config>` for runtime behavior that depends on the current on-disk config. For any `execute()`/client-creation path (e.g., `ProviderContext::execute()` and `ProviderContext::backend_client()`), reload the live config at the start of each invocation using `config_rpc::load_config_with_timeout()` before calling `create_composio_client()`. Treat any `Arc<Config>` stored on `self` as init-time reference only (not for mode selection/key lookup). In tests that assert behavior driven by `load_config_with_timeout()`, isolate config reads by using the `TEST_ENV_LOCK` + an `OPENHUMAN_WORKSPACE` tempdir, and call `config.save()` so the per-call reload reads a deterministic snapshot.
| 'webhooks.tunnels.toggleFailed': 'فشل تبديل الصدى', | ||
| 'composio.authExpired': 'انتهت صلاحية المصادقة', | ||
| 'composio.reconnect': 'إعادة الاتصال', | ||
| 'composio.previewBadge': 'Preview', |
There was a problem hiding this comment.
Deferring: the PR description explicitly calls out the English-text seeding as intentional ("the new follow-up commit seeds the same keys into all 11 non-English locale chunks so the i18n-coverage script stops reporting drift"). Adding a // TODO: translate marker across 11 files would itself drift the chunks vs the script's expected shape. The right follow-up is a dedicated i18n sweep PR that translates these two keys (and other recently-added English-only strings) in one pass.
coderabbitai
Bot
added
the
rust-core
…t stale uncurated test Two CI failures from the merge of main into pr/2361: 1. i18n Coverage + I18nContext/coverage Vitest tests: the original PR seeded composio.previewBadge/previewTooltip into 11 non-English locale chunks but missed German (de-5.ts). Add them with the same English placeholder pattern the other chunks use, so the missing-key gate is satisfied. 2. openhuman::composio::tools::tests::empty_uncurated_toolkits_message_names_agent_unsupported_toolkits: this test was added on main to assert the uncurated-toolkit message names onedrive/excel/todoist — but this PR curates those three, so they're no longer uncurated and uncatalogued_toolkits filters them out, returning None. Repoint the test at slugs that remain uncurated (sharepoint, monday, intercom). Companion test empty_uncurated_toolkits_message_ignores_catalogued_toolkits already covers the gmail/googlesheets curated path.
# Conflicts: # app/src/pages/Skills.tsx # src/openhuman/composio/providers/catalogs.rs
…merge Commit e9c9374 (tinyhumansai#2361) accidentally removed all settings.developerMenu.mcpServer.* and settings.mcpServer.* keys from de-5.ts, leaving German as the only locale without translations for the MCP server settings panel. Other locales (en, ar, bn, es, fr, hi, id, it, ko, pt, ru, zh-CN) still have them. Restored the 18 missing keys with the same German translations they had before tinyhumansai#2361.
3 participants
Summary
composio.list_agent_ready_toolkitsas the single source of truth for the UI guard.Problem
OneDrive, Excel, Todoist (plus ~90 other Composio toolkits) can be connected in the UI but have no curated backend catalog.
composio_list_toolsreturns no usable actions for them, the agent keeps trying to recover, burns throughmax_iterations = 10, and surfaces a generic max-iterations error instead of a clear "toolkit not supported" message.Solution
Rust core
src/openhuman/composio/providers/catalogs_microsoft.rswithONE_DRIVE_CURATED(16 actions) +EXCEL_CURATED(18 actions), Read/Write/Admin scoped.TODOIST_CURATED(22 actions) tocatalogs_productivity.rs.catalog_for_toolkit(including theONE_DRIVE_*→"one"multi-segment prefix alias, mirroring themicrosoft_teamspattern) andCAPABILITY_TOOLKITS.agent_ready_toolkits()helper +composio.list_agent_ready_toolkitsRPC returns sorted slugs.Frontend
listAgentReadyToolkitsRPC wrapper +useAgentReadyComposioToolkits()hook (one-shot fetch, cached) → returnsSet<string>.ComposioConnectorTilerenders a small "Preview" badge with tooltip "Agent integration coming soon" for toolkits without a curated catalog. Badge does not block authorization.Skills.*.test.tsxmocks oflib/composio/hooksto include the new hook (withloading: trueso legacy aria-label assertions keep passing).Submission Checklist
cargo test --lib composio::providers composio::schemas composio::ops_test→ 263 passed, 0 failed.pnpm test src/lib/composio src/pages/__tests__/Skills→ 17 files, 50 tests passing.agent_ready_toolkits,catalog_for_toolkit(new aliases), and the new hook/RPC wrapper has at least one focused test. UI tile change (Preview badge) is small and exercised via the existing Skills tests.## Related— N/A: no matrix row touched.Closes #NNNin the## Relatedsection — see below.Impact
agent_ready_toolkitsis a static slice; the hook fetches once per session and caches.composio.list_toolkitsandcomposio.list_capabilitiesare unchanged.Tests
catalog_for_toolkitresolves the new slugs including theONE_DRIVE_*→oneprefix alias; capability matrix integration;agent_ready_toolkits()ordering + content.cargo test --lib composio::providers composio::schemas composio::ops_test→ 263 passed, 0 failed.pnpm test src/lib/composio src/pages/__tests__/Skills→ 17 files, 50 tests passing.pnpm exec tsx scripts/i18n-coverage.ts→ no drift on the new keys.Related
AI Authored PR Metadata (required for Codex/Linear PRs)
Linear Issue
Commit & Branch
fix/2283-uncurated-toolkit-catalogs(branched fromorigin/mainafter fresh fetch)Validation Run
pnpm --filter openhuman-app format:check— applied via prettier hook (commitfb94e08f).pnpm typecheck— clean.cargo test --lib composio→ 263 passed;pnpm test src/lib/composio src/pages/__tests__/Skills→ 50 passed.cargo fmt --manifest-path Cargo.tomlapplied;cargo checkclean.app/src-tauri/changes.Validation Blocked
command:N/Aerror:N/Aimpact:N/ABehavior Changes
Parity Contract
composio.list_toolkits, andcomposio.list_capabilitiesare unchanged. Legacy Skills page tests pass with the same aria-label and role text.loading: truedefault for the new hook in test mocks keeps Preview badges off so legacy assertions don't drift.Duplicate / Superseded PR Handling
Summary by CodeRabbit
New Features
Tests
Localization