In Parallel MCP tools reference
The full set of tools your AI assistant can use once you connect it to In Parallel over MCP — grouped by what they do, and which ones read versus write.
Updated 23 June 2026
When you connect an AI assistant to In Parallel over MCP, it gains a set of tools — named actions it can call to read your execution data and act on it. This article lists those tools, grouped by what they do, so you know exactly what a connected assistant can reach.
You don’t need to memorise tool names — your assistant picks the right tool for what you ask. This reference is here for when you want to understand the surface, set per-tool permissions, or write a prompt that’s specific about what to use.
How to read this list
- Read tools look things up and never change your data. They’re safe to allow without review.
- Write tools create or change records. In clients that support per-tool permissions (like Claude), write and delete tools default to Needs approval — see Connect your AI tool.
- ⚠ Irreversible marks a tool whose effect can’t be undone (deleting a Workspace, publishing a changelog entry). Keep these on Needs approval.
A few names to translate between the UI and the tools:
- Workspace — In Parallel’s unit of work (a team, project, account, or mission). Most tools take a
workspace_id; your assistant gets it fromlist_workspacesfirst. - Observations — the structured signals In Parallel captures from meetings, such as action items, decisions, and risks. The tools use the word observations; the UI surfaces specific kinds in pages such as Action Items or Risk Register.
- Topics — the topic vocabulary on observations and meetings. The
..._tag_taxonomytool uses the older word tags.
The exact list your assistant sees is always the live one. In Parallel adds and refines tools over time, and your AI client fetches the current set when it connects. If a tool here doesn’t appear (or a new one does), trust the client’s list — this reference is a snapshot verified on the date above.
Find your way around
Start-here tools: which Workspaces you can reach, who your connection is, and how to search.
| Tool | What it does | Type |
|---|---|---|
list_workspaces | List every Workspace you can access. Almost every other tool needs a Workspace from here. | Read |
whoami | Show who your connection authenticates as, what it can reach, and when the credential expires. | Read |
search_observations | Keyword search across observations (title and linked detail) using full-text search. | Read |
list_observation_tag_taxonomy | List your organisation’s Topics vocabulary — the tags used on observations and meetings. | Read |
Meetings and transcripts
Read what happened in meetings, and re-file a meeting under the right Workspace.
| Tool | What it does | Type |
|---|---|---|
list_meeting_records | List recent meetings (newest first); optionally search by topic. | Read |
get_meeting_record | Get one meeting in full — attendees, summary, decisions, and action items. | Read |
get_meeting_records | Get several meetings at once (the batch version of the above). | Read |
get_transcript | Get the speaker-attributed transcript — exactly who said what, with timestamps. | Read |
update_meeting_workspaces | Reassign a meeting to different Workspace(s). | Write ⚠ |
Observations
Observations are grouped by kind. Each kind has a list_* tool to find them and a get_* tool to open one in full. Some kinds also support creating, updating, or closing.
| Tool | What it does | Type |
|---|---|---|
list_action_items / get_action_item | Find tasks and action items (filter by status, owner, due date, overdue) and open one. | Read |
create_action_item | Create a new action item owned by a person. | Write |
update_action_item | Change an action item’s status, title, description, owner, or due date. | Write |
close_action_items | Mark one or more action items done. | Write |
list_decisions / get_decision | Find decisions and open one with its rationale and source. | Read |
create_decision | Record a decision made in a Workspace. | Write |
update_decision | Change a decision’s status, title, or rationale. | Write |
list_opportunities / get_opportunity | Find growth/business opportunities and open one. | Read |
list_risks / get_risk | Find risks — threats and blockers that could delay the plan — and open one. | Read |
create_risk | Record a risk raised in a Workspace. | Write |
list_open_questions / get_open_question | Find unanswered questions and open one. | Read |
create_open_question | Record a question that needs an answer. | Write |
resolve_open_questions | Mark one or more open questions answered. | Write |
list_learnings / get_learning | Find captured lessons and insights and open one. | Read |
get_observations | Get full detail for a batch of observations of any kind in one call. | Read |
reassign_observations | Link observations to a goal, or unlink them. | Write |
Legacy plan tools and health
Some older accounts expose plan-specific MCP tools. These tool names describe the legacy API surface; they are not a standard product element. Trust the live tool list in your client. Health and goal tools can appear independently.
| Tool | What it does | Type |
|---|---|---|
get_execution_plan | Get the legacy plan view for a Workspace when that tool is exposed. | Read |
list_plan_versions | List saved snapshots of the plan, newest first, to pick a comparison point. | Read |
get_plan_diff | Show how the plan changed between two points in time, as a diff. | Read |
get_drift_report | Run an execution-health check: overdue action items, stale decisions, unresolved risks, and other drift signals. | Read |
get_goal_metrics | Report goal health for a Workspace (time-to-first-goal, coverage, outdated-goal rate). | Read |
Goals
Propose goals from a meeting or a document, create and curate them, and manage each goal’s action-plan summary.
| Tool | What it does | Type |
|---|---|---|
propose_goals_from_text | Suggest goal candidates from freeform text you paste in. | Read |
propose_goals_from_meeting | Suggest goal candidates from a meeting’s transcript. | Read |
create_goal | Create a goal (typically after accepting a proposed candidate). | Write |
curate_goal | Act on an outdated goal — extend its deadline, mark it done, archive it, or keep it as-is. | Write |
list_outdated_goals | List goals whose deadline has passed and which haven’t been reviewed recently. | Read |
upload_document_and_propose_goals | Upload a document and extract goal candidates from it. | Write |
re_propose_goals_from_document | Re-run goal extraction on a previously uploaded document. | Read |
list_workspace_documents | List documents uploaded to a Workspace (metadata only). | Read |
generate_goal_action_plan_summary | Generate (or regenerate) a goal’s action-plan summary from its observations. | Write |
get_goal_action_plan_summary | Read a goal’s current action-plan summary. | Read |
update_goal_action_plan_summary | Replace a goal’s action-plan summary with your own text. | Write |
Changelog — review what changed
The changelog is the “what changed and who cares” layer: meeting-derived changes an owner accepts or rejects. See Changelogs for the feature itself.
| Tool | What it does | Type |
|---|---|---|
get_changelog | List changelog entries for a Workspace (filter by status or meeting). | Read |
get_changelog_entry | Open one changelog entry with its source observations. | Read |
publish_changelog_entry | Accept a pending entry and publish its proposed changes. | Write ⚠ |
reject_changelog_entry | Reject a pending entry — discard the proposed change. | Write ⚠ |
Changelog tools only reach entries in Workspaces you own.
Workspaces (admin)
Create and manage Workspaces and their members.
| Tool | What it does | Type |
|---|---|---|
create_workspace | Create a new Workspace for your organisation. | Write |
update_workspace | Change a Workspace’s name, description, category, or “how we work” soul. | Write |
manage_workspace_members | Add or remove members, or assign ownership. | Write ⚠ |
delete_workspace | Permanently delete a Workspace and its dependent data. | Write ⚠ |
Pre-wired workflows (prompts)
Beyond individual tools, In Parallel ships a couple of prompts — ready-made multi-step workflows your client can run by name (look for them in your client’s prompt or slash-command menu).
| Prompt | What it does |
|---|---|
weekly_workspace_review | A weekly review of a Workspace: pulls the health report, then open decisions and questions, then recent meetings. |
triage_action_items | Triage a Workspace’s open action items: lists the open work (optionally for one owner), then helps assign and schedule it. |
Privacy
When a Workspace Owner turns on MCP privacy, recognized people in MCP responses use stable pseudonyms instead of their directory names and email addresses. Filenames, URLs, object names, and unrecognized names in free text can remain visible.
Related
- Connect your AI tool to In Parallel — set up the connection and per-tool permissions.
- Use the In Parallel MCP with Jira in Claude — a worked example that combines these tools with a second connector.
- MCP privacy: mask people in AI responses — hide names and emails from your assistant.
- AI in In Parallel — how the AI layer works, and what it doesn’t do.