Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@NoFuss for OmniFocus Show me the tasks in my inbox"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
CLI + MCP interface for AI agents, with two interfaces over the same verified read-only core.
NoFuss helps agents inspect OmniFocus without changing your database. Read exact tasks and projects, query Inbox roots or project work, inspect complete project hierarchies, and prepare workload and review overviews. Selected fields include states, dates, notes, tags and supported notifications. Explicit waiting-tag IDs can classify waiting work for a request without saving a preference.
Version 0.1.0-alpha.1 is a read-only source release on GitHub. There is no npm registry release. With Git and Node.js 22+ installed, build locally:
git clone https://github.com/pgerhardt/nofuss-omnifocus.git
cd nofuss-omnifocus
git checkout v0.1.0-alpha.1
npm ci
npm run buildRun from that checkout; no global executable installation is assumed:
node dist/cli.js doctor
node dist/cli.js overview
node dist/cli.js query tasks --scope inbox --fields id,name --limit 20
node dist/cli.js get project PROJECT_ID --view detail
node dist/cli.js mcpReplace PROJECT_ID with an exact persistent project ID. The get example reads
metadata; use the tree request documented in the CLI guide for a
complete hierarchy. CLI output is JSON, with explicit errors and stable exit codes.
doctor reports connectivity, build information, capabilities and verification limits.
Related MCP server: OmniFocus MCP Server
Configure your client to launch a stdio server using this conceptual configuration; its file format and executable-path requirements depend on the client:
command: /ABSOLUTE/PATH/TO/node
args: ["/ABSOLUTE/PATH/TO/nofuss-omnifocus/dist/cli.js", "mcp"]Replace both paths with your local Node executable and built checkout paths.
The server exposes exactly four read tools: nofuss_get, nofuss_query,
nofuss_overview and nofuss_status. The compatibility executable
nofuss-omnifocus-mcp maps to dist/index.js; from source, run
node dist/index.js for the equivalent MCP entry. Both interfaces use the same core and read semantics.
This alpha exposes no task/project mutation, write tools, sync trigger or arbitrary native-script execution. Task text and notes remain untrusted data, not instructions. Output is bounded, with explicit coverage, unavailable fields and continuation when needed. Follow cursors to obtain complete requested data. Live continuations are not snapshots: concurrent edits can affect later pages. See the read contract.
MCP is recommended for the specifically tested Codex read-only sandbox. In that controlled evaluation, MCP completed the tested workflows while direct CLI native execution was unavailable. Direct CLI remained functional in host-level tests and is useful where shell/native Automation execution is permitted.
No direct-CLI workflow token advantage was demonstrated; failed workflows are not equivalent successful completions. The small evaluation establishes no general performance or token-efficiency winner. Interface choice is client/environment dependent.
For agent authors: retrieve complete needed data, retain/process it deterministically, aggregate before display when appropriate, and expose one domain representation to model context where the client permits. Explicitly requested hierarchies must remain complete; do not shrink them merely to reduce token counts.
Requires macOS, Node.js 22+, and an already running OmniFocus with native Automation access from the execution environment. Live validation covers OmniFocus 4.9.2 (188.3); other versions are not certified. Attachment CRUD, perspectives and sync-completion verification are unavailable. This alpha does not promise universal client compatibility.
MIT license. Preserve the dependency notices. NoFuss for OmniFocus is an independent project and is not affiliated with or endorsed by The Omni Group.
Get 1–20 exact IDs with per-ID outcomes, preserving order and duplicates. entity defaults to task; project returns metadata for any state. Optional tree accepts one project ID: all descendants in native sibling preorder, including completed/dropped work. Tree has its own task view/fields/limit/cursor; follow its live continuation to completion. Brief by default; fields overrides view. Project brief: name/status/type/folder_id; detail adds notes/tags/dates, native direct counts and review interval. Project roots are excluded from task rows. Notes/name use Unicode text windows; tag_ids/notifications use collection windows on exact gets. Follow truncated[field].next_cursor with the matching text/collection field and ID. Unavailable fields are explicit.
| Name | Required | Description | Default |
|---|---|---|---|
| ids | Yes | ||
| text | No | ||
| tree | No | Project only, one ID. All descendants in native sibling preorder; id/parent_id/project_id always present. | |
| view | No | brief | |
| entity | No | task | |
| fields | No | Overrides view. Project fields: id, name, status, type, folder_id, note, tag_ids, due_at, defer_at, effective_due_at, effective_defer_at, created_at, modified_at, completed_at, dropped_at, floating_time_zone, direct_task_count, direct_completed_task_count, last_review_at, next_review_at, review_interval, planned_at, flagged. Task-only fields reject for projects and vice versa. | |
| collection | No | One exact ID and selected field. Native-order element windows; default limit 100, or continued cursor limit. |
| Name | Required | Description |
|---|---|---|
| read_at | Yes | |
| results | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With annotations already covering read-only/idempotent/non-destructive, the description adds substantial behavior beyond them: tree includes completed/dropped descendants in native preorder, project roots are excluded from task rows, defaults, truncation cursors, and explicit unavailable fields. Pagination continuation via truncated[field].next_cursor is genuinely useful. It stops short of explaining error modes.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Very terse for a 7-parameter nested tool and front-loads the core behavior. It is dense and run-on rather than structured, which slightly hurts scannability, but almost every clause carries information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given an output schema exists, return formatting needn't be described, and the description covers defaults, ordering, tree scope, truncation, and cursor continuation. It is close to complete for a complex read tool, with only minor ambiguity about text-window paging.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is only 43%, so the description must compensate and largely does: it explains entity/state behavior, the one-project-ID restriction on tree, that fields overrides view, and how text vs collection windows pair with their cursor fields. Gaps remain around offset/length semantics for text windows.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb and resource ('Get 1–20 exact IDs with per-ID outcomes'), and the phrase 'exact IDs' implicitly contrasts with the bulk-oriented sibling nofuss_query. However, no sibling is named, so the differentiation is left for the agent to infer.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives useful context cues ('entity defaults to task', tree applies to 'one project ID', brief by default) but never states when to choose this over nofuss_query or nofuss_overview. Usage is implied by the exact-ID framing rather than spelled out.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Workload overview in one fresh read: unfinished Inbox roots and all active library projects, including nested/inactive folders. Full-scope counts; one compact project list with next_review_at, review_due (at evaluated_at), and work_state. Remaining work excludes native Completed/Dropped descendant task statuses; available actions have Available/Next/DueSoon/Overdue status, including groups, excluding project roots. Optional waiting_tag_ids (1–20 exact IDs, match any native task tag) adds full-scope waiting counts and compact task IDs; omitted means not requested, never zero. No ancestor-tag expansion or priority judgment. Counts stay complete when list coverage is byte-limited; drill down using nofuss_query active library and nofuss_get project trees. No notes or notifications.
| Name | Required | Description | Default |
|---|---|---|---|
| waiting_tag_ids | No |
| Name | Required | Description |
|---|---|---|
| scope | Yes | |
| counts | Yes | |
| waiting | No | |
| coverage | Yes | |
| projects | Yes | |
| consistency | Yes | |
| unavailable | No | |
| evaluated_at | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already cover the safety profile (readOnly, idempotent, non-destructive, closed-world), and the description adds substantial non-obvious behavior: counts stay complete when list coverage is byte-limited, 'omitted means not requested, never zero', no ancestor-tag expansion, no priority judgment, and no notes/notifications. These are exactly the traits an agent needs and cannot infer from the structured fields.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The scope statement is front-loaded, which is good, but the remainder is a compressed run-on of jargon-dense clauses ('excluding project roots', 'excluding native Completed/Dropped descendant task statuses') that takes real effort to parse. Not wasteful in content, but the structure impedes quick comprehension.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
An output schema exists, so return shape need not be explained, and the description still cues the key returned fields (next_review_at, review_due, work_state). For a one-parameter read tool it covers what's needed, though the boundary between 'counts' and 'compact project list' could be slightly clearer.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description carries the full burden for the single parameter, and it does: waiting_tag_ids takes 1–20 exact IDs matching any native task tag, and its presence adds full-scope waiting counts plus compact task IDs, while omission means 'not requested, never zero'. This meaning is entirely absent from the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific resource and scope: 'Workload overview in one fresh read: unfinished Inbox roots and all active library projects.' The description also names the sibling tools used for the next step ('drill down using nofuss_query active library and nofuss_get project trees'), so an agent can differentiate it from its siblings. It falls short of 5 only because the dense jargon ('work_state', 'full-scope counts') partially obscures the plain purpose.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives clear routing context: this is the broad overview, and nofuss_query/nofuss_get are the drill-down tools for deeper coverage. However, it never states an explicit when-not-to-use or the condition that should send the agent straight to a sibling instead of here.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Query tasks in inbox_roots or one exact project, or projects in library (including nested folders). Project inventories default to all statuses/flags; optional status and flagged filters combine with AND, including flagged:false. Project records reuse exact-get projections. Task project depth defaults to descendants; direct selects immediate children. Tasks exclude project roots and locally dropped work; include_completed admits local completion. Ancestors do not prune tasks. Order: created_at ASC (null first), then ID. Follow live query-bound cursors to completion.
| Name | Required | Description | Default |
|---|---|---|---|
| sort | No | created_at | |
| view | No | brief | |
| depth | No | Project scope only; defaults to descendants. | |
| limit | No | ||
| scope | Yes | ||
| cursor | No | ||
| entity | Yes | ||
| fields | No | ||
| status | No | ||
| flagged | No | ||
| project_id | No | Exact project ID; required only for project scope. | |
| include_completed | No | Task queries only; defaults to false. |
| Name | Required | Description |
|---|---|---|
| items | Yes | |
| read_at | Yes | |
| has_more | Yes | |
| returned | Yes | |
| consistency | Yes | |
| next_cursor | Yes | |
| stop_reason | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Far exceeds the annotations (readOnly/idempotent/non-destructive): it discloses default status/flag inclusion, AND-combination including flagged:false, projection reuse, depth defaults, exclusion of project roots and locally dropped work, include_completed semantics, that ancestors do not prune tasks, exact ordering (created_at ASC, nulls first, then ID), and live cursor semantics.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
It is front-loaded with scope and every clause carries information, but the dense, semicolon-chained fragments make it read like a spec dump rather than flowing guidance. Length is economical yet structure suffers for readability.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a complex 12-parameter tool with an output schema (so return values need not be explained), the description covers query semantics, defaults, filtering, ordering, and pagination well. Minor gaps remain around view/limit/fields behavior, but nothing essential for correct invocation is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
With only 25% schema coverage the description must compensate, and it does for most parameters: entity/scope, project_id (exact match), depth (descendants default vs direct), status, flagged, include_completed, sort order, and cursor usage are all given meaning. It leaves view, limit, and fields largely to the schema, so it is strong but not exhaustive.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The opening clause names a specific verb and resource ('Query tasks ... or projects') and enumerates the three scopes (inbox_roots, one exact project, library). It clearly distinguishes task vs project querying, though it never names the sibling tools (nofuss_get, nofuss_overview) it should be chosen over.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Usage is implied through scope semantics (inbox_roots vs project_id vs library) and filter behavior, giving an agent enough to route by scenario. However there is no explicit when-to-use versus alternatives such as nofuss_get or nofuss_overview, so selection guidance is incomplete.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Observe build/readiness, implemented operations, fresh native declaration support and build-scoped verification/gaps. Declarations do not prove behavior. No private traces, sync trigger or claim of sync completion.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
| Name | Required | Description |
|---|---|---|
| sync | Yes | |
| build | Yes | |
| native | Yes | |
| worker | Yes | |
| capabilities | Yes | |
| verification | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare read-only, idempotent, non-destructive, and closed-world behavior. The description adds meaningful boundaries beyond safety hints: declarations do not prove behavior, and it neither exposes private traces, triggers sync, nor claims sync completion. This helps prevent misinterpretation, though some phrasing is jargon-heavy.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Three sentences with purpose front-loaded and caveats following. No obvious filler, and it is appropriately sized for a zero-parameter status tool. The first sentence is a dense list that could be clearer, but it remains compact and usable.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
An output schema exists, so return-value details need not be explained. With no parameters and safety annotations present, the description supplies enough scope and behavioral boundaries for correct invocation. Some domain-specific terms remain unexplained, but the definition is broadly complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has zero parameters, so there are no parameter semantics to document. Per the rubric, zero parameters yields a baseline of 4; the description appropriately avoids discussing inputs.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb ('Observe') and enumerates what is observed: build/readiness, implemented operations, native declaration support, and build-scoped verification/gaps. Clear enough to understand the tool's purpose, but it does not distinguish itself from siblings nofuss_get, nofuss_query, or nofuss_overview.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No explicit when-to-use or when-not-to-use guidance, and no alternatives are named among the sibling tools. The description implies a status-observation use case, but the agent must infer when to select this tool over nofuss_overview or nofuss_query.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.0-alpha.1nofuss_getnofuss_overviewnofuss_querynofuss_statusScored across 4 tools
The four tools occupy mostly distinct niches: nofuss_get fetches by exact IDs, nofuss_query filters by inbox/project/library, nofuss_overview returns aggregate workload counts, and nofuss_status reports build/readiness. The boundary between nofuss_query and nofuss_overview is somewhat blurry since both read tasks and projects, but the descriptions clarify that overview is a fresh-read summary and query is a drill-down.
All tools share the nofuss_ prefix with snake_case single-word suffixes, giving a predictable pattern. The minor deviation is that get/query read as verbs while status/overview read as nouns, but the convention is still clearly consistent.
Four tools is a compact, coherent set for what appears to be a read-only inspection server, and each earns a place (exact fetch, filtered query, overview, status). It is slightly thin and could arguably merge overview into query, but nothing is redundant or bloated.
Read coverage is solid (exact gets, filtered queries, tree traversal, aggregate overview, status), but there are no mutation tools such as create/update/complete/delete task or project. If the server is intentionally read-only this is fine, but for an OmniFocus task-management domain the surface leaves notable gaps.
Read-only AI access to HAVN properties, leads, tasks, files, media, and analytics.
Read-only AI access to Osboon business card analytics, viewers, links, connections and contacts.
Secure AI access to OpenOak tasks, notes, and Kanban boards.
Task management for people and AI agents, with scoped OAuth access to issues, projects, and docs.
