mcp server — v0.1.0

your team’s living memory.

every ai agent in your organisation shares one source of truth — a git repo. no database. no hosted service. just your team’s context in markdown.

14 mcp tools
pr-based write flow
1 env var needed

context your ai agents can actually read

fork the repo, fill in your team’s context, and connect any mcp-compatible ai tool.

1

fork the repo

copy organisation.md into your github org. the template includes identity, team, projects, decisions, and preferences sections.

2

write your context

edit the markdown files with your org’s actual data — who does what, which projects are active, past decisions.

3

run the server

start with npx @shashank-sn/organisation-md. your ai agents connect over stdio and read/write context naturally.

agent memory as reviewable diffs

every context change goes through a pull request. no silent writes into agent state.

1

add a decision

an agent proposes a context update — becomes a pr.

2

codeowners review

teammates discuss, approve, or request changes.

3

merge

approved changes become permanent context.

4

next agent reads it

every agent that starts sees the updated context.

14 mcp tools, one purpose

read, search, and propose updates to your organisation’s context — all through pull requests.

📄
read_org

read the full organisation.md file from the github repo.

🔍
read_section

read a specific section by heading (e.g. team, decisions).

✏️
update_section

propose an update to any section — creates a pull request.

🔎
search_context

search across organisation.md and all context files.

📝
propose_change

propose a change to any tracked file — creates a pr.

📁
list_context_files

list all files in the context directory.

📥
import_file

import a txt, md, or docx file into the knowledge base.

add_info

add info via natural language — auto-detects the section.

remove_info

remove info via natural language — auto-detects what to remove.

🛡️
check_roles

check which users can modify which files.

🔑
check_permissions

check a user’s permissions across the repo.

📋
configure_codeowners

set up codeowners for path-level access control.

🐛
report_bug

auto-detect the area and file a github issue.

💡
suggest_feature

suggest a feature — creates a github issue automatically.

from zero to running in 30 seconds

one environment variable. one npx command. no database setup, no deployment pipeline.

# generate a pat and export it export github_token=ghp_your_token export github_owner=your-org export github_repo=organisation.md # run with npx npx @shashank-sn/organisation-md
# or clone and run git clone https://github.com/your-org/organisation.md cd organisation.md npm install # start the server npx tsx src/server.ts
# add to your .mcp.json { "mcpServers": { "organisation.md": { "command": "npx", "args": ["@shashank-sn/organisation-md"], "env": { "github_token": "ghp_...", "github_owner": "your-org", "github_repo": "organisation.md" } } } }
# connect and read your context read_org → your full organisation.md read_section team → your team section search_context "decisions" → matching entries update_section projects → creates a pr
need help? see the full quickstart guide, or check the example workflows.

bring your own files

upload folders of .txt, .md, or .docx files. the server parses them, creates context entries, and updates organisation.md as your navigation index.

📂

folder upload

point the server at a folder of files. every .txt, .md, and .docx file becomes a context entry under context/.

.txt .md .docx
💬

natural language crud

add, edit, or remove information by speaking naturally. the server parses your intent and updates the right files.

ai-powered pr-based
🔑

role-based access

control who can add, delete, or approve changes through git itself — codeowners, branch protection, and commit signing.

git-native team-ready
organisation.md/
organisation.md # your navigation index
context/
projects.md
architecture.md
people.md
+ your uploaded files here
src/
docs/

everything in one place

the repo structure is designed to grow with your team. start with the template, add context files as you go.

organisation.md

canonical context file — identity, mission, team, projects, decisions, and more.

context/

supporting docs for projects, architecture, people, and anything else that deserves its own file.

src/

typescript mcp server with octokit integration, markdown parser, and tool implementations.

docs/

quickstart guide, agent prompt template, example flows, and product strategy.