Orion

Trigger execution of a metric to refresh its value. Every execution adds a new run to the metric's history (get_metric_results). For table and view metrics that run is a fresh snapshot of the dataset. Args: metric_id: The metric ID to execute.

Get the content of a slide deck, dashboard, or report as markdown. Args: artifact_id: The artifact ID to retrieve.

Read messages from an existing Orion conversation. Args: conversation_id: The conversation ID to get history for. limit: Maximum number of messages to return (default 50).

Get full metric detail including definition, Python code, data sources, and latest result. When a result's value_data.storage is "gcs", the rows are not in the result; they live in storage. value_data holds a column profile (row count, per-column distinct and null counts, numeric min/max/mean), a short preview (preview.n rows, cells truncated at 256 bytes), and storage pointers this tool cannot follow. The profile has no sums. Read row count and column stats from it, and treat the preview as a sample: never sum or count it as if it were the full dataset. For computations over the full rows, use ask_orion. Older results without storage "gcs" carry their rows inline. Args: metric_id: The metric ID.

Get auto-detected metric insights (significant changes, anomalies) for a project. Args: project_id: The project ID. change_threshold: Minimum change percentage to flag (0.0-1.0, default 0.2 = 20%).

Get historical execution results for a metric, newest first. Each result is one run. A run whose value_data.storage is "gcs" carries a column profile and a short preview, not the full rows; see get_metric. Older runs may carry their rows inline. Args: metric_id: The metric ID. limit: Maximum number of results (default 20).

Read a wiki page's Markdown prose and metadata. For page_type=state, content does not include table rows. Use describe_state for structure and saved views, then query_state for records. write_state edits records; define_state_schema designs tables and cards. Args: page_id: The wiki page ID. Find one via search_wiki or list_wiki_pages.

Get a wiki page by its hierarchical path (e.g. /internal/icp). State page content is prose; describe_state and query_state read its tables. Args: path: The full hierarchical path to the page.

Get workflow detail including its definition (steps, outputs) and schedule. Args: workflow_id: The workflow ID.

Read a workflow report artifact's content. Text formats (markdown report, slide deck, dashboard) return their content as markdown. Binary formats (podcast audio) are not readable inline; instead this returns a short-lived signed download URL to give to the user. Args: artifact_id: A report artifact ID from get_workflow_run_reports.

List the report artifacts a workflow run produced. This is where every output of a run lives: markdown reports, slide decks, podcasts, and CSV exports each appear as one item with a format and an id. Pass an item's id to get_workflow_report_content to read it. When present, an item's share_url is its full public link; only share that exact URL, never assemble one yourself. Args: run_id: The workflow run ID.

List a workflow's runs, newest first. Each run has an id — pass it to get_workflow_run_reports to see the reports, slide decks, and podcasts that run produced. Args: workflow_id: The workflow ID. limit: Maximum number of runs to return (default 10).

List slides, dashboards, or reports for a project. When present, each item's share_url is the artifact's full public link. Only share that exact URL; never assemble one from the vanity_url slug. Args: project_id: The project ID to list artifacts for. artifact_type: Filter by type: 'html_slides', 'html_dashboard', or 'markdown_report'. limit: Maximum number of artifacts to return (default 20). offset: Number of artifacts to skip for pagination (default 0). visibility: Filter by visibility: 'user_board' or 'shared_board'. created_by: Filter by creator user ID (UUID format). has_workflow: Filter by workflow status: true for workflow, false for static.

List chat conversations for a user. Args: user_id: The user ID to list conversations for. project_id: Optional project ID to filter by. limit: Maximum number of conversations to return (default 20).

List metrics (KPIs) and views for a project. Each item's value_type is the declared type. number, percentage, currency, string and boolean are scalars; array and object are inline collections. table and view are datasets: a view is a large checkpoint meant as a base for further analysis, a table is the same shape at a size the Orion UI shows whole. graph is a promoted chart spec plus a reference to the rows it draws. A result can differ from the declared type: an over-budget table runs as a view, and a large array or object becomes tabular. The result's own value_type and coerced_from say so. Args: project_id: The project ID. pinned: If True, only return pinned metrics.

List the knowledge base pages enabled for a project. Args: project_id: The project ID.

List Orion projects the user has access to. Args: search: Optional search text to filter projects by name. limit: Maximum number of projects to return (default 20).

Get AI-generated recommendations for a project. Args: project_id: The project ID. limit: Maximum number of recommendations (default 10).

List state pages: knowledge base pages that hold structured tables (trackers, missions, decision logs, status boards). Match the user's words against table names too, not only titles: call describe_state on likely candidates. If an existing page fits, add rows there instead of creating a new page.

List all users in the Orion organization.

List wiki pages, folders and State workspaces in the knowledge base. State pages contain structured records as well as prose. Use list_state_pages for project-scoped discovery, then describe_state and query_state for their data. Args: parent_id: Filter to children of this parent page. page_type: Filter by 'document', 'folder' or 'state'. Use describe_state for State tables. search: Filter by text in title/content. limit: Maximum number of pages to return (default 50). offset: Number of pages to skip for pagination (default 0).

List workflows for a project (the workflow-builder pipelines that produce reports). Args: project_id: The project ID. search: Filter by text in the workflow name. limit: Maximum number of workflows to return (default 50).

Read a state page. Give exactly one of view, sql or table. Prefer a saved view when one answers the question: it is the page owner's definition. sql runs a read-only SQLite SELECT (joins, aggregates); writes are rejected, use write_state. table returns rows as {column: value} objects with a total count. Results over 200 rows or 64 KB come back with truncated true and a hint; narrow the query rather than paging blindly.

Trigger execution of a workflow. Returns the created run, including its id. Runs take minutes; poll get_workflow_run_reports with the run id to see outputs as they land. Args: workflow_id: The workflow ID to run.

Search the Orion wiki/knowledge base. Args: text: Search text. project_id: Optional project ID to scope the search. limit: Maximum number of results (default 20).

Rename a project or replace its description. Only the provided fields change. Requires analyst-level access to the project. Args: project_id: The project ID. name: New project name. description: New project description.

Update wiki prose and metadata. Only the provided fields are changed. For State pages this edits the introduction, not table data. Use write_state for records and define_state_schema for tables, saved views and cards. Edits to an unlocked page apply immediately with no review. On a locked page, only the page's creator edits directly; for anyone else the change is submitted as a change request for review. Args: page_id: The wiki page ID to update. title: New page title. content: New page content in markdown. summary: New short description. tags: New list of tags. edit_summary: Short description of the change, for revision history.