MCP App Store
by KeenEthics
Revelica icon

Revelica MCP Server for Claude and ChatGPT

By Revelica

Overview

Revelica is a continuous product discovery engine that loops strategy, execution, and learning into a single system. This connector empowers Claude and coding agents to read and write directly to your team’s workspace, keeping strategic context, user research, product bets, and discovery experiments synchronized in one shared knowledge graph. A Revelica workspace consolidates your technical discovery stack into a unified graph: Customer & Market Research: Interview snapshots with cited sources, segmented opportunities, competitive analyses, and scored market maps. Strategy & Frameworks: Unified product knowledge graph, RICE assessments, and unified product specs with story map, document, and markdown export/import views. * Hypotheses & Execution: OKR-driven product bets, assumption finding, validation experiments, and MCP plugins for coding agents. Every request is authenticated with OAuth and scoped to the user's own workspace and individual permission. Revelica does not read your chat history, conversation summaries, or files. Built for product managers, product leaders, and the engineering teams collaborating with modern AI coding agents. Bring continuity to your discovery workflow. Get started for free at revelica.com.

How to connect

  1. Claude: Open the Revelica listing in the Claude connectors directory and choose Connect. Claude listing Revelica exposes 6 tools here.
  2. ChatGPT: Open the Revelica listing in the ChatGPT app directory and choose Connect. ChatGPT listing Revelica exposes 10 tools here.

Tools

create

ChatGPT
Create a new artifact or a new version of an existing one. Call query or read with include_template=true for this type first. Content is validated against the Pydantic schema for the artifact type. A validation failure returns the error and the type's template, so you can correct the call. Link records with links, a list of {"rel": <verb>, "to": <ref>}: inside content on create, or as updates={"links": [...]} on update (adds links). Example: {"rel": "affects", "to": "STORY-42"}; a part's ref (STORY-42) links its record, anchored to that part. Create an edge directly only when it carries properties a link can't (a predicts edge's magnitude); its ends are refs too. Templates include link_rules (out/in rules: rel, targets, min/max) and allowed_edge_types; min > 0 means a required link in that direction. On success, links lists persisted relationships for the returned id: edge_type/direction for edges, or field for FK/id-list references, plus to (type, id, and name when available). links_summary counts distinct targets. Report links from these fields, not from the requested intent. Removed links are absent. links_truncated=true means a partial receipt; links=null means the write succeeded but links could not be verified. If parent_id is provided, a new version is created in the version chain; otherwise a brand-new artifact (version 1) is created.

list_skills

ChatGPT
List the Revelica skills you can load in this workspace. Each entry has name and description; the description says what the skill does and when to use it. A skill that needs existing records first also lists requirements: the input, its artifact type and the skill that sets it up. Load one with load_skill(skill_name=...), which returns its full instructions. Pass project_id or context_entities to also get each skill's readiness for that project or those records. Use it at the start of a session, or when the job in hand matches none of the skills named in the server instructions.

load_skill

ChatGPT
Load an agent skill's full instructions and reference materials. Skills are specialized instruction sets for specific tasks. Only skills declaring available-in: mcp in their frontmatter are loadable via this tool. Call list_skills first to see what's available in your workspace.

open_project_panel

ChatGPT
Open the Revelica Canvas beside the conversation, on a project's home. Opened by the user as a tab in the host's conversation, never by the model. With no project_id, it opens the project the user is working in.

open_revelica

ChatGPT
Open the Revelica Canvas on a project's home: its goal, plan, bets and open decisions. Opened by the user from the host's sidebar, never by the model. With no project_id, it opens the project the user is working in.

query

ChatGPT
Search/filter records in the Revelica workspace by criteria. Returns multiple matches with snippets and ranks. Backed by Postgres full-text search (name weighted above content). For fetching a specific known resource by ID, use read instead.

read

ChatGPT
Fetch a specific resource by ID. Returns the full content with type-aware rendering. Two access modes: - No `path`: returns the whole entity for structured content, or paginates by offset/limit for text-shaped content (long markdown docs, etc.). - With `path`: navigates to a subtree using dotted-path syntax (same as update's field paths — dict keys, list items by ref or index, e.g. stories.STORY-42). Use this to target large entities without paying for the whole shape. For finding resources matching criteria, use query instead. A resource may carry open_in_revelica, a link that opens it in the Revelica app. Give it to the user when they'd want to see or edit it there.

send_feedback

ChatGPT
Send Revelica's developers a short report when a Revelica tool got in your way or worked notably well, then tell the user in one line that you sent it. Most valuable: a missing capability, when no tool fits your task, even if you worked around it. Also: a tool that failed, an error that didn't say how to fix it, a description or schema that misled you. One specific report per problem. It records feedback; it doesn't change any tool or the user's workspace. No customer data or personal details.

show

ChatGPT
Show the user one thing: navigate them to focus, the record or view they should look at now, and list up to five supporting records they can open next. Say in your reply why the focus matters and what the supporting records are. Call it once at the end of a turn in which you read or changed records, including tasks you run from a plan, and mid-conversation whenever you discuss a specific record or view the user should look at, or they ask to see something ('what source is that?' means show that source); not after every save, and not for records you only read for yourself. The Canvas is Revelica's workspace view, where the user sees and edits records. Args: focus: Required: the one record or view to look at now. A record is its ref, e.g. "SPEC-7"; it opens on its own Canvas page, and a part's ref (e.g. "STORY-42") opens its record. A view is {layer, view} with slugs from canvas_layers_catalog, e.g. {layer: "context", view: "matrix"}; add id (a ref) to open a record inside that view. A project opens its home. Invalid refs or slugs return what's valid. The focus always appears first and navigates the user's live Canvas immediately. supporting: Up to five other records the user can open next, by ref, e.g. ["IDEA-9", "OPP-14"]. Say in your reply what they are and why the focus matters. mode: "inline" (default) or "fullscreen". A new MCP frame stays inline unless fullscreen is requested; an already-open Canvas follows this conversation's focus in either mode. The app navigates immediately in either mode. Never request pip. title: Optional one-line card header explaining what to look at, for example "The bet to test first". project_id: Optional project for a view or workspace record; project-scoped records open in their own project. Hosts with UI present the focus as an interactive Canvas the user can expand to fullscreen; terminal hosts show links instead, so there include the focus link in your reply. Returns the focus link first, then supporting records' names and links.

update

ChatGPT
Update specific fields on an existing artifact or entity. Fetches the current content, merges your updates, validates against the schema, and saves. For artifacts this creates a new version; for entities this updates in place. A validation failure returns the error and the type's template, so you can correct the call. Link records with links, a list of {"rel": <verb>, "to": <ref>}: inside content on create, or as updates={"links": [...]} on update (adds links). Example: {"rel": "affects", "to": "STORY-42"}; a part's ref (STORY-42) links its record, anchored to that part. Create an edge directly only when it carries properties a link can't (a predicts edge's magnitude); its ends are refs too. Templates include link_rules (out/in rules: rel, targets, min/max) and allowed_edge_types; min > 0 means a required link in that direction. On success, links lists persisted relationships for the returned id: edge_type/direction for edges, or field for FK/id-list references, plus to (type, id, and name when available). links_summary counts distinct targets. Report links from these fields, not from the requested intent. Removed links are absent. links_truncated=true means a partial receipt; links=null means the write succeeded but links could not be verified. Args: id: The record's ref (e.g. SPEC-7). updates: Dictionary of field paths to new values. Supports: - Top-level fields: {"memorable_quote": "new quote"} - Nested dot-paths: {"experience_map.moments.2.what_feeling": "frustrated"} - A list item by its ref (prefer this to an index): {"stories.STORY-42.title": "..."}; removing two items by ref in one call can't shift either. - List append with "-" (one item): {"opportunities.-": {"name": "...", "type": "pain_point"}} - List extend with "+" (several items): {"features.+": [{...}, {...}]} - A plain list value replaces the whole list: {"features": [...]} - String append with ".append": {"description.append": " New section..."} concatenates onto an existing string field instead of replacing it (value may be str or list[str]; lists are newline-joined). Use this to grow a long text/markdown field section by section without re-sending the whole value. - Remove with null: {"stories.STORY-42": null} or {"insights.3": null} Use the field's own name (no content. prefix). A warning names any path that isn't a field, or any change that didn't save. Returns: JSON with success status, ref, and fields updated. created lists the refs of any parts the update added (e.g. a new release). It may carry open_in_revelica, a link to the record in the Revelica app; give it to the user when they'd want to see it there.

create

Claude

list_skills

Claude

load_skill

Claude

query

Claude

read

Claude

update

Claude

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/revelica"><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/revelica)