debug: probe the pinned action itself, dump execution-file result

Author: kyle-sexton

.github/workflows/debug-oauth-probe.yml

@@ -2,18 +2,31 @@ name: debug-oauth-probe
 on:
   push:
     branches: [debug/oauth-probe]
-permissions: {}
+permissions:
+  contents: read
+  id-token: write
 jobs:
   probe:
     runs-on: ubuntu-24.04
     timeout-minutes: 10
     steps:
-      - name: Probe Claude CLI auth from hosted runner
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+      - name: Run pinned claude-code-action with trivial prompt
+        id: cca
+        continue-on-error: true
+        uses: anthropics/claude-code-action@12531344451323133b0493233c759991ac61da12 # v1.0.174
+        with:
+          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+          prompt: "Reply with exactly: ACTION_PROBE_OK"
+          claude_args: --model claude-sonnet-4-6 --max-turns 2
+      - name: Dump SDK result detail
+        if: always()
         env:
-          CLAUDE_CODE_OAUTH_TOKEN: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+          EXECUTION_FILE: ${{ steps.cca.outputs.execution_file }}
         run: |
-          npm install -g @anthropic-ai/claude-code@2.1.210 >/dev/null 2>&1
-          claude -p "reply with exactly: PROBE_OK" --model claude-sonnet-4-6 --output-format json > out.json 2> err.txt || true
-          echo "--- stderr (first 20 lines) ---"; head -20 err.txt || true
-          echo "--- structured result ---"
-          jq '{subtype, is_error, api_error_status, num_turns, duration_ms, total_cost_usd, result: ((.result // "")|tostring|.[0:400])}' out.json || head -5 out.json
+          echo "outcome=${{ steps.cca.outcome }}"
+          if [ -n "${EXECUTION_FILE:-}" ] && [ -f "$EXECUTION_FILE" ]; then
+            jq '.[-1] | {subtype, is_error, api_error_status, num_turns, duration_ms, total_cost_usd, result: ((.result // "")|tostring|.[0:600])}' "$EXECUTION_FILE" || tail -3 "$EXECUTION_FILE"
+          else
+            echo "(no execution file)"
+          fi