@projectsolo/solo-mission-mcp 0.21.13 → 0.21.14

package/DEVELOPER_README.md

@@ -99,7 +99,7 @@ but the package itself is `@projectsolo/solo-mission-mcp` (org-scoped) — insta
    ```
 2. **Register Trusted Publishing.** On https://www.npmjs.com, open
    `@projectsolo/solo-mission-mcp` → **Settings → Trusted Publisher → GitHub Actions**:
-   - Organization/repo: `ProjectSoloXYZ/solo_mission_mcp`
+   - Organization/repo: `ProjectSoloAI/solo_mission_mcp`
    - Workflow filename: `release.yml`
    - Environment: (leave blank)
 3. **Grant team publish access** so other org members aren't stuck relying on step 1's

package/dist/index.js

@@ -19,7 +19,7 @@ import {
 var RESPONSE_SCHEMA_PROPERTY = {
   type: "array",
   maxItems: 20,
-  description: 'Structured completion questions a human answers to complete the mission. Question ids must be unique within one schema \u2014 a duplicate id is rejected at creation/update, not merged or silently overwritten. When set, finalize_qualification auto-derives qualified UIDs from completion instead of requiring an explicit qualified_human_uids list \u2014 same mechanism media_review has always used for track ratings, generalized to any mission type. media_review defaults to a canonical two-question schema when this is omitted at creation \u2014 literally { id: "rating", kind: "stars", required: true } and { id: "comment", kind: "long", required: false } \u2014 so read/append against those exact ids if you rely on the default rather than supplying your own schema. Every other type has no default (stays fully manual/chat-based unless you set this).',
+  description: 'Structured completion questions a human answers to complete the mission. Question ids must be unique within one schema \u2014 a duplicate id is rejected at creation/update, not merged or silently overwritten. When set, finalize_qualification auto-derives qualified UIDs from completion instead of requiring an explicit qualified_human_uids list \u2014 same mechanism media_review has always used for track ratings, generalized to any mission type. media_review defaults to a canonical two-question schema when this is omitted at creation \u2014 literally { id: "rating", kind: "stars", required: true } and { id: "comment", kind: "long", required: true } \u2014 so read/append against those exact ids if you rely on the default rather than supplying your own schema. Every other type has no default (stays fully manual/chat-based unless you set this).',
   items: {
     type: "object",
     properties: {
@@ -43,7 +43,10 @@ var RESPONSE_SCHEMA_PROPERTY = {
       },
       label: { type: "string", description: "Question text shown to the human (max 200 chars). For the checkbox kind this is the checkbox's own clickable text, not a separate heading." },
       help: { type: "string", description: "Optional helper text shown under the label (max 500 chars)." },
-      required: { type: "boolean", description: "Whether this question must be answered to complete the mission." },
+      required: {
+        type: "boolean",
+        description: "Whether this question must be answered to complete the mission. A required: false question does NOT block mission completion or finalize_qualification's auto-qualification \u2014 both derive purely from the required questions, so an optional question left blank (or still being typed into) will not stop either from firing the instant the required ones are answered. If every question in the schema should actually be answered, mark them all required: true; reserve required: false for a genuinely optional/skippable field."
+      },
       options: {
         type: "array",
         items: { type: "string" },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@projectsolo/solo-mission-mcp",
-  "version": "0.21.13",
+  "version": "0.21.14",
   "description": "MCP server for Solo Mission Platform — lets AI agents create missions, browse humans, and chat.",
   "type": "module",
   "main": "dist/index.js",

package/src/tools/missions.ts

@@ -26,7 +26,7 @@ const RESPONSE_SCHEMA_PROPERTY = {
     'media_review has always used for track ratings, generalized to any mission type. ' +
     'media_review defaults to a canonical two-question schema when this is omitted at creation — ' +
     'literally { id: "rating", kind: "stars", required: true } and { id: "comment", kind: "long", ' +
-    'required: false } — so read/append against those exact ids if you rely on the default rather ' +
+    'required: true } — so read/append against those exact ids if you rely on the default rather ' +
     'than supplying your own schema. Every other type has no default (stays fully manual/chat-based ' +
     'unless you set this).',
   items: {
@@ -63,7 +63,16 @@ const RESPONSE_SCHEMA_PROPERTY = {
       },
       label: { type: 'string', description: 'Question text shown to the human (max 200 chars). For the checkbox kind this is the checkbox\'s own clickable text, not a separate heading.' },
       help: { type: 'string', description: 'Optional helper text shown under the label (max 500 chars).' },
-      required: { type: 'boolean', description: 'Whether this question must be answered to complete the mission.' },
+      required: {
+        type: 'boolean',
+        description:
+          'Whether this question must be answered to complete the mission. A required: false question ' +
+          'does NOT block mission completion or finalize_qualification\'s auto-qualification — both ' +
+          'derive purely from the required questions, so an optional question left blank (or still ' +
+          'being typed into) will not stop either from firing the instant the required ones are ' +
+          'answered. If every question in the schema should actually be answered, mark them all ' +
+          'required: true; reserve required: false for a genuinely optional/skippable field.',
+      },
       options: {
         type: 'array',
         items: { type: 'string' },