Executes a CloudWatch Logs Insights query against trace span records stored in CloudWatch Logs with the OpenTelemetry semantic-convention schema (@data_format = "AWS-OTEL-TRACE-V1"). Scope: only log records tagged `@data_format = "AWS-OTEL-TRACE-V1"` will match, so span records must follow the O...
Part of the CloudWatch Application Signals MCP Server MCP server. Enforce policies on this tool with Intercept, the open-source MCP proxy.
AI agents call search_transaction_spans to retrieve information from CloudWatch Application Signals MCP Server without modifying any data. This is common in research, monitoring, and reporting workflows where the agent needs context before taking action. Because read operations don't change state, they are generally safe to allow without restrictions -- but you may still want rate limits to control API costs.
Even though search_transaction_spans only reads data, uncontrolled read access can leak sensitive information or rack up API costs. An agent caught in a retry loop could make thousands of calls per minute. A rate limit gives you a safety net without blocking legitimate use.
Read-only tools are safe to allow by default. No rate limit needed unless you want to control costs.
tools:
search_transaction_spans:
rules:
- action: allowSee the full CloudWatch Application Signals MCP Server policy for all 22 tools.
Agents calling read-class tools like search_transaction_spans have been implicated in these attack patterns. Read the full case and prevention policy for each:
Other tools in the Read risk category across the catalogue. The same policy patterns (rate-limit, allow) apply to each.
Executes a CloudWatch Logs Insights query against trace span records stored in CloudWatch Logs with the OpenTelemetry semantic-convention schema (@data_format = "AWS-OTEL-TRACE-V1"). Scope: only log records tagged `@data_format = "AWS-OTEL-TRACE-V1"` will match, so span records must follow the OpenTelemetry semantic-convention schema (e.g. `attributes.aws.local.service`, `attributes.aws.remote.operation`). If your spans are stored in some other shape or log group, pass an explicit `log_group_name` and include your own filter in `query_string`. The tool adds the `@data_format = "AWS-OTEL-TRACE-V1"` filter automatically, so you do not need to include it in `query_string`. You can override by providing your own `@data_format` reference, or take control of log-group scoping by starting `query_string` with a `SOURCE logGroups(...)` clause. Don't combine `log_group_name` with a user-supplied `SOURCE` — `log_group_name` is dropped and `log_group_name_ignored: True` appears in the response. The volume of returned logs can easily overwhelm the agent context window. Always include a limit in the query (| limit 50) or via the limit parameter. Usage: Write CloudWatch Logs Insights queries over OpenTelemetry span attributes (filter, stats, sort, limit, etc.). If source code is not accessible, consider querying with code-level attributes. ⚠️ Use CORRECT attribute names: attributes.code.file.path, attributes.code.function.name, attributes.code.line.number ``` FILTER attributes.aws.local.service = "customers-service-java" and attributes.aws.local.environment = "eks:demo/default" and attributes.aws.remote.operation="InvokeModel" | STATS sum(`attributes.gen_ai.usage.output_tokens`) as `avg_output_tokens` by `attributes.gen_ai.request.model`, `attributes.aws.local.service`,bin(1h) | DISPLAY avg_output_tokens, `attributes.gen_ai.request.model`, `attributes.aws.local.service` ``` Returns: -------- A dictionary containing the final query results, including: - status: one of 'Scheduled', 'Running', 'Complete', 'Failed', 'Cancelled', 'Polling Timeout', 'Transaction Search Not Available', 'Invalid Input'. 'Invalid Input' is returned synchronously when query_string is empty or whitespace-only. - results: A list of the actual query results if the status is Complete. - statistics: Query performance statistics - messages: Any informational messages about the query - transaction_search_status: Information about transaction search availability - log_group_name_ignored (only when true): set when log_group_name was supplied but dropped because query_string already contained a SOURCE clause. Accompanied by log_group_name_ignored_reason.. It is categorised as a Read tool in the CloudWatch Application Signals MCP Server MCP Server, which means it retrieves data without modifying state.
Add a rule in your Intercept YAML policy under the tools section for search_transaction_spans. You can allow, deny, rate-limit, or validate arguments. Then run Intercept as a proxy in front of the CloudWatch Application Signals MCP Server MCP server.
search_transaction_spans is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the search_transaction_spans rule in your Intercept policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the Intercept policy for search_transaction_spans. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
search_transaction_spans is provided by the CloudWatch Application Signals MCP Server MCP server (awslabs.cloudwatch-applicationsignals-mcp-server). Intercept sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic policy on every MCP tool call. Per-identity grants. Full audit log.