fix(dsql): bump databases-on-aws to 1.6.0, add Workflow 12 evals

Resolves the 1.5.0 version collision with EF Core PR awslabs#208 (both this
branch's system-diagnostics work and awslabs#208 independently claimed 1.5.0).
Bumps to 1.6.0 across plugin.json, codex plugin.json, and marketplace.json.

Adds eval coverage for Workflow 12 (system diagnostics), which previously
had none: 3 triggering cases and 1 functional AAS-interpretation eval.
Corrects stale eval README counts (37 trigger cases, 15 functional
prompts / 54 assertions) and documents evals 13-15.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: Morlej

.claude-plugin/marketplace.json

@@ -164,7 +164,7 @@
       "name": "databases-on-aws",
       "source": "./plugins/databases-on-aws",
       "tags": ["aws", "database", "aurora", "dsql", "serverless", "postgresql"],
-      "version": "1.5.0"
+      "version": "1.6.0"
     },
     {
       "category": "deployment",

plugins/databases-on-aws/.claude-plugin/plugin.json

@@ -22,5 +22,5 @@
   "license": "Apache-2.0",
   "name": "databases-on-aws",
   "repository": "https://github.com/awslabs/agent-plugins",
-  "version": "1.5.0"
+  "version": "1.6.0"
 }

plugins/databases-on-aws/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "databases-on-aws",
-  "version": "1.5.0",
+  "version": "1.6.0",
   "description": "Expert database guidance for the AWS database portfolio. Design schemas, execute queries, handle migrations, and choose the right database for your workload.",
   "author": {
     "name": "Amazon Web Services",

tools/evals/databases-on-aws/README.md

@@ -14,8 +14,8 @@ scripts, and unit tests for that database's skill.
 tools/evals/databases-on-aws/
 ├── README.md                        # This file — top-level index
 └── dsql/                            # Aurora DSQL skill evals
-    ├── evals.json                   # Tier 2: functional evals (9 prompts, 31 assertions)
-    ├── trigger_evals.json           # Tier 1: triggering evals (31 test cases)
+    ├── evals.json                   # Tier 2: functional evals (15 prompts, 54 assertions)
+    ├── trigger_evals.json           # Tier 1: triggering evals (37 test cases)
     ├── safe_query_evals.json        # Tier 3: safe_query enforcement (6 prompts, ~30 expectations)
     ├── query_explainability_evals.json  # Workflow 9: query plan diagnostics (9 prompts, 70 assertions)
     ├── query_plan_rewrite_evals.json   # Query rewrites: type coercion, subquery unnesting, etc. (11 prompts, manual)
@@ -55,7 +55,7 @@ PYTHONPATH="<skill-creator-path>:$PYTHONPATH" python -m scripts.run_eval \
 
 **What it checks:**
 
-- 16 should-trigger prompts (Aurora DSQL, distributed SQL, DSQL migrations, query plan explainability, data loading, etc.)
+- 22 should-trigger prompts (Aurora DSQL, distributed SQL, DSQL migrations, query plan explainability, system diagnostics / cluster performance, data loading, etc.)
 - 15 should-not-trigger prompts (DynamoDB, Aurora/RDS PostgreSQL with EXPLAIN ANALYZE, Redshift, generic SQL, non-DSQL bulk loading, etc.)
 
 ### Tier 2: Functional Evals
@@ -81,22 +81,25 @@ python tools/evals/databases-on-aws/dsql/scripts/run_functional_evals.py \
   --verbose
 ```
 
-**What it checks** (12 eval prompts, 42 assertions total):
-
-| Eval                           | Focus                 | Grader    | Key assertions                                                                                            |
-| ------------------------------ | --------------------- | --------- | --------------------------------------------------------------------------------------------------------- |
-| 1. Transaction limits          | MCP delegation        | regex     | Calls `awsknowledge`, cites 3,000 row limit, recommends batching                                          |
-| 2. Multi-tenant schema         | Correctness           | regex     | Uses `tenant_id`, `CREATE INDEX ASYNC`, no foreign keys, separate DDL txns                                |
-| 3. Index limits                | MCP delegation        | regex     | Calls `awsknowledge`, cites 24 index limit, suggests alternatives                                         |
-| 4. Python connection           | Language routing      | regex     | Recommends DSQL Python Connector, IAM auth, 15-min token expiry, SSL                                      |
-| 5. Column type change          | DDL migration routing | regex     | Table Recreation Pattern, DROP TABLE warning, batching, user confirmation                                 |
-| 6. JSON column storage         | Type guidance         | LLM judge | Recommends `JSONB` (or `JSON`) as the column type for queryable structured data                           |
-| 7. Array storage               | Type guidance         | LLM judge | Flags `TEXT[]` / array column as unsupported, recommends storing the array as `JSONB`                     |
-| 8. INACTIVE cluster error      | Troubleshooting       | LLM judge | Identifies INACTIVE state, uses `aws dsql get-cluster` to poll until `ACTIVE`, retries afterwards         |
-| 9. Backup on IDLE/INACTIVE     | Troubleshooting       | LLM judge | Identifies `FailedPrecondition`, connects to wake cluster to ACTIVE, retries backup                       |
-| 10. Loader stuck at 3K rec/s   | Data loading          | LLM judge | Identifies partition-constrained fresh table, advises to keep running, does NOT recommend more workers    |
-| 11. Loader crash lost manifest | Data loading          | LLM judge | Identifies /tmp as tmpfs, recommends --on-conflict do-nothing for recovery, --manifest-dir for prevention |
-| 12. Header row parse error     | Data loading          | LLM judge | Identifies missing --header flag, explains default behavior, recommends fix                               |
+**What it checks** (15 eval prompts, 54 assertions total):
+
+| Eval                           | Focus                 | Grader    | Key assertions                                                                                                        |
+| ------------------------------ | --------------------- | --------- | --------------------------------------------------------------------------------------------------------------------- |
+| 1. Transaction limits          | MCP delegation        | regex     | Calls `awsknowledge`, cites 3,000 row limit, recommends batching                                                      |
+| 2. Multi-tenant schema         | Correctness           | regex     | Uses `tenant_id`, `CREATE INDEX ASYNC`, no foreign keys, separate DDL txns                                            |
+| 3. Index limits                | MCP delegation        | regex     | Calls `awsknowledge`, cites 24 index limit, suggests alternatives                                                     |
+| 4. Python connection           | Language routing      | regex     | Recommends DSQL Python Connector, IAM auth, 15-min token expiry, SSL                                                  |
+| 5. Column type change          | DDL migration routing | regex     | Table Recreation Pattern, DROP TABLE warning, batching, user confirmation                                             |
+| 6. JSON column storage         | Type guidance         | LLM judge | Recommends `JSONB` (or `JSON`) as the column type for queryable structured data                                       |
+| 7. Array storage               | Type guidance         | LLM judge | Flags `TEXT[]` / array column as unsupported, recommends storing the array as `JSONB`                                 |
+| 8. INACTIVE cluster error      | Troubleshooting       | LLM judge | Identifies INACTIVE state, uses `aws dsql get-cluster` to poll until `ACTIVE`, retries afterwards                     |
+| 9. Backup on IDLE/INACTIVE     | Troubleshooting       | LLM judge | Identifies `FailedPrecondition`, connects to wake cluster to ACTIVE, retries backup                                   |
+| 10. Loader stuck at 3K rec/s   | Data loading          | LLM judge | Identifies partition-constrained fresh table, advises to keep running, does NOT recommend more workers                |
+| 11. Loader crash lost manifest | Data loading          | LLM judge | Identifies /tmp as tmpfs, recommends --on-conflict do-nothing for recovery, --manifest-dir for prevention             |
+| 12. Header row parse error     | Data loading          | LLM judge | Identifies missing --header flag, explains default behavior, recommends fix                                           |
+| 13. EF Core data layer setup   | ORM routing (.NET)    | LLM judge | Recommends Amazon.AuroraDsql.EntityFrameworkCore, Guid keys w/ gen_random_uuid(), app-layer FK, DsqlExecutionStrategy |
+| 14. .NET / C# support          | Language routing      | LLM judge | Confirms .NET support, recommends Npgsql connector for IAM auth and EF Core adapter                                   |
+| 15. System diagnostics (W12)   | AAS interpretation    | LLM judge | Identifies the shifted wait event vs baseline, rules out load growth, no absolute-AAS claim, defers to Workflow 9     |
 
 ### Grader modes
 

tools/evals/databases-on-aws/dsql/evals.json

@@ -170,6 +170,20 @@
         "Recommends the Amazon.AuroraDsql.EntityFrameworkCore adapter for EF Core users"
       ],
       "llm_judge": true
+    },
+    {
+      "id": 15,
+      "prompt": "My Aurora DSQL cluster is performing differently than last week and I can't point to one bad query. I pulled the Active Average Sessions (db.active_sessions.avg) broken down by wait event for the same hour, this week vs last week. Last week: OnCpu 226, SequentialScanRead 89, UniqueConstraintCheck 89, Commit 74, everything else flat. This week: OnCpu 213, SequentialScanRead 336, UniqueConstraintCheck 87, Commit 73, everything else flat. What does this tell you and what should I do next?",
+      "expected_output": "Interprets the data as a wait-event distribution shift: SequentialScanRead grew ~3.7x (from ~12% to ~35% of total AAS) while every other wait event, including OnCpu and Commit, stayed flat. Concludes this is NOT overall load growth (which would raise all events together) but a scan/plan regression. Does NOT treat any absolute AAS number as inherently bad — reasons from the change vs the temporal baseline. Recommends identifying the top SQL on SequentialScanRead and handing those queries off to per-query EXPLAIN analysis (Workflow 9). Does not recommend schema or index changes from CloudWatch data alone.",
+      "files": [],
+      "expectations": [
+        "Identifies SequentialScanRead as the shifted wait event and quantifies the change relative to the baseline (roughly 3.7x, or a jump from ~12% to ~35% of total AAS)",
+        "Concludes this is a scan/plan regression rather than proportional load growth, citing that OnCpu / Commit / other events stayed flat",
+        "Does NOT claim any absolute AAS value is inherently problematic — reasons from change over time against the temporal baseline",
+        "Recommends identifying the top SQL statements attributed to SequentialScanRead as the next step",
+        "Defers per-query root cause to EXPLAIN / query plan analysis (Workflow 9) and does NOT recommend specific index or schema changes from the CloudWatch metrics alone"
+      ],
+      "llm_judge": true
     }
   ]
 }

tools/evals/databases-on-aws/dsql/trigger_evals.json

@@ -51,6 +51,18 @@
     "query": "I think there's a bug in DSQL's EXPLAIN ANALYZE output — my Storage Lookup is reporting trillions of actual rows on a table that only has 300k rows. Can you diagnose this?",
     "should_trigger": true
   },
+  {
+    "query": "my Aurora DSQL cluster is performing differently than it was last week — can you diagnose what changed?",
+    "should_trigger": true
+  },
+  {
+    "query": "my dsql cluster feels slow but no single query stands out. how do I find where the time is going — wait events, active sessions, that kind of thing?",
+    "should_trigger": true
+  },
+  {
+    "query": "where is compute time being spent on my Aurora DSQL cluster? I want to see the AAS breakdown by wait event.",
+    "should_trigger": true
+  },
   {
     "query": "I need to load a 500MB CSV file into my Aurora DSQL table. What's the fastest way?",
     "should_trigger": true