E2 ChatBot Investigation MCP

Version 1.0.0 ยท 11 tools

MCP URL: https://dev.mcp.s-tech-services.at/e2-chat-bot-investigation

Instructions

E2 ChatBot Investigation MCP exposes read-only analytics tools for investigating chatbot conversations, AI calls, pipeline runs, tool calls, and entity resolution issues.

Preferred Workflow

  • Start with conversation investigation suggestions when the issue is broad or unclear.
  • Use the conversation dossier before drilling into raw logs.
  • Use AI call logs, tool logs, pipeline runs, and resolver logs only for focused follow-up.

Important Rules

  • Prefer exact UUID values when an analytics endpoint uses strict search behavior.
  • Preserve large request, response, arguments, result, metadata, and debug payloads.
  • Some nested lookups are correlation-scoped or parent-scoped, not generic free joins.

Tool List

  1. Read Guide
  2. List Conversation Investigation Suggestions
  3. Get Conversation Dossier
  4. List AI Call Logs
  5. Get AI Call Log
  6. List AI Call Tool Logs For Call
  7. List Pipeline Runs
  8. Get Pipeline Run
  9. List Pipeline Step Runs
  10. List Entity Resolver Logs
  11. List Entity Resolver Log Entries

Read Guide

read_guide

Read the investigation guide with workflow, search guidance, data locations, and edge cases for chatbot analytics.

readOnlyHint: true

Parameters

This tool does not accept parameters.

Response

This tool returns JSON. A formal output schema is not declared.

List Conversation Investigation Suggestions

list_conversation_investigation_suggestions

List ranked conversations worth investigating based on failures, problematic calls, unresolved entities, or high runtime.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
user_id integer No Optional user ID to scope the conversation suggestions.
from string No Optional start datetime.
to string No Optional end datetime.
page integer No Page number.
per_page integer No Page size, maximum 50.
sort string No Optional suggestion ranking mode. problematic, failed_pipeline_runs, failed_ai_calls, problematic_tool_calls, unresolved_entity_events, long_runtime
exclude_without_reasons boolean No Exclude conversations without active investigation reasons. Defaults to true.

Response

This tool returns JSON. A formal output schema is not declared.

Get Conversation Dossier

get_conversation_dossier

Fetch a bounded investigation dossier for a single conversation, including summary metrics and related resources.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
conversation_id integer Yes Conversation ID.
include_message_previews boolean No Include message previews in the dossier. Defaults to true.

Response

This tool returns JSON. A formal output schema is not declared.

List AI Call Logs

list_ai_call_logs

List AI call logs for investigation, including failed, slow, or otherwise problematic model calls.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
search string No Numeric ID or exact UUID-style search value.
status string No Optional AI call status filter. pending, success, failed
problematic boolean No Filter to problematic AI calls.
tenants string No Optional comma-separated tenant IDs such as 5,9.
page integer No Page number.
per_page integer No Page size, maximum 100.
sort string No Sort column. id, provider_key, model_key, cost_total_in_dollar, runtime_in_ms, status, created_at
direction string No Sort direction. asc, desc

Response

This tool returns JSON. A formal output schema is not declared.

Get AI Call Log

get_ai_call_log

Fetch a single AI call log including request, response, runtime, cost, and problematic reasons.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
ai_call_log_id integer Yes AI call log ID.

Response

This tool returns JSON. A formal output schema is not declared.

List AI Call Tool Logs For Call

list_ai_call_tool_logs_for_call

List tool logs associated with a single AI call log via correlation ID.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
ai_call_log_id integer Yes AI call log ID.
page integer No Page number.
per_page integer No Page size, maximum 100.
sort string No Sort column. id, tool_name, mcp_tool_name, status, runtime_in_ms, created_at
direction string No Sort direction. asc, desc

Response

This tool returns JSON. A formal output schema is not declared.

List Pipeline Runs

list_pipeline_runs

List pipeline runs so investigation can start from failed or slow execution flows.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
search string No Numeric ID or exact UUID-style search value.
status string No Optional pipeline run status filter. pending, running, completed, failed
pipeline_id integer No Optional pipeline ID filter.
tenants string No Optional comma-separated tenant IDs such as 1,2,3.
page integer No Page number.
per_page integer No Page size, maximum 100.
sort string No Sort column. id, status, started_at, finished_at, runtime_in_ms, created_at
direction string No Sort direction. asc, desc

Response

This tool returns JSON. A formal output schema is not declared.

Get Pipeline Run

get_pipeline_run

Fetch a single pipeline run including status, input and output payloads, runtime, and errors.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
pipeline_run_id integer Yes Pipeline run ID.

Response

This tool returns JSON. A formal output schema is not declared.

List Pipeline Step Runs

list_pipeline_step_runs

List step runs belonging to a pipeline run to inspect where an execution failed or slowed down.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
pipeline_run_id integer Yes Pipeline run ID.
page integer No Page number.
per_page integer No Page size, maximum 100.
sort string No Sort column. id, pipeline_step_order, status, started_at, finished_at, runtime_in_ms
direction string No Sort direction. asc, desc

Response

This tool returns JSON. A formal output schema is not declared.

List Entity Resolver Logs

list_entity_resolver_logs

List aggregate entity resolver logs to investigate not-resolved and ambiguous resolution issues.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
search string No Numeric ID or prefix search for requested or normalized values.
entity_key string No Optional entity key filter.
status string No Optional resolver status filter. resolved, not_resolved
source string No Optional resolver source filter. local_db, external_api
page integer No Page number.
per_page integer No Page size, maximum 100.
sort string No Sort column or supported camelCase alias. id, entity_key, latest_status, latest_source, total_count, resolved_count, not_resolved_count, cache_hit_count, entry_count, first_seen_at, last_seen_at, entityKey, latestStatus, latestSource, totalCount, resolvedCount, notResolvedCount, cacheHitCount, entryCount, firstSeenAt, lastSeenAt
direction string No Sort direction. asc, desc

Response

This tool returns JSON. A formal output schema is not declared.

List Entity Resolver Log Entries

list_entity_resolver_log_entries

List raw entity resolver log entries for a single aggregate resolver log.

readOnlyHint: true

Parameters

Name Type Required Description Constraints
resolver_log_id integer Yes Entity resolver log ID.
page integer No Page number.
per_page integer No Page size, maximum 100.

Response

This tool returns JSON. A formal output schema is not declared.

Tool List