Zunou is the platform where AI and people work together. One company, one team, one workplace, with AI staff who hold real jobs alongside your people. A small company, a big company’s power.
This listing identifies a third-party MCP server. MCPBundles is not affiliated with, sponsored by, or endorsed by the service or its operator unless stated on this page.
This server may ask you to sign in when your AI client connects.
Last probed Nov 16, 2026 · mcp.zunou.ai
155tools discovered
For AI agents — MCPBundles Directory API
For MCP client apps (human connect):
MCP connect URL: https://mcp.zunou.ai/mcp
Skill page: https://www.mcpbundles.com/skills/zunou-mcp-c6e70a93f5
Explore capabilities (155)
read 9
write 16
Showing 25 of 155
Showing 25 of 155 tools from the public preview. Connect the server to inspect the full live tool list.
Propose4
Get3
List3
Delete Table2
Update2
Other capabilities11
List Every Tool
read
list_all_tools
Names and one-line summaries of every Zunou tool, including the ones not described in your tool list (lean tool set). Use it to find the tool for an action you do not see, then run it with call_tool. Cheap and read-only.
List Every Tool — Names and one-line summaries of every Zunou tool, including the ones not described in your tool list (lean tool set). Use it to find the tool for an action you do not see, then run it with call_tool. Cheap and read-only.
Call Any Tool By Name — Run any Zunou tool by name with its arguments object, exactly as if you had called it directly (same validation, same permissions). For tools not described in your tool list; find names with list_all_tools.
Get Managed Agent Connections — Read one managed teammate’s current MCP connector assignments before planning a work split. agentAccountId comes from list_agents. Requires active workforce-management capability and membership in the same workspace, and the target must have you in its managers circle. For yourself use get_my_connections. Returns connector names, enabled/consent state, the configured readOnly restriction, and the last reported runtime handshake with timestamp. configuredScopes are requested OAuth scopes, NOT pro
Get My Connections — Check your own assigned integrations before claiming access. Returns connection names, consent state and whether they are enabled and authorised; never credentials or another agent’s assignments. Assignment alone is not completed authorisation. available means ready to configure/use, not proof that a vendor data call succeeded. runtimeStatus is only the last reported handshake, with its timestamp. requestedScopes are requests, approvedScopes are provider-confirmed (null means unknown). permissio
Get My Notebook — Read your notebook — your own recorded lessons and working notes, the one instruction layer you can edit. Returns your instruction mode too (LEARNING: edits apply immediately; PROPOSE: edits await human approval; LOCKED: read-only), your capacity (limit / length), any PARKED lessons the cap refused (fold them in with consolidate_my_notes), and the health of your [[mem:key]] memory links. Read it before updating so you never lose existing lessons.
Update My Notebook — Update your notebook. ONE LESSON = append: true (a non-append write under half the size of your live notebook is REFUSED as a probable mistake - it would erase every lesson not in it; shrinking on purpose is consolidate_my_notes). In PROPOSE mode an append is sized as it would LAND (live + delta) and parked if that is over the cap - get_my_notes.pendingLength / willExceedOnApproval show the same for a queued one. CAPACITY: your notebook has a per-agent limit (get_my_notes.limit; 12000 by default
Get your MCP into directories
A working endpoint is step one. Directory coverage is the coordinated launch across ChatGPT, Claude, Cursor, the MCP Registry, and community indexes.
What is the Zunou · Do the work of a much bigger company MCP server?
Zunou is the platform where AI and people work together. One company, one team, one workplace, with AI staff who hold real jobs alongside your people. A small company, a big company’s power.
How do I connect Zunou · Do the work of a much bigger company to my AI agent?
Use the MCP endpoint listed on this page in your MCP client configuration. One-click install pills support Claude, Cursor, VS Code, and other hosts. Copy the remote MCP URL if your client needs a manual entry.
Operate this server?
Request removal if you do not want Zunou · Do the work of a much bigger company listed here, or claim the listing if you want to manage it on MCPBundles.
Operate Zunou · Do the work of a much bigger company?
This server appears in the MCPBundles directory. Verify you operate it to take over the listing — name, description, logo, contact email, and skill content. We email a 6-digit code to a maintainer address your server publishes in /.well-known/security.txt or /.well-known/mcpbundles.json. Free, takes about a minute.
MCPBundles publishes this directory for MCP discovery. Except where we host or operate an offering, third-party services run under their own terms. Product and company names on this page are used in a descriptive, identifying way (including under nominative fair use where applicable); they remain the property of their owners. Nothing here grants you rights in those marks, and nothing here is an offer to sell a third party's services. Terms · Request removal
More in developer tools
Other MCP servers in this category from the directory index
Tidy Up My Notebook — The tidy-up pass - run it when get_my_notes shows a capWarning, PARKED lessons, or you are asked to. ONE atomic move: (1) `memories` are written to your own memory (facts, measurements - add verifyWith for environment numbers), (2) every [[mem:key]] link in `content` is verified, (3) the parked lessons in foldParkedIds are folded in, (4) `content` lands as your new notebook - immediately in LEARNING mode, as a reviewable proposal marked 'tidy-up' in PROPOSE mode (the reviewer sees the size delta
Set Agent Notebook Capacity — Raise or lower the notebook capacity tier (12000 / 24000 / 36000 / 48000 chars) of an agent you MANAGE - never your own. Growth is the LAST resort: the API refuses a raise unless the agent has landed a tidy-up (consolidate_my_notes) in the last 14 days, so ask for the tidy-up first and raise only if the notebook is still genuinely irreducible. The reason is shown on the agent's page.
Propose a Skill — Draft a reusable skill into your organization's catalog for human review (lands as DRAFT, never auto-assigned). THE BAR - a skill must clear at least one: (a) a procedure you will REPEAT, (b) a procedure OTHER agents doing this job would follow, (c) a human asked for it. How you happened to finish ONE task is not a skill - that is a notebook lesson or a memory fact; if you cannot name the SECOND time this would run, do not draft it. SEARCH FIRST: list_skills(search) before drafting - a near-dupl
Propose a Platform Improvement — Tell ZUNOU (the platform makers, not your org) about platform friction: a missing MCP tool, a core-skill problem, a blocker you hit, or a capability you need to do your job. Zunou staff read and triage every one; their decision (PLANNED / SHIPPED / DECLINED + note) comes back in whats_new.proposalDecisions as kind PLATFORM. Use this ONLY for how the PLATFORM works - org procedures are propose_skill, personal lessons are your notebook, org knowledge is propose_adapter. Max 10 awaiting triage; mak
Propose a Skill Update — Propose new content for an EXISTING skill you think is wrong or outdated. ADDING a section? Pass append: true with just the new text - it is composed under the live body server-side, nothing to retype. REPLACING? FIRST call get_skill, edit FROM the live body keeping every section you are not changing, and pass baseSha256 (get_skill.contentSha256) so a rewrite from a stale copy is refused instead of silently dropping sections for everyone. Proposals carry currentLength / proposedLength so you and
List Adapters — Your organization's ADAPTERS - reusable shared-context documents (company facts, tone, playbooks) that humans attach to agents; attached ones compose into your CLAUDE.md. Read the catalog before proposing a new one to avoid duplicates.
Propose an Adapter — Draft a new ADAPTER (shared org context - company facts, tone guides, client background) for human review. Use when you learn durable knowledge the WHOLE org's agents should carry, not just you: it lands as DRAFT and a human activates/attaches it. Facts private to you stay in memory; procedures are skills; shared always-on context is an adapter. The result echoes id and slug - that is your proof it landed.
List Tables — Zunou Tables in one scope: structured, typed tables that you AND humans co-edit (they see a live grid in the pulse). Returns each table's id, slug, column schema ({id, name, type, options?}) and rowCount. Scope PULSE (pass pulseId) for shared project data, ORG for org-wide, AGENT (default) for your private working sets. Structured/tabular data belongs HERE - not in notes, memory or chat walls.
Create a Table — Create a Zunou Table with typed columns. Column types: text, number, date, select (pass options), url, checkbox, person (org user id). Column ids are derived from names and STAY STABLE forever - reference them in cells. Received a list (CSV, PDF contents, pasted rows)? Promote it into a table in the relevant pulse instead of a note, then upsert_table_rows. Scope PULSE (pass pulseId) to share with the room, AGENT (default) for private working data.
Update a Table — Rename/re-describe a table or evolve its schema. When passing columns, include EVERY column you want to keep WITH its existing id (from list_tables) - a column whose id you omit is removed (its cell data stays orphaned, not lost). New columns need only name+type.
Query Table Rows — Filtered, sorted, paginated rows of a table (max 100/page - check hasMore). Filters AND together: [{columnId, op, value}] with ops EQ, NEQ, CONTAINS (case-insensitive), GT, LT, EMPTY, NOT_EMPTY. Sort: {columnId, direction: asc|desc}. semantic: search by MEANING instead ('japanese electronic music press') when you don't know exact values. columns: return only those cell columns - use it on wide tables. QUERY for what you need instead of dumping whole tables into context.
Aggregate Table Rows — groupBy counts over a table, with optional filters - the digest tool ("12 warm, 3 overdue") that never dumps rows into context. Returns buckets: [{value, count}] sorted by count.
Import a File Into a Table — SERVER-SIDE bulk import of a CSV/TSV into a table: upload the file first (get_upload_url + curl PUT from your shell), then pass its fileKey - the SERVER parses and inserts, so file contents never pass through your context and the 100-row cap does not apply (table row limit still does). Headers map to column NAMES case-insensitively; unmatched headers are reported back. THE way to promote a big received list into a table.
Upsert Table Rows — THE workhorse: insert or update rows (max 100/call). cells = {columnId: value} using the table's stable column ids. Pass a row id to UPDATE that row (cells you name are merged in - update one cell of one row without touching the rest); omit id to INSERT. Never rebuild a whole table to change a few rows.
Delete Table Rows — Delete rows by id (max 100/call). Soft-deleted - recoverable server-side, but treat as final.
Delete a Table — Delete a whole Zunou Table from its pulse, rows included (soft-deleted server-side, but treat as final). Use it to remove a decommissioned or empty table that would otherwise sit as clutter; delete_table_rows keeps the table. Pass confirm: true after you have checked list_tables/query_table_rows and are sure this is the one.
Remember a Fact — Store an addressable fact in Zunou Memory (scope > category > key > value). Use IMMEDIATELY after learning something durable you'd otherwise ask for twice - client data, contacts, decisions, states. Facts are NOUNS (what is X). A procedure you or a teammate will RUN AGAIN belongs in a skill instead (propose_skill; keep a one-line pointer here) - but a ONE-OFF procedure's steps are fine here as history. Never skill what has no second run. Defaults to your PRIVATE agent memory; scope ORG shares th
Recall Facts — Look up facts in Zunou Memory BEFORE asking a human or re-deriving something - a fact you were told once is probably here. A record that carries verifyWith comes back with verifyFirst: RUN that check before relying on the value, and never paste such a value into a notebook, skill or instruction unverified - it is a measurement of an environment, true when stored, not a durable fact. Exact lookup: pass category + key. Browse/search: pass category, query (case-insensitive; words are ANDed - every
Forget a Fact — Delete a fact from Zunou Memory by scope + category + key (soft-delete; versioned, so history survives and re-remembering the same key restores it). Use for facts that are wrong or obsolete - for facts that CHANGED, just remember the new value instead.
Showing the first 25 tools — full list and schemas: Directory API.
How many tools does Zunou · Do the work of a much bigger company provide?
MCPBundles probed 155 tools on the live server. The tool list on this page reflects what was discovered at the last refresh — connect your client to see the full set available to your session.
What authentication does Zunou · Do the work of a much bigger company require?
Zunou · Do the work of a much bigger company may require signing in to the provider before tools can run. Connect through MCPBundles or your MCP client and complete any provider login when prompted.
Don't want this listed?
Request removal from the MCPBundles directory. We stop publishing this listing on mcpbundles.com.
MCPBundles is an independent platform built on the open Model Context Protocol standard. MCPBundles is not affiliated with, sponsored by, or endorsed by Anthropic PBC.