MCP App Store
by KeenEthics
impactlake icon

impactlake MCP Server for ChatGPT

By IMPACTLAKE

Overview

Understand how your business or investment could affect people, society, and the environment. Explore relevant impact data, map how your activities could lead to change, and choose KPIs to measure that change. Review your saved impact models and record new KPI measurements through conversation. Some features are available without signing in. Connect your impactlake account to access and manage your saved models and KPI data. Available features depend on your account and permissions. Changes are previewed and saved only after your approval.

How to connect

  1. ChatGPT: Open the impactlake listing in the ChatGPT app directory and choose Connect. ChatGPT listing impactlake exposes 31 tools here.

Tools

apply_artifact_package

ChatGPT
Apply exactly the artifact package approved through preview_artifact_package to source data within the signed-in user's authorized Impactlake account or workspace. Does not publish content or create or update published copies, including KPIs with disclosure=PUBLIC. Returns per-item saved, failed-with-no-effect, dependency-blocked, or outcome-unknown state and never automatically resends an unknown write.

apply_domain_edit

ChatGPT
Apply a previously previewed domain edit to source data within the signed-in user's authorized Impactlake account or workspace, only after explicit user approval and optimistic-lock baseVersion validation. Does not publish content or create or update published copies. Allowed operations by target: logic_model, logic_model_wizard, and issue_map: rename_node, move_node, update_node_data, add_node, remove_node, connect_nodes, disconnect_edge, update_edge_data. calc_model and calc_wizard: add_kpi_node, update_kpi_node_data, remove_kpi_node, link_item_kpi_node, unlink_item_kpi_node. kpi: set_fields only. link_item_kpi_node and unlink_item_kpi_node are calculation-model operations, even though logic-model items contain kpiNodeIds. Use preview_domain_edit to validate a proposal before applying it.

compare_with_open_benchmark

ChatGPT
Compare local KPI/indicator with open KPI benchmark evidence.

convert_logic_model_json_to_operations

ChatGPT
Convert client-AI impactlake-logic-v1 JSON into Impactlake domain edit operations without saving.

create_impact_model

ChatGPT
Create a new impact model inside an owned portfolio and link it to that portfolio. Requires approval information from preview_create_impact_model. Enforces the unpublished-model plan limit.

create_issue

ChatGPT
Create a new issue (課題マップ) assigned to your account with an empty graph. Requires approval information from preview_create_issue.

create_kpi

ChatGPT
Create the approved standalone KPI within the signed-in user's Impactlake account and assign the creator account. Does not publish the KPI or create or update published copies, including when disclosure=PUBLIC. Requires the approval information from preview_create_kpi: pass that preview's nextApplyArguments unchanged with a new idempotencyKey, and set confirmMutation=true only after the user approves. Initial measurements and model edits remain separate approved operations.

explore_business_impact

ChatGPT
Add Impactlake perspectives to a business question: external change, unmet needs, capabilities, market opportunities and business effects; provide optional handoffs to detailed guides and tools. Use for open-ended questions about who benefits, what may change, how to measure or estimate it, supporting evidence, risks, comparisons, or user feedback. No existing Impactlake ID is needed.

export_impact_model

ChatGPT
Export an owned impact model as a pptx or pdf file. Returns a small descriptor plus a resourceUri; read that resourceUri via resources/read to download the base64 file blob (the binary is not inlined in this response). As in the app, PPTX requires a Plus plan on Open accounts; the PDF summary report is available on every plan.

get_client_ai_task_guide

ChatGPT
Retrieve specialist reasoning for logic models, measurement, calculation, evidence, comparison or survey design with mode=explore and no saved IDs. Use mode=existing for saved-resource operations. For initial perspective expansion, use explore_business_impact.

get_guided_domain_editing_state

ChatGPT
Get current state, follow-up questions, and a safe iterative workflow for natural-language logic model, KPI model, issue map, or KPI editing. The response identifies the task-specific authoring guide that must be retrieved before constructing the first proposal. Model targets require BOTH portfolioId and impactModelId from the search result; issue_map requires issueId and kpi requires kpiId.

get_impact_dimensions

ChatGPT
Read the structured Five Dimensions saved on one accessible Impact Model, in the same item shape as the Impactlake editor. This is not graph-node layout or Calculation Model dimensions.

get_kpi_measurement_history

ChatGPT
Read up to 200 recent observations of an owned KPI. Use after a saved or uncertain observation without asking permission to read. Match known historical IDs; equal values alone cannot identify an uncertain write, and absence from this bounded history does not prove failure. This is not a model-node KPI read.

get_mcp_usage_guide

ChatGPT
Explain how Impactlake can help with business value, measurement, estimates, evidence and user feedback. Technical details are opt-in.

get_save_context

ChatGPT
Return the authenticated account, available Portfolio destinations, current save-policy ceiling, and candidate scopes for saving Impact Models and KPIs. This tool accepts no business content and never saves, previews, or approves a package.

preview_artifact_package

ChatGPT
Validate one variable Impactlake artifact package and return a single human-readable save plan and package receipt without writing.

preview_create_impact_model

ChatGPT
Validate creating a new (empty) impact model inside an owned portfolio and issue approval information WITHOUT writing. Confirms portfolio ownership and the intention category. Populate the model afterward with apply_domain_edit or logic-model import.

preview_create_issue

ChatGPT
Validate creating a new issue (課題マップ) and issue approval information WITHOUT writing. Returns the issue creation plan. On apply the issue is assigned to your account with an empty graph you can fill in later in the editor.

preview_create_kpi

ChatGPT
Preview creation of a standalone Impactlake KPI without writing. It does not create an external integration or initial measurement. Resolve ID fields with resolve_creation_reference first. Returns resolved labels, duplicate candidates (not merged automatically), and the approval information and nextApplyArguments for create_kpi.

preview_domain_edit

ChatGPT
Validate and preview minimal structured operations for a logic model, KPI model, issue map, or KPI without saving. Allowed operations by target: logic_model, logic_model_wizard, and issue_map: rename_node, move_node, update_node_data, add_node, remove_node, connect_nodes, disconnect_edge, update_edge_data. calc_model and calc_wizard: add_kpi_node, update_kpi_node_data, remove_kpi_node, link_item_kpi_node, unlink_item_kpi_node. kpi: set_fields only. link_item_kpi_node and unlink_item_kpi_node are calculation-model operations, even though logic-model items contain kpiNodeIds. Use preview_domain_edit to validate a proposal before applying it. Retrieve the target-specific client AI task guide before constructing the first proposal; reading state alone is not the complete authoring contract. Returns normalized nextApplyArguments; pass them unchanged to apply_domain_edit and add only the matching approval information and confirmMutation after explicit approval.

preview_logic_model_import

ChatGPT
Convert a client-AI logic model JSON into operations and preview the import against a real Impactlake impact model without saving.

preview_record_kpi_measurement

ChatGPT
Validate recording a KPI measurement (a new KPI historical data point) and issue approval information WITHOUT writing. Confirms KPI ownership and value constraints; returns the measurement plan. This records an actual value, distinct from editing the KPI definition.

preview_update_impact_dimensions

ChatGPT
Preview replacing the structured Five Dimensions array on one Impact Model, in the same item shape as the Impactlake editor. Keep facts, assumptions and unknowns distinct in the conversation; they are not stored as separate fields. This does not write.

record_kpi_measurement

ChatGPT
Record a KPI measurement (create a KPI historical data point) for an owned KPI. Requires approval information from preview_record_kpi_measurement. At least one of amount/forecast/minVolatility/maxVolatility is required.

resolve_creation_reference

ChatGPT
Discover or resolve allowed reference options only for fields required by supported creation tools. Use mode=options when the user knows their goal but not the canonical intention or KPI-type name, or to list report frequencies for create_kpi; intention options require a business_activity or management_activity perspective. Use mode=search only when the user supplied a concrete reference name, and do not retry guessed or catch-all queries. Results are operation/field scoped, visibility filtered, and never expose an unrestricted master-data listing.

search_impact_models

ChatGPT
Find accessible existing impact models by a user-provided impact-model name, optionally within a selected portfolio. Use only for impact models; do not use for issue maps, logic models, workflows, or portfolios.

search_issues

ChatGPT
Find accessible Issue Maps by an explicit issue name. Distinguishes exact and partial matches and returns classification plus authorized related Impact Model and Portfolio context; it never treats no match as permission to create.

search_kpis

ChatGPT
Find accessible existing KPIs by a user-provided KPI name. Use only for existing KPIs; do not use for KPI classifications, impact models, portfolios, or master data.

search_open_impact_evidence

ChatGPT
Search open KPI and open impact model evidence. This read tool does not fetch arbitrary external URLs.

search_portfolios

ChatGPT
Find accessible existing portfolios by a user-provided portfolio name. Use only for portfolios; do not use for impact models, workflows, KPIs, meeting minutes, companies, or master data.

update_impact_dimensions

ChatGPT
Save the approved structured Five Dimensions to the existing Impact Model using optimistic version control and the matching approval information.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/impactlake"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/impactlake)