[AZ-836] Defer In Testing transition: CallMcpTool unavailable

The harness's MCP shim stopped accepting CallMcpTool mid-/autodev,
so the In Testing transition after batch 106 could not fire. Two
earlier MCP calls in the same turn succeeded (To Do -> In Progress
on AZ-836), so Jira itself is reachable; the shim is the problem.

Recorded under _docs/_process_leftovers/ with full replay payload
(transition id 32) per .cursor/rules/tracker.mdc. Will replay on
next /autodev Bootstrap step B1.

Updated _docs/_autodev_state.md sub_step.detail to point at the
leftover so the resumer doesn't lose track.

Co-authored-by: Cursor <cursoragent@cursor.com>

_docs/_autodev_state.md

@@ -8,7 +8,7 @@ status: in_progress
 sub_step:
   phase: 7
   name: batch-loop
-  detail: "AZ-836 landed (batch 106); AZ-838 next — start in fresh session"
+  detail: "AZ-836 batch 106 committed; In Testing transition deferred (leftover 2026-05-22 az836); AZ-838 next"
 retry_count: 0
 cycle: 3
 tracker: jira

_docs/_process_leftovers/2026-05-22_az836_in_testing_transition_deferred.md

@@ -0,0 +1,71 @@
+# AZ-836 In Testing transition deferred — CallMcpTool unavailable
+
+**Recorded**: 2026-05-22T17:50+03:00 (Europe/Kyiv)
+**Status**: deferred-non-user (replay when CallMcpTool returns)
+**Last replay attempt**: 2026-05-22T17:50+03:00 — `CallMcpTool` returned
+`Tool not found: CallMcpTool. Available tools: Shell, Glob, Grep, ...`
+twice in a row. The In Progress transition earlier in the same
+`/autodev` turn succeeded; the In Testing transition now cannot fire.
+
+## What is blocked
+
+Transition AZ-836 from **In Progress** → **In Testing** in Jira
+(`denyspopov.atlassian.net`, project AZ).
+
+Required by the implement skill's Step 12 (Update Tracker Status →
+In Testing) after a batch commits. Implementation commit landed
+locally as `5e52779` ([AZ-836] TlogRouteExtractor: tlog -> RouteSpec
+for Epic AZ-835 C1) — the code is done; only the Jira status move
+remains.
+
+## Payload (to be replayed when unblocked)
+
+Tool call:
+
+```
+CallMcpTool(
+  server="user-atlassian-mcp",
+  toolName="transitionJiraIssue",
+  arguments={
+    "cloudId": "denyspopov.atlassian.net",
+    "issueIdOrKey": "AZ-836",
+    "transition": {"id": "32"}  # workflow-confirmed via
+                                # getTransitionsForJiraIssue at
+                                # 2026-05-22T17:35: id=32 = "In Testing"
+  }
+)
+```
+
+Post-replay read-back: `getJiraIssue` on AZ-836 → expect
+`fields.status.name == "In Testing"`.
+
+## Why
+
+The harness's MCP tool calling shim (`CallMcpTool`) became
+unavailable mid-session. Two earlier MCP calls in the same
+`/autodev` turn succeeded:
+
+- `getTransitionsForJiraIssue` on AZ-836 — returned the workflow
+  transition table (id 21 = In Progress, id 32 = In Testing).
+- `transitionJiraIssue` on AZ-836 with `id=21` — moved To Do →
+  In Progress; Jira read-back confirmed `status.name == "In Progress"`.
+
+The third call (`transitionJiraIssue` with `id=32`) returns
+`Tool not found: CallMcpTool` from the harness, not from Jira. Jira
+itself is reachable.
+
+## Replay procedure
+
+1. Verify `CallMcpTool` is available again.
+2. Replay the payload above.
+3. Read back AZ-836 and confirm `status.name == "In Testing"`.
+4. Delete this leftover.
+
+If the harness shim is still missing on next `/autodev` invocation,
+re-record the failure here (update the `Last replay attempt`
+timestamp) and surface to the user.
+
+## Owner
+
+Autodev orchestrator — replay on next `/autodev` Bootstrap step B1
+(Process leftovers).