OpenAPI operationId truncated to 60 chars without uniqueness → shadowed RestApiTool names

[michaelkillgta]

Description

Gemini’s ~64-char function name limit is handled by a hard slice to 60 characters with no uniqueness suffix:

• google/adk/tools/openapi_tool/openapi_spec_parser/operation_parser.py — get_function_name ([:60])
• google/adk/tools/openapi_tool/openapi_spec_parser/rest_api_tool.py — RestApiTool.__init__ (name[:60])

Codegen-style OpenAPI specs routinely produce long operationIds that share a long snake_case prefix. Distinct operations collapse to one tool name. LlmRequest.append_tools still sends two FunctionDeclarations with that name to the model, but tools_dict keeps only the last registration (warning logged; calls always hit the survivor).

Environment

• google-adk==2.9.2
• Offline unit repro

Repro

Minimal OpenAPI with two long operationIds that share a 60-char snake_case prefix after conversion, e.g.:

• listAllCompanyOrganizationDepartmentUsersByFilterCriteriaV1 → /users/v1
• listAllCompanyOrganizationDepartmentUsersByFilterCriteriaV2 → /users/v2

Build OpenAPIToolset / register tools into an LlmRequest and inspect names + tools_dict.

Observed (google-adk==2.9.2)

• Both tools named list_all_company_organization_department_users_by_filter_cri (len 60)
• Warning: duplicate tool name … previously registered tool is shadowed
• tools_dict has 1 key; survivor path = /users/v2; first endpoint unreachable
• Declarations list still contains two entries with the same name

Expected

After truncation, colliding names should get a short stable disambiguator so each operationId maps to a unique callable tool, and tools_dict / declarations stay 1:1.

Suggested fix

After [:60], if the name collides with an already-emitted tool in the toolset, append a short stable disambiguator (e.g. _2, hash of full operationId, or path+method suffix).

Notes

#60 is an unrelated tools_dict miss for agent-as-tool naming. Happy to open a PR.

[chelsealong]

Picking this one up now — opening a PR shortly. Flagging it here so nobody duplicates the work; if someone is already on it, say so and I will drop mine.

[michaelkillgta]

Thanks for picking this up so quickly — and for the PRs (#7285 / #7284 / #7283). Happy to review or clarify anything from the original repros if useful.