openloomi-loop
openloomi's Loop — the proactive execution brain that runs inside the OpenLoomi desktop app. Use this skill to inspect state, run a tick, schedule / cancel decision actions, tune preferences, and extend Loop with user-defined decision types, Composio-backed signal channels, or deterministic classifier rules. Triggers: 'openloomi loop', 'loop tick', 'loop schedule', 'loop inbox', 'loop run', 'proactive decisions', 'signal → decision → execute', 'pull signals', 'decision queue', 'register loop type', 'add loop decision type', 'register custom channel', 'add composio channel', 'add loop rule', 'register classifier rule', 'force loop type', 'dry-run loop rule', 'list my loop extensions', 'remove loop type', 'delete loop channel'
git clone --depth 1 https://github.com/melandlabs/openloomi /tmp/openloomi-loop && cp -r /tmp/openloomi-loop/skills/openloomi-loop ~/.claude/skills/openloomi-loopSKILL.md
> **Note:** If OpenLoomi readiness is unknown, use `openloomi-setup` first. If OpenLoomi Desktop is not installed, follow [Getting Started](https://openloomi.ai/docs/getting-started).
# OpenLoomi Loop — The Proactive Execution Brain
Loop pulls signals from connected integrations, classifies them into
typed decisions, and lets the user approve execution from the pet or
the web UI. This skill is a thin Claude-side wrapper around Loop's
HTTP API.
## Where things live
| Concern | Location |
|---|---|
| Business logic | Loop's TypeScript core (closed `DecisionType` + classifier + scheduler) |
| HTTP API | `/api/loop/*` — `state`, `decisions`, `decision/[id]`, `card/[id]`, `connectors`, `brief`, `wrap`, `tick`, `preferences`, `action/*`, `types`, `types/[id]`, `channels`, `channels/[id]`, `classifier-rules`, `classifier-rules/[id]`, `classifier-rules/dry-run` |
| Persistence | `~/.openloomi/loop/{signals.jsonl,decisions.json,status.json,connectors.json,config.json}` |
| Scheduler | Three `ScheduledJob` rows: `loop.tick`, `loop.brief`, `loop.wrap` (registered by the loop scheduler) |
| Pet surface | Tauri Rust thread `loomi-pet-decision-watcher` polls `decisions.json` mtime every 2s and emits `loop:state` / `loop:decision` to bubble + card webviews. The widget supports two built-in themes (`fox`, `capybara`) and a `presenting` state surfaced when a decision moves to `done` before the user has reviewed it — click the bubble to flip back to `happy`. User-editable theme config lives at `~/.openloomi/pet-config.json`. |
| Desktop notifications | Opt-in via `LoopPreferences.desktopNotifications` (default `false`). The pet bubble/card is the primary surface; OS notifications only fire for filtered, actionable decisions. |
## Base URL
| Environment | Base |
|---|---|
| Local desktop (Tauri) — default | `http://localhost:3414` |
| Dev server (`pnpm dev`, `pnpm tauri:dev`) | `http://localhost:3515` |
If unsure, start with `http://localhost:3414`. Loop ships inside the
desktop bundle; the dev port is only relevant when you're running
the web app standalone.
## Auth
Per-user routes (`/tick`, `/decision/[id]` POST, `/preferences`,
`/action/*`) require the same auth as the rest of the app. Token is
the base64-encoded JWT stored at `~/.openloomi/token` — decode it
before use:
```bash
TOKEN=$(cat ~/.openloomi/token | base64 -d)
```
Then pass `-H "Authorization: Bearer $TOKEN"` on every call below.
## API quick reference
| Verb | Path | Use |
|---|---|---|
| GET | `/api/loop/state` | dashboard payload (prefs + counts + connectors + lastTickAt) |
| GET | `/api/loop/decisions?status=pending\|done\|dismissed` | inbox |
| GET | `/api/loop/decision/[id]` | full decision JSON |
| GET | `/api/loop/card/[id]` | card-shaped JSON (`why` / `source_chain` / `dialogue` / `nextStep`) |
| POST | `/api/loop/tick` | run one tick (signals → classify → enqueue) |
| POST | `/api/loop/action/schedule` | `{decision_id, action:"run\|dry\|dismiss\|promote"}` → `{action_id, fire_at}`. Job fires ~30s later; cancellable. |
| DELETE | `/api/loop/action/[id]` | cancel a not-yet-fired scheduled action (409 if already fired) |
| GET | `/api/loop/action/by-decision/[id]` | look up `action_id` for a decision (pet "Open" button) |
| POST | `/api/loop/brief` `{force?}` | build morning brief + enqueue card |
| GET | `/api/loop/brief/content` | render the morning brief as text without enqueuing |
| POST | `/api/loop/wrap` `{force?}` | build evening wrap + enqueue card |
| GET | `/api/loop/wrap/content` | render the evening wrap as text without enqueuing |
| GET | `/api/loop/preferences` | read prefs |
| PUT | `/api/loop/preferences` `{...patch}` | write prefs + sync the 3 `ScheduledJob` rows |
| GET | `/api/loop/connectors?refresh=1` | list integration health |
| GET | `/api/loop/types` | list user-defined decision types (per-user extension to the closed `DecisionType` union) |
| PUT | `/api/loop/types` `{id,label,icon,actionKind,description?}` | upsert a custom decision type. `actionKind` must be one of the 15 built-in `ActionKind` literals; `id` must not collide with a built-in `DecisionType`. |
| DELETE | `/api/loop/types/[id]` | remove a custom decision type |
| GET | `/api/loop/channels` | list user-defined signal channels (Composio-backed pullers) |
| PUT | `/api/loop/channels` `{id,label,toolkit,toolSlug,pollIntervalSec,signalType,payloadShape?,eventFilter?}` | upsert a custom channel. `toolSlug` follows the `VENDOR_ACTION` convention (e.g. `STRIPE_LIST_CHARGES`); the watcher invokes it via the `composio` CLI on the registered cadence. |
| DELETE | `/api/loop/channels/[id]` | remove a custom signal channel |
| GET | `/api/loop/classifier-rules` | list user-defined deterministic classifier rules (override the LLM's classification when conditions match) |
| PUT | `/api/loop/classifier-rules` `{id,label?,when[],then{type,actionKind?,confidence?},description?}` | upsert a classifier rule. `when` is a non-empty array of up to 8 `{field,op,value?\|pattern?}` predicates (`signal.type` / `signal.payload.*` paths; ops: `eq` `neq` `contains` `matches` `startsWith` `endsWith` `gt` `lt` `gte` `lte` `exists` `absent`). `then.type` is a built-in or custom `DecisionType`, or `"noop"` to suppress the decision entirely. |
| DELETE | `/api/loop/classifier-rules/[id]` | remove a classifier rule |
| POST | `/api/loop/classifier-rules/dry-run` `{signal}` | preview which rules would match a given signal — returns `{matches:[{ruleId,then}], trace:[{ruleId,matched}], totalRules}`. Pure read; does not mutate state. |
## Examples
```bash
BASE="http://localhost:3414" # or http://localhost:3515
TOKEN=$(cat ~/.openloomi/token | base64 -d)
# Dashboard snapshot
curl -sS "$BASE/api/loop/state" -H "Authorization: Bearer $TOKEN" | jq .
# Run one tick
curl -sS -X POST "$BASE/api/loop/tick" -H "Authorization: Bearer $TOKEN"
# List pending decisions
curl -sS "$BASE/api/loop/decisions?status=pending" \
-H "Authorization: Bearer $TOKEN" |Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
Drive a native macOS app via the cua-driver CLI (default) or MCP server — snapshot its AX tree, click/type/scroll by element_index, verify via re-snapshot. Use when the user asks you to operate, drive, automate, or perform a GUI task in a real macOS application on the host (e.g. "open a file in TextEdit", "navigate to /Applications in Finder", "click the Save button in Numbers").
Use this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files). Triggers include: any mention of "Word doc", "word document", ".docx", or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a "report", "memo", "letter", "template", or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation.
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML/CSS layouts, or when styling/beautifying any web UI). Generates creative, polished code and UI design that avoids generic AI aesthetics.
openloomi HTTP API reference (local-first, served from the OpenLoomi Desktop app at http://localhost:3414). Use when working with openloomi backend routes — auth, AI, files, integrations, RAG, memory, Loop, pet, workspace, platform callbacks. Triggers: API endpoints, backend routes, /api/*, local API, port 3414, integrations REST, OAuth start, RAG search, loop state, memory search, pet state, audit logs
openloomi Connectors tools - manage the native 7 messaging integrations and pair with the composio skill for the 1000+ apps OAuth layer (Slack, Discord, X, Gmail, Outlook, Google Calendar/Drive/Docs, GitHub, Notion, Linear, HubSpot, LinkedIn, Jira, Asana). Triggers: connect platform, integration status, list accounts, disconnect, list-accounts, status, connect, send-reply, native vs composio, 1000+ apps, list connections.
Use this when users ask about openloomi features, capabilities, or how to use it. Examples: 'openloomi 怎么用', '你能做什么', 'What can you do?', 'How does openloomi work?', 'Tell me about openloomi features', 'What platforms does openloomi support?', 'How do I use scheduled tasks?', 'What is Loop?', 'How does the attention agent work?', 'What is a Decision Card?', 'How do connectors work?', 'How do I extend Loop with custom types?', 'What is a classifier rule?', 'How do I plug openloomi into Claude Code / Codex?'
openloomi Memory tools - search and manage the holistic context (people, projects, decisions, knowledge base, chat insights). Triggers: memory search, knowledge base, search documents, list insights, who is John, what did we decide about X, tiered memory, knowledge graph, people/projects/decisions, search-all, conversation memory