Tools available to your AI agent

fetch_entities

Any project, meeting, spec sheet, requirement or story, one at a time or as a list, across the workspace or within one project.

fetch_ai_memory

A project's AI Memory: the documents Spectr keeps current about the product, its features, its architecture, open questions and the team.

Reading a file's content needs a plan that includes AI Memory.

fetch_budget

A project's Harvest budget: percent used, health, hours and the projection of committed work. Or every active project's status at once.

Money figures only for admins and owners, unless the budget is shared with everyone. The same as the Budget page.

fetch_team

A project's team across every connected tool, with Forecast allocation. Or every active project's team at once.

Roles and titles only for admins and owners.

fetch_team_time_stats

Each team member's logged Harvest hours on a project: all time, this week and last week.

fetch_scope_creep

Items flagged as outside a project's scope, with their status history and a summary of where they stand.

fetch_integrations

The tools a project is connected to, and whether each connection is working or needs a reconnect.

list_provider_meetings

The recent meetings in a project's Fathom, Granola, Read.ai or Fireflies, marked when already imported or matching the project's keywords.

fetch_workstreams

The Development board's workstreams by state, or one workstream with its stories, which of them were sent to a coding agent, and its agent runs with their links.

fetch_connected_accounts

Your own accounts on coding agent services such as Conductor: their names and whether each key still works. Never the keys.

Only ever your own accounts, whatever your role.

list_agent_run_projects

The repositories one of your Conductor accounts can start work in, with the one matching the project's GitHub repository picked out.

list_agent_run_link_options

The open Conductor workspaces one of your accounts can see, searchable by name and repository, to link to a workstream.

preview_agent_run_message

The exact message Spectr would send a coding agent for some of a workstream's stories and a note, to read before it is sent.

list_workspaces

The workspaces you approved for this agent, with your role in each.

fetch_workspace_members

The workspace's members: name, email, role and when they joined.

fetch_plan

The workspace's plan, how much of each monthly allowance is used, what higher plans add, and how to upgrade.

fetch_billing_activity

The Billing page's activity: every spec sheet and requirement revision with the AI tokens it used, and AI Assistant usage.

assign_project_roles

Matches the project's team to its roles from AI Memory, as the Team page's AI fill does. A role someone set by hand is never changed.

Needs a paid plan with AI Memory.

Counts toward AI Assistant tokens.

regenerate_project_memory

Rewrites the project's six AI Memory documents from scratch, from its requirements, spec sheets, context files and meetings. Custom documents are left alone.

Needs a paid plan with AI Memory.

Counts toward AI Assistant tokens.

refine_memory_document

Rewrites one AI Memory document following your instruction, as Refine with AI does.

Needs a paid plan with AI Memory.

Counts toward AI Assistant tokens.

generate_project_description

Replaces the project's description with one written from its AI Memory, and refreshes its keywords.

Needs a paid plan with AI Memory.

Counts toward AI Assistant tokens.

refresh_story_status

Re-reads published stories from Linear, Shortcut or Jira: status, assignees, labels, type and priority. Run it before refreshing a budget so its projection uses current stories.

A project refreshed in the last five minutes is skipped.

sync_project_team

Re-reads a project's team from its connected tools, or every active project's.

refresh_product_health

Re-reads a project's errors, feedback and rollouts from PostHog and Sentry, or every active project's.

import_meetings

Imports up to three meetings from a project's note-takers, as the meeting picker does. Importing never starts processing on its own.

process_meetings

Starts a spec sheet from one meeting, or one combined spec sheet from up to three, following the project's context settings.

Counts as a spec sheet and toward spec sheet AI tokens.

approve_meeting_context

Lets a meeting waiting on its gathered context go on to its spec sheet, for projects that review context first.

Counts as a spec sheet and toward spec sheet AI tokens.

create_spec_sheet_from_brief

Turns a written brief into a spec sheet and its stories. This is how stories are created directly.

Counts as a spec sheet and toward spec sheet AI tokens.

generate_spec_sheet_from_requirement

Starts a spec sheet from a requirement's latest tagged version, or its draft when nothing is tagged.

Counts as a spec sheet and toward spec sheet AI tokens.

retry_spec_sheet

Re-runs a failed spec sheet made from a Brief or a requirement, in place.

Counts as a spec sheet and toward spec sheet AI tokens.

create_requirement

Saves text word for word as a new requirement, with no AI, as a plain paste does.

generate_requirement_from_text

Writes a new requirement from text and an instruction, as "Enhance with AI" does.

Counts as a requirement revision and toward requirement AI tokens.

import_requirement_from_notion

Turns a whole Notion page tree, under one of the project's selected pages, into one requirement.

Counts as a requirement revision and toward requirement AI tokens.

generate_requirement_from_meetings

Writes a new requirement from one to three of the project's meetings, with their gathered context.

Counts as a requirement revision and toward requirement AI tokens.

update_requirement_from_meetings

Revises a requirement from one to three meetings into a new draft. Tagged versions are kept.

Counts as a requirement revision and toward requirement AI tokens.

rewrite_requirement_with_ai

Rewrites a requirement's draft following an instruction, as "Update with AI" does. Tagged versions are kept.

Counts as a requirement revision and toward requirement AI tokens.

retry_requirement

Re-runs a failed first generation from its files, Notion page or pasted text.

Counts as a requirement revision and toward requirement AI tokens.

update_story

Changes a story's title, description, acceptance criteria, type, priority, estimate, labels or assignees.

set_story_status

Moves stories between Pending review, Approved and Rejected, as the board does.

publish_story

Creates the story in the project's Linear, Shortcut or Jira. Once published it stays published.

regenerate_task_stories

Writes a spec sheet task's stories again with AI.

refresh_budget

Re-syncs a project's Harvest budget and each person's logged hours, or every active project's. Refresh its stories first so the projection is current.

Admins and owners, as on the Budget page.

update_project_settings

Changes a project's name, description, keywords, generation settings and budget inputs, as Project Settings does.

set_time_classification_rules

Sets the rules that mark a project's Harvest time entries as meeting or miscellaneous time, then re-syncs its logged hours, as the budget settings do.

Admins and owners, as in the budget settings.

create_workstream

Groups stories of one project into a new workstream on the Development board. Stories still in review are approved as they join, as Start development does in the app.

update_workstream

Changes a workstream's title or who it is assigned to.

move_workstream

Moves a workstream to another state on the Development board, or to a place in its lane. Archiving is final.

add_workstream_stories

Adds stories of the workstream's project to it. Stories still in review are approved as they join.

remove_workstream_story

Takes a story out of a workstream. A story already sent to a coding agent stays in it.

start_agent_run

Creates a Conductor workspace for a workstream with your own connected Conductor account, and sends it the stories' details as one message.

Uses your own Conductor account and its compute, not the workspace's allowances.

link_agent_run

Links a Conductor workspace that already exists to a workstream, optionally sending it stories.

Uses your own Conductor account.

send_stories_to_agent_run

Sends more of a workstream's stories to the chat of a run you started.

Only the person whose account the run uses.

unlink_agent_run

Ends a run in Spectr. Nothing changes in Conductor, and the stories sent to it stay in the workstream.

The person whose account the run uses, or a workspace admin or owner on any run.

get_feature_guide

What each part of Spectr is for, who can use it, and how workspace roles work.

get_integration_guide

What each integration does for a project and the steps to connect it.