[finklen-amazon]

Fixes

Summary

Changes

Savings percentage (list_recommendation_summaries). The tool never sent metrics, so the service always answered metrics.savingsPercentage with the placeholder "Percentage not requested.", and the tool returned that string as metrics.savings_percentage. The request now sends metrics=['SavingsPercentage']. A percentage value ('12%', or the clamped '>=80%' / '<=2%') is returned as metrics.savings_percentage. When the service does not compute it, it puts an explanation in the same field; that text is now returned as metrics.savings_percentage_unavailable_reason. The service gives an explanation in three cases: savings estimation mode BeforeDiscounts, a filter other than accountIds, regions and tags, or a failed calculation. The description says what the value is: estimated deduplicated monthly savings as a percent of the last 30 days' net amortized cost, excluding credits and refunds (the service filters out the Credit and Refund record types), for the whole filtered result, not per group.

Boolean filters. restartNeeded / rollbackPossible given as "true" / "false" strings failed botocore parameter validation. They are now coerced to booleans. Other values pass through unchanged.

Summaries group_by. Fixed dimensions now match case-insensitively (resourcetype -> ResourceType). Underscores are not ignored, so the legacy RESOURCE_TYPE spelling is still rejected. The missing-group_by error now lists the same values as the invalid-group_by error, including TagKey:<tag key>. The unreachable 'RESOURCE_TYPE' fallback is removed.

Filter docs. DatabaseSavingsPlans support (folding and valid_filter_values) now comes from main (#4948). The description now also covers three more things:

• The resourceIds, resourceArns and recommendationIds filters are listed with the other filter parameters as list_recommendations only (summaries reject them locally, since feat(billing-cost-management-mcp-server): add COH summaries total cou… #4934).
• There is no per-service filter, so to cover a service you list each of its types.
• Summary fields are the same on every page, including continuation calls. The description previously said they were not repeated; a live continuation call returns them.

No tool parameter is added, renamed, retyped or given a new default.

User experience

Before:

• "metrics": {"savings_percentage": "Percentage not requested."} on every summaries call.
• filters='{"restartNeeded": "true"}' gave a parameter validation error.
• group_by='resourcetype' gave a validation error.

After:

• "metrics": {"savings_percentage": "12%"}, or {"savings_percentage_unavailable_reason": "<service text>"} when the service does not compute a percentage.
• The string boolean is accepted.
• group_by='resourcetype' is sent as ResourceType; RESOURCE_TYPE is still rejected locally.

Testing

Run from src/billing-cost-management-mcp-server:

• uv run --frozen pytest: 3515 passed (after the latest rebase onto main); patch coverage 21/21 added executable lines.
• ruff format --check, ruff check and pyright on the changed files: clean.
• pre-commit on the changed files: clean.

The new tests failed before the fix (16 failures) and pass after.

Live check (read-only): a harness started the server over stdio and read the description as served. It then made 4 list_recommendation_summaries calls and 2 list_recommendations calls (max_items=1).

• On this change, 21/21 passed:
  • an unfiltered summary returns a percentage string;
  • an actionTypes filter returns the service's incompatible-filter reason, not a metric;
  • group_by='resourcetype' is accepted (an earlier revision folded underscores too; it now ignores letter case only);
  • summaries reject resourceIds;
  • restartNeeded: "false" is accepted;
  • DatabaseSavingsPlans is accepted as a filter value.
• On the base commit, 5/21 passed. Those 5 are API facts both versions share; every fixed claim fails.
• After the rebase onto main, the full harness passed 24/24 with 5 read-only calls. Since feat(billing-cost-management-mcp-server): add COH summaries total cou… #4934 the resourceIds check expects the tool's local validation_error. On current main, the description checks pass 1/13.

Merge conflict

After #4945 and #4946 merged, this branch conflicted with main in cost_optimization_hub_helpers.py, cost_optimization_hub_tools.py and tests/tools/test_cost_optimization_hub_tools.py. I merged main in (a merge commit, no rebase) and resolved it:

• helpers docstring: took main's wording, "Grouping dimension (e.g. ResourceType)".
• Description: kept this PR's savings-percentage paragraph and docs(billing-cost-management-mcp-server): warn that summing Cost Opti… #4946's SAVINGS TOTALS section, both unchanged.
• effective_group_by: both sides removed the legacy 'RESOURCE_TYPE' fallback; kept a single copy.
• Reject tests: kept fix(billing-cost-management-mcp-server): drop legacy RESOURCE_TYPE group_by and test COH enum lists against the API #4945's RESOURCE_TYPE case. Dropped main's lowercase resourcetype reject line, because this PR matches fixed group_by dimensions case-insensitively (underscores are not ignored, so RESOURCE_TYPE and resource_type stay rejected).
• Earlier, this branch had also been reconciled with feat(billing-cost-management-mcp-server): add COH summaries total cou… #4934 and fix(billing-cost-management-mcp-server): accept DatabaseSavingsPlans and document Cost Optimization Hub region behavior #4948, keeping main's code where both added the same behavior.

After the merge: 4271 tests pass, ruff and pyright are clean, and the description checks pass 14/14.

Checklist

If your change doesn't seem to apply, please leave them unchecked.

• I have reviewed the contributing guidelines
• I have performed a self-review of this change
• Changes have been tested
• Changes are documented

Is this a breaking change? (Y/N) Y. When no percentage is computed, metrics.savings_percentage is now absent and the service's text is in metrics.savings_percentage_unavailable_reason. Before this change the field always held the "Percentage not requested." placeholder, never a percentage. This is recorded in CHANGELOG.md.

RFC issue number:

Checklist:

• Migration process documented
• Implement warnings (if it can live side by side)

Acknowledgment

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of the project license.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.79%. Comparing base (c6951b9) to head (032b66d).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4906   +/-   ##
=======================================
  Coverage   93.79%   93.79%           
=======================================
  Files        1080     1080           
  Lines       94586    94599   +13     
  Branches    15402    15404    +2     
=======================================
+ Hits        88715    88728   +13     
  Misses       3498     3498           
  Partials     2373     2373           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.