@yemi33/minions 0.1.2454 → 0.1.2455

package/engine.js

@@ -11706,7 +11706,7 @@ function discoverFromWorkItems(config, project) {
     newWork.push({
       type: workType,
       agent: agentId,
-      agentName: config.agents[agentId]?.name || tempAgents.get(agentId)?.name || agentId,
+      agentName: config.agents[agentId]?.name || tempAgents.get(agentId)?.name || shared.tempAgentCallSign(agentId) || agentId,
       agentRole: config.agents[agentId]?.role || tempAgents.get(agentId)?.role || 'Agent',
       task: `[${project?.name || 'project'}] ${item.title || item.description?.slice(0, 80) || item.id}`,
       prompt,
@@ -12952,7 +12952,7 @@ function resolvePendingDispatchAgent(item, config) {
 function assignPendingDispatchAgent(item, agentId, config) {
   const agents = config.agents || {};
   item.agent = agentId;
-  item.agentName = agents[agentId]?.name || tempAgents.get(agentId)?.name || agentId;
+  item.agentName = agents[agentId]?.name || tempAgents.get(agentId)?.name || shared.tempAgentCallSign(agentId) || agentId;
   item.agentRole = agents[agentId]?.role || tempAgents.get(agentId)?.role || 'Agent';
   delete item._agentBusySince;
   delete item.skipReason;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yemi33/minions",
-  "version": "0.1.2454",
+  "version": "0.1.2455",
   "description": "Multi-agent AI dev team that runs from ~/.minions/ — five autonomous agents share a single engine, dashboard, and knowledge base",
   "bin": {
     "minions": "bin/minions.js"

package/playbooks/fix.md

@@ -232,8 +232,29 @@ Pick the disposition that matches the finding and post the matching reply, then
 - **Won't fix / disagree** (comment is invalid, stale, already addressed, out of scope, or harmful) → Reply with the evidence-backed rationale explaining why no code change was made. Leave the thread **active** for the reviewer to decide; do NOT resolve unilaterally.
 
 Host-specific mechanics:
-- **GitHub**: Reply to each review comment, then resolve the conversation for the four resolve-eligible dispositions above. Leave Won't-fix threads unresolved.
-- **ADO**: Use `az` CLI first to reply to each thread and update status when supported; use ADO MCP only as a fallback when `az` is unavailable or insufficient. Set status to `fixed` for Fixed/Satisfied-elsewhere/Tracked-as-follow-up and `closed` for Deferred; leave `active` for Won't-fix rationale replies.
+- **Preferred (both hosts): reply on the ORIGINATING thread and resolve it in one call** with
+  `minions pr comment <repo> --thread <threadId> --resolve --agent <id> --kind fix [--wi <finding-id>] --body-file <path>`
+  (ADO: add `--host ado --ado-org … --ado-project … --repo-id …` and, for Deferred, `--resolve-status closed`;
+  GitHub: `--thread` is the review-thread GraphQL node id). The `--thread` id is the **structured thread/finding
+  identity carried on the work item / your `reviewFindingResolution` evidence** — never resolve by matching title or
+  body text, and never resolve sibling threads or all threads in bulk. This path is idempotent: re-running after a
+  retry, restart, or a provider timeout re-checks the thread, skips a reply it already posted (matched by the hidden
+  minions marker), and re-attempts only the unfinished resolve.
+- **Do NOT open a fresh active thread to summarize an addressed finding.** A new top-level/active thread for a fix
+  you already made just adds triage noise the author must re-close. Post the summary as a reply on the finding's own
+  thread via `--thread … --resolve` above.
+- **GitHub**: `--thread … --resolve` replies to the review thread and resolves it via `resolveReviewThread` for the
+  four resolve-eligible dispositions. Leave Won't-fix threads unresolved. If no review-thread id exists (the finding
+  was a top-level conversation comment), a top-level `minions pr comment` reply is informational only — GitHub has no
+  resolved state for conversation comments, so do **not** claim it is "resolved."
+- **ADO**: `--thread … --resolve` posts the reply on the originating thread and transitions it (`fixed` for
+  Fixed/Satisfied-elsewhere/Tracked-as-follow-up, `--resolve-status closed` for Deferred). Prefer it over raw `az`;
+  if you must fall back, use `az` first (ADO MCP only when `az` is insufficient), reply on the existing thread, and
+  set its status — do not create a new thread just for the fix summary. Leave `active` for Won't-fix rationale replies.
+
+Never auto-resolve a finding the agent did NOT actually address. When the fix report says the finding was not
+addressed, is ambiguous, needs human input, or lacks the required commit/`file:line` resolution evidence, reply with
+the rationale and leave the thread **active**. Human feedback is not auto-resolved merely because a dispatch completed.
 
 Hard rule: never resolve, close, or mark-fixed a thread without first posting a reply that explains the disposition. The chunk-5 / Caleb-Tseng / MaiLibraryViewModel thread (ADO 5215549 thread 65692221) is the canonical bad example — closed silently with no audit trail.
 

package/playbooks/shared-rules.md

@@ -262,6 +262,18 @@ non-actionable thread is posted pre-resolved in a single call. Prefer
 thread `status` field yourself: `closed` (4) for non-actionable notes (or `fixed`
 (2) when the comment records something already handled), `active` (1) for
 actionable findings. The verdict/summary comment itself stays `active`.
+
+**Fix summaries reply on the originating thread — never a new active thread.**
+When you are recording that an existing review finding was *addressed*, reply on
+that finding's own thread and transition it, rather than opening a fresh thread:
+`minions pr comment <pr> --host ado --ado-org … --ado-project … --repo-id … --thread <threadId> --resolve [--resolve-status closed] --agent <id> --kind fix [--wi <finding-id>] --body-file <path>`.
+The `--thread` id is **structured thread/finding identity** from the work item or
+your `reviewFindingResolution` evidence — resolve only that thread, keyed by id,
+never by matching comment title/body text and never in bulk across sibling
+threads. The reply+status transition is idempotent across retries/restarts: it
+re-reads the thread, skips a reply already carrying your minions marker, and only
+re-applies the status if the earlier PATCH failed — so a partial failure stays
+visible and retryable instead of being silently marked resolved locally.
 {{/ado_shared_rules}}
 {{#github_shared_rules}}
 **GitHub mechanism.** GitHub has no "resolved" state for PR *conversation*
@@ -273,9 +285,19 @@ single API call that creates a thread pre-resolved. So on GitHub:
   `<details><summary>Non-blocking observations</summary>…</details>` block inside
   the verdict comment so they are visible but de-emphasized and create no separate
   unresolved threads. `--resolved` is rejected on the GitHub path for this reason.
+- **To record that an existing review finding was addressed**, reply on that
+  finding's own review thread and resolve it in one idempotent call:
+  `minions pr comment <repo> --thread <reviewThreadNodeId> --resolve --agent <id> --kind fix [--wi <finding-id>] --body-file <path>`.
+  The `--thread` node id is **structured thread identity** from the work item or
+  your `reviewFindingResolution` evidence — resolve only that thread, never by
+  matching comment text and never in bulk. Do **not** open a fresh active thread
+  to summarize a fix you already made. If the finding was a *top-level*
+  conversation comment (no review-thread node id), a top-level `minions pr comment`
+  reply is informational only — never call it "resolved."
 - If you DO post a non-actionable note as its own **inline review thread**, resolve
-  it immediately after posting via the GraphQL `resolveReviewThread` mutation
-  (issue comments cannot be resolved this way; only review threads can):
+  it immediately after posting (equivalent to `--thread … --resolve`, or the raw
+  GraphQL `resolveReviewThread` mutation below; issue comments cannot be resolved
+  this way, only review threads can):
   ```bash
   # 1. find the thread node id for the comment you just posted
   gh api graphql -f query='query($owner:String!,$repo:String!,$pr:Int!){