overlord-cli 0.2610051433.0 → 0.2610051609.0

package/dist/connectors/VERSION

@@ -1 +1 @@
-0.3.49
+0.3.50

package/dist/connectors/adapters/antigravity/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://antigravity.google/schemas/v1/plugin.json",
   "name": "overlord",
-  "version": "0.3.49",
+  "version": "0.3.50",
   "description": "Chat-driven access to Overlord mission protocol workflows through skills and MCP.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/claude/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "overlord",
-  "version": "0.3.49",
+  "version": "0.3.50",
   "description": "Overlord mission protocol workflow for Claude Code — attach/update/ask/deliver, slash commands, and hook-driven activity capture.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/codex/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "overlord",
-  "version": "0.3.49",
+  "version": "0.3.50",
   "description": "Chat-driven access to Overlord mission protocol workflows with hook-driven activity capture.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/cursor/.cursor-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "overlord",
-  "version": "0.3.49",
+  "version": "0.3.50",
   "description": "Overlord mission protocol workflow for Cursor — attach/update/ask/deliver, slash commands, MCP bridge, and follow-up capture.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/core/scripts/overlord-mcp.mjs

@@ -210,9 +210,15 @@ const tools = [
     name: 'overlord_search_missions',
     title: 'Search Overlord missions',
     description:
-      "Search mission anchors with their matching objectives and deliveries across authorized workspaces. An objective displayId can be passed directly to overlord_load_mission_context; matches are matched-only, not the full objective list. Artifacts are not indexed. Use absolute ISO from/to bounds (no relative dates or implicit window), inspect workspaceCounts, entityCounts, and truncatedCandidates before claiming completeness, and treat appliedFilters.mode 'fallback' as a recency listing. Compact detail is default; request full for child snippets and metadata.",
+      "Search mission anchors with their matching objectives and deliveries across authorized workspaces. An objective displayId can be passed directly to overlord_load_mission_context; matches are matched-only, not the full objective list. Artifacts are not indexed. Use absolute ISO from/to bounds (no relative dates or implicit window), inspect workspaceCounts, entityCounts, and truncatedCandidates before claiming completeness, and treat appliedFilters.mode 'fallback' as a recency listing. Compact detail is default; request full for child snippets and metadata. Ranked results never prove absence: to check an exact token such as a canonical kb-feature: Knowledgebase Feature reference, pass reference with projectId (only workspaceId, limit 1-100 and cursor may accompany it) and follow nextCursor until complete is true.",
     inputSchema: objectSchema({
       query: stringProperty('Search query text.'),
+      reference: stringProperty(
+        'Exact-reference mode: case-sensitive whole token (8-512 characters, no whitespace) in live objective text. Requires projectId; cannot be combined with ranked filters. Returns { results (with statusType and matching objectives), nextCursor, complete }.'
+      ),
+      cursor: stringProperty(
+        'Exact-reference mode only: nextCursor from the previous page for the same reference and projectId.'
+      ),
       status: stringProperty(
         'Comma-separated status TYPES, such as draft,next,execute,review. Types are workspace-invariant (draft, next, execute, review, complete, blocked, cancelled). Project-defined status names are not accepted here.'
       ),
@@ -829,6 +835,43 @@ async function callOverlordTool(name, args) {
     });
   }
   if (name === 'overlord_search_missions') {
+    const reference = optionalString(args, 'reference');
+    if (reference) {
+      // Exact-reference mode: ranked filters are forwarded so Protocol rejects them.
+      const ranked = {};
+      for (const [key, flag] of [
+        ['query', 'query'],
+        ['status', 'status'],
+        ['resourceKey', 'resource-key'],
+        ['dateField', 'date-field'],
+        ['from', 'from'],
+        ['to', 'to'],
+        ['entityTypes', 'entity-types'],
+        ['objectiveStates', 'objective-states']
+      ]) {
+        if (optionalString(args, key)) ranked[flag] = requiredString(args, key);
+      }
+      if (typeof args.matchesPerResult === 'number') {
+        ranked['matches-per-result'] = String(args.matchesPerResult);
+      }
+      return runProtocol('search', {
+        reference,
+        ...(optionalString(args, 'projectId')
+          ? { 'project-id': requiredString(args, 'projectId') }
+          : {}),
+        ...(optionalString(args, 'workspaceId')
+          ? { 'workspace-id': requiredString(args, 'workspaceId') }
+          : {}),
+        ...(optionalString(args, 'cursor') ? { cursor: requiredString(args, 'cursor') } : {}),
+        ...(typeof args.limit === 'number' && Number.isFinite(args.limit)
+          ? { limit: String(Math.trunc(args.limit)) }
+          : {}),
+        ...ranked
+      });
+    }
+    if (optionalString(args, 'cursor')) {
+      throw new Error('cursor applies only to an exact reference search');
+    }
     const detail = optionalString(args, 'detail') || 'compact';
     if (detail !== 'compact' && detail !== 'full') {
       throw new Error("detail must be 'compact' or 'full'");
@@ -1228,7 +1271,7 @@ process.stdin.on('data', async chunk => {
         result: {
           protocolVersion: PROTOCOL_VERSION,
           capabilities: { tools: { listChanged: false } },
-          serverInfo: { name: 'overlord-__OVERLORD_ADAPTER_KEY__', version: '0.3.49' }
+          serverInfo: { name: 'overlord-__OVERLORD_ADAPTER_KEY__', version: '0.3.50' }
         }
       });
       continue;

package/dist/index.js

@@ -8065,6 +8065,13 @@ var init_launch_variables = __esm({
   }
 });
 
+// ../packages/contract/dist/mission-reference.js
+var init_mission_reference = __esm({
+  "../packages/contract/dist/mission-reference.js"() {
+    "use strict";
+  }
+});
+
 // ../packages/contract/dist/mission-search.js
 var init_mission_search = __esm({
   "../packages/contract/dist/mission-search.js"() {
@@ -8185,6 +8192,7 @@ var init_dist2 = __esm({
     init_agent_launch_flags();
     init_chat();
     init_launch_variables();
+    init_mission_reference();
     init_mission_search();
     init_notifications();
     init_objective_ref();
@@ -39591,8 +39599,16 @@ search:
     --entity-types <csv>        V3 only: mission,objective,delivery (default all)
     --objective-states <csv>    V3 only: restrict matching objective states
     --matches-per-result <n>    V3 only: child matches per mission (default 3, max 10)
+    --reference <token>         Exact-reference mode instead of ranked search: every
+                                mission in --project-id (required) whose live objective
+                                text contains this case-sensitive whole token, e.g. a
+                                kb-feature: Knowledgebase Feature reference. Accepts only
+                                --workspace-id, --limit (1-100, default 50) and --cursor.
+    --cursor <cursor>           Exact-reference mode: nextCursor from the previous page
   Returns:
     JSON with matching missions (v1) or the versioned search envelope (v2/v3).
+    With --reference: { results, nextCursor, complete }. Only a page with
+    complete: true ends the scan; ranked search never proves absence.
 
 statuses:
   Purpose: