LABSCO SUMMARY
Access following the account rather than the URL is the right answer to a per-org address.
A URL containing your org id looks like a secret and is not one — people paste them into tickets. CoreStory's documentation is explicit that the URL is not the grant: sign-in is, and removal from the org cuts access at once. That makes the address safe to share internally. On the tools, sections_only with paging is the detail that keeps a large tech spec usable — an agent can read the headings and then one section rather than the whole document.
What it is
CoreStory's endpoint, addressed per organisation — the URL contains your org slug and id. Fourteen tools split between reading generated documents, holding conversations about the codebase, and searching the index behind them.
What you get
- list_projects, get_project_prd, get_project_techspec — the generated documents, whole or by named section with paging
- list_conversations, get_conversation, create_conversation, rename_conversation, send_message — a persistent conversation per topic
- refine_document_definition, generate_document, get_document_generation_result — shape what a document should contain, then generate it asynchronously
- semantic_search, describe_index, filter_chunks — query the index, and inspect what is in it
Requirements
A CoreStory account with a project already ingested, and browser sign-in through Clerk. Access follows the account rather than the URL: being removed from the organisation ends MCP access immediately.
Setup effort
Get your endpoint, then paste it — add your endpoint to your client, then approve the OAuth consent screen
USE CASES
Understanding a codebase you did not write
Read the generated tech spec's section list, then pull only the sections that matter.
Asking follow-up questions over time
Create a conversation and keep sending into it rather than starting fresh each session.
Finding the relevant code
Search semantically, then filter the chunks and check what the index actually covers.
Producing a document to a shape you choose
Refine the definition first, then generate and collect the result asynchronously.
14 tools
Listed in project docs. Your client may show a different list.
get_project_prd
sections_only, sections, limit, offset
get_project_techspec
sections_only, sections, limit, offset
list_conversations
get_conversation
create_conversation
rename_conversation
send_message
refine_document_definition
generate_document
get_document_generation_result
semantic_search
describe_index
filter_chunks
BEFORE YOU INSTALL
What to know first
Tools you get
14 tools named in the project's documentation.
Requires
- an MCP client that can reach a remote endpoint
- OAuth through Clerk; the older bearer mcp_ tokens are deprecated
INSTALLATION
Connect your client
Add your endpoint to your client, then approve the OAuth consent screen.
Server setup
https://app.corestory.ai/mcp/<YOUR_ORG_SLUG>-<YOUR_ORG_ID>Requirements
- OAuth through Clerk; the older bearer mcp_ tokens are deprecated