@myspec/mcp-server 1.1.1-next.175 → 1.1.1-next.176

package/README.md

@@ -363,9 +363,9 @@ Local state is stored under `~/.myspec/` (alongside the on-disk cache root), spl
 | `reply_spec_file_comment_thread` | `file_id`, `thread_id`, `body` | Replies to an existing thread. Conflicts on a trashed file |
 | `update_spec_file_comment` | `file_id`, `comment_id`, `body`, optional `expected_version` | Edits the body of one of YOUR OWN comments; the platform refuses somebody else's. `expected_version` is the **comment's** own `content_version`, not the file's — pairing it with the file would fail an edit because somebody saved the document. Works on a trashed file |
 | `delete_spec_file_comment` | `file_id`, `comment_id` | Soft-deletes one of YOUR OWN comments. Three outcomes, reported in the result: a reply disappears (`comment: null`); a first comment with live replies becomes a tombstone so the thread keeps a head (`comment` is the tombstone); a first comment with no live replies takes the whole thread (`thread_deleted: true`) |
-| `list_tasks` | `project_id`, optional `status`, `assignee_user_id` (a user id, `me` or `none`), `cycle`, `label`, `file_id`, `include_details`, `limit`, `offset` | The project's task board: every checkbox task in its tasks.md files, in file order, each with a stable `task_id`, `status` (`pending`/`in_progress`/`done`), assignee, `priority` (0 none … 4 low, plus `priority_name`), `story_points`, `cycle`, `start_date`/`end_date` (calendar days, `YYYY-MM-DD`), `labels`, and with `include_details` a 500-character preview of its details. Also the files (with the `content_version` their tasks reflect) and the labels and cycles in use. Filtered and paged by the platform: 100 tasks by default (`limit` up to 500), with `total_tasks` and, while more remain, `next_offset` |
+| `list_tasks` | `project_id`, optional `status`, `assignee_user_id` (a user id, `me` or `none`), `cycle`, `label`, `file_id`, `include_details`, `limit`, `offset` | The project's task board: every checkbox task in its tasks.md files, in file order, each with a stable `task_id`, `status` (`pending`/`in_progress`/`done`), assignee, `priority` (0 none … 4 low, plus `priority_name`), `story_points`, `cycle`, `start_date`/`end_date` (calendar days, `YYYY-MM-DD`), `labels`, `relationships` (`{type, task_id}`: this task `blocks`, is `blocked_by`, or is `related` to that task), and with `include_details` a 500-character preview of its details. Also the files (with the `content_version` their tasks reflect) and the labels and cycles in use. Filtered and paged by the platform: 100 tasks by default (`limit` up to 500), with `total_tasks` and, while more remain, `next_offset` |
 | `list_task_assignees` | `project_id` | Who the project's tasks can be assigned to: the org's people who can open the project (exactly who `update_tasks` accepts), each with `user_id`, `username` and `display_name`, plus your own `self_user_id`. May include people who left the org but whose profile remains; at most 1000 (`truncated`) |
-| `update_tasks` | `project_id`, `task_ids[]` (≤ 200), any of `assignee_user_id`, `priority`, `story_points`, `cycle`, `start_date`, `end_date`, `add_labels[]`, `remove_labels[]` | Edits planning fields on every task given. A field left out is left alone; `null` clears it. The assignee must be an org member who can open the project (`list_task_assignees`); `me` means you. All or nothing: an unknown id fails the call. Returns the tasks as they are now |
+| `update_tasks` | `project_id`, `task_ids[]` (≤ 200), any of `assignee_user_id`, `priority`, `story_points`, `cycle`, `start_date`, `end_date`, `add_labels[]`, `remove_labels[]`, `add_relationships[]` (`{type, task_id}`, ≤ 50), `remove_relationships[]` (task ids, ≤ 50) | Edits planning fields on every task given. A field left out is left alone; `null` clears it. Two tasks have at most one relationship: setting one replaces the pair's old one, and the other task sees the inverse. The assignee must be an org member who can open the project (`list_task_assignees`); `me` means you. All or nothing: an unknown id fails the call. Returns the tasks as they are now |
 | `set_task_status` | `project_id`, `task_ids[]` (≤ 200), `status`, optional `expected_versions[]` (`{spec_file_id, content_version}` from `list_tasks`) | Rewrites the tasks' checkboxes in tasks.md (`[ ]`, `[-]`, `[x]`), one revision per file, recorded as an AI change. A file that changed since the pinned version is not written (`file_changed`). Partial: returns `updated`/`unchanged`/`removed`/`failed` counts, an outcome per task (with an error code such as `conflict` or `quota_exceeded`) and per file, and the tasks as they are now; an error result only when every task failed |
 
 There is deliberately **no tool to resolve, unresolve or re-anchor a thread**. Those three are human-only: resolving declares a review finished, and re-anchoring is the repair a person makes after the automatic pass has already failed to locate the passage. The platform refuses them for an AI caller with a 403, so a tool would only ever return an error.

package/dist/index.js

@@ -2390,12 +2390,15 @@ var SPEC_FILE_LOCK_TTL_MS = 2 * 6e4;
 // ../../packages/platform-client/project/types/task.types.ts
 var TASK_STATUSES = ["pending", "in_progress", "done"];
 var TASK_PRIORITIES = [0, 1, 2, 3, 4];
+var TASK_RELATIONSHIP_TYPES = ["blocks", "blocked_by", "related"];
 var TASK_LIMITS = {
   maxLabels: 20,
   maxLabelLength: 40,
   maxStoryPoints: 9999.99,
   maxCycle: 1e5,
   maxTasksPerRequest: 200,
+  /** The most relationships one task has, and the most one change adds (or removes). */
+  maxRelationships: 50,
   /** The most tasks one page of a listing holds. */
   maxPageSize: 500,
   /** The most tasks.md files one board lists (and the most expected_versions a status change takes). */
@@ -2414,6 +2417,11 @@ function toStatus(raw) {
 function toPriority(raw) {
   return TASK_PRIORITIES.includes(raw) ? raw : 0;
 }
+function toRelationships(raw) {
+  return (raw ?? []).flatMap(
+    (r) => TASK_RELATIONSHIP_TYPES.includes(r.type) ? [{ type: r.type, taskId: r.task_id }] : []
+  );
+}
 function toOutcome(raw) {
   return raw === "updated" || raw === "unchanged" || raw === "removed" ? raw : "failed";
 }
@@ -2440,6 +2448,7 @@ function transformProjectTask(raw) {
     startAt: parseTimestamp(raw.start_at),
     endAt: parseTimestamp(raw.end_at),
     labels: raw.labels ?? [],
+    relationships: toRelationships(raw.relationships),
     updatedBy: raw.updated_by ?? null,
     createdAt: parseTimestamp(raw.created_at),
     updatedAt: parseTimestamp(raw.updated_at)
@@ -2527,8 +2536,15 @@ function toRawUpdateTasksRequest(taskIds, patch) {
   const body = { task_ids: taskIds, set: toRawPlanningSet(patch) };
   if (patch.addLabels?.length) body.add_labels = patch.addLabels;
   if (patch.removeLabels?.length) body.remove_labels = patch.removeLabels;
+  addRawRelationshipEdits(body, patch);
   return body;
 }
+function addRawRelationshipEdits(body, patch) {
+  if (patch.addRelationships?.length) {
+    body.add_relationships = patch.addRelationships.map((r) => ({ type: r.type, task_id: r.taskId }));
+  }
+  if (patch.removeRelationships?.length) body.remove_relationships = patch.removeRelationships;
+}
 function toRawPlanningSet(patch) {
   const set = {};
   if ("assigneeUserId" in patch) set.assignee_user_id = patch.assigneeUserId ?? null;
@@ -2548,6 +2564,7 @@ function toRawTaskChange(change) {
     if (Object.keys(set).length > 0) raw.set = set;
     if (patch.addLabels?.length) raw.add_labels = patch.addLabels;
     if (patch.removeLabels?.length) raw.remove_labels = patch.removeLabels;
+    addRawRelationshipEdits(raw, patch);
   }
   return raw;
 }
@@ -7532,6 +7549,7 @@ var ME = "me";
 var NONE = "none";
 var PRIORITY_NAMES = ["none", "urgent", "high", "medium", "low"];
 var STATUS = z39.enum(["pending", "in_progress", "done"]);
+var RELATIONSHIP_TYPE = z39.enum(["blocks", "blocked_by", "related"]);
 function toDay(date) {
   return date ? date.toISOString().slice(0, 10) : null;
 }
@@ -7558,6 +7576,8 @@ function taskSummary(task, opts = {}) {
     start_date: toDay(task.startAt),
     end_date: toDay(task.endAt),
     labels: task.labels,
+    // "this task <type> task_id": blocks, blocked_by or related.
+    relationships: task.relationships.map((r) => ({ type: r.type, task_id: r.taskId })),
     ...withDetails ? { details: truncated ? task.body.slice(0, DETAILS_PREVIEW_CHARS) : task.body } : {},
     // Only ever present and true.
     ...truncated ? { details_truncated: true } : {},
@@ -7610,7 +7630,7 @@ function pageLocally(board, query) {
 }
 var listTasksTool = {
   name: "list_tasks",
-  description: `List a project's tasks: every checkbox task in its tasks.md files, in file order, each with a stable task_id, status (pending, in_progress, done), assignee, priority (0 none, 1 urgent, 2 high, 3 medium, 4 low), story points, cycle, start/end dates (YYYY-MM-DD) and labels. Also returns the files (with the content_version their tasks reflect), and the labels and cycles in use. Filter by status, assignee ("me" for yourself, "none" for unassigned), cycle, label or file. Paged: 100 tasks by default (limit up to 500), with total_tasks and, while more remain, next_offset. Details (sub-bullets) only with include_details. Use the task_id values with update_tasks and set_task_status.`,
+  description: `List a project's tasks: every checkbox task in its tasks.md files, in file order, each with a stable task_id, status (pending, in_progress, done), assignee, priority (0 none, 1 urgent, 2 high, 3 medium, 4 low), story points, cycle, start/end dates (YYYY-MM-DD), labels and relationships with other tasks ({type, task_id}: this task blocks, is blocked_by, or is related to that task). Also returns the files (with the content_version their tasks reflect), and the labels and cycles in use. Filter by status, assignee ("me" for yourself, "none" for unassigned), cycle, label or file. Paged: 100 tasks by default (limit up to 500), with total_tasks and, while more remain, next_offset. Details (sub-bullets) only with include_details. Use the task_id values with update_tasks and set_task_status.`,
   inputSchema: listInputSchema2,
   handler: async (args, ctx) => {
     let assignee;
@@ -7690,11 +7710,22 @@ var updateInputSchema2 = {
   start_date: z39.iso.date().nullable().optional().describe("Start day as YYYY-MM-DD, or null to clear it."),
   end_date: z39.iso.date().nullable().optional().describe("Due day as YYYY-MM-DD, or null to clear it. Not before the start day."),
   add_labels: z39.array(labelField).max(TASK_LIMITS.maxLabels).optional().describe(`Labels to add. A task carries at most ${String(TASK_LIMITS.maxLabels)}.`),
-  remove_labels: z39.array(labelField).max(TASK_LIMITS.maxLabels).optional().describe("Labels to remove.")
+  remove_labels: z39.array(labelField).max(TASK_LIMITS.maxLabels).optional().describe("Labels to remove."),
+  add_relationships: z39.array(
+    z39.object({
+      type: RELATIONSHIP_TYPE.describe(
+        "How each changed task stands to task_id: blocks (task_id waits for it), blocked_by (it waits for task_id) or related."
+      ),
+      task_id: z39.uuid().describe("The other task (a task_id from list_tasks).")
+    })
+  ).max(TASK_LIMITS.maxRelationships).optional().describe(
+    `Relationships to set. Two tasks have at most one: setting one replaces whatever the pair had, and the other task sees the inverse (A blocks B is B blocked_by A). A task has at most ${String(TASK_LIMITS.maxRelationships)}.`
+  ),
+  remove_relationships: z39.array(z39.uuid()).max(TASK_LIMITS.maxRelationships).optional().describe("Task ids to unlink each changed task from, whatever the relationship.")
 };
 var updateTasksTool = {
   name: "update_tasks",
-  description: "Edit the planning fields of one or more tasks: assignee, priority, story points, cycle, start and due dates, and labels. The same change applies to every task given. A field left out is left alone; null clears it. Labels are added and removed, not replaced. Does not change status or the task's text, which live in tasks.md: use set_task_status for status. All or nothing: an unknown task id fails the whole call. Returns the tasks as they are now.",
+  description: "Edit the planning fields of one or more tasks: assignee, priority, story points, cycle, start and due dates, labels, and relationships with other tasks (blocks, blocked_by, related). The same change applies to every task given. A field left out is left alone; null clears it. Labels and relationships are added and removed, not replaced. Does not change status or the task's text, which live in tasks.md: use set_task_status for status. All or nothing: an unknown task id fails the whole call. Returns the tasks as they are now.",
   inputSchema: updateInputSchema2,
   handler: async (args, ctx) => {
     const patch = {};
@@ -7716,9 +7747,16 @@ var updateTasksTool = {
     }
     if (args.add_labels?.length) patch.addLabels = args.add_labels;
     if (args.remove_labels?.length) patch.removeLabels = args.remove_labels;
+    if (args.add_relationships?.length) {
+      patch.addRelationships = args.add_relationships.map((r) => ({
+        type: r.type,
+        taskId: r.task_id
+      }));
+    }
+    if (args.remove_relationships?.length) patch.removeRelationships = args.remove_relationships;
     if (Object.keys(patch).length === 0) {
       return errorResult(
-        "Nothing to change: pass at least one of assignee_user_id, priority, story_points, cycle, start_date, end_date, add_labels or remove_labels."
+        "Nothing to change: pass at least one of assignee_user_id, priority, story_points, cycle, start_date, end_date, add_labels, remove_labels, add_relationships or remove_relationships."
       );
     }
     const tasks = await ctx.client.updateProjectTasks(args.project_id, args.task_ids, patch);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@myspec/mcp-server",
-  "version": "1.1.1-next.175",
+  "version": "1.1.1-next.176",
   "description": "MySpec MCP server — exposes MySpec platform projects, files and attachments to MCP-aware clients via OAuth-authenticated access tokens.",
   "type": "module",
   "repository": {