task_search
Search tasks in a project (lightweight summaries). Always pass projectId. Every successful response includes delegation (your human supervisor, advisory responsibility role, and credential-authorized capabilities) plus projectPolicy; follow both. Responsibility roles never filter results. For multi-workspace OAuth connections, pass workspaceId from statelode://workspace to target a non-default workspace; ungranted ids are rejected. Never guess a projectId: searching never creates anything, and an unknown projectId returns an unknown_project error listing the selected workspace's real slugs (knownProjects) — pick one of those and retry. Every response also includes workflow: this project's statuses as { key, name, category, hidden } — projects define their own columns, so use a key (stable) or a category for status, never a display name. With NO status, results are limited to 'active' tasks only — so an empty result does NOT prove a task is absent; pass an explicit status (or an array, e.g. backlog+active+in_progress+review+done+blocked) before concluding nothing matches. Default limit is 10 (max 50). The response echoes appliedStatus and appliedLimit, and sets hasMore:true when the limit truncated the results. Optional sprint filter: a sprint id, 'active' (the project's active sprint), or 'none' (unsprinted). Use this FIRST; only call task_get when you need a full body.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | ||
| owner | No | ||
| query | No | ||
| sprint | No | ||
| status | No | ||
| worker | No | ||
| assignee | No | ||
| priority | No | ||
| requests | No | ||
| projectId | Yes | ||
| workspaceId | No | A workspace id from statelode://workspace. Explicit input overrides every default. For repository work, pass the workspaceId pinned in statelode.json. Project-creating mutations require this when a multi-workspace grant has no human-selected default. |