Cody
Cody is an AI-powered assistant that uses the magic of ChatGPT meets the expertise of a custom-made AI assistant trained on your specific business.
Verdict
Common use cases
- Onboard new hires with company wiki bot
- Answer policy questions from HR docs
- Surface past customer support threads
- Upload meeting notes to team knowledge base
- Organize training materials into folders
Integration
- Vendor
- Cody
- Category
- other
- Auth
- API_KEY
- Tools
- 21
- Composio slug
cody
Tools
- Create Conversation
Tool to create a new conversation with a specified bot. Use when starting a new conversation thread with optional focus mode to limit bot's knowledge base to specific documents.
- Create Document
Tool to create a new document with text or HTML content in Cody AI. Use when you need to add documents to Cody's knowledge base with up to 768 KB of content.
- Create Document From File
Tool to create a document by uploading a file (up to 100 MB). Supports txt, md, rtf, pdf, ppt, pptx, pptm, doc, docx, docm formats. Use when you need to add file-based documents to Cody's knowledge base. The file is processed asynchronously
- Create Document from Webpage
Tool to create a document from a publicly accessible webpage URL. Use when you need to import content from a webpage into Cody AI. The webpage must be accessible without login. If request fails, ensure the URL is publicly accessible and not
- Create Folder
Tool to create a new folder in Cody AI for organizing content. Use when you need to create a folder to organize documents or conversations.
- Delete Conversationdestructive
Tool to delete a conversation by its ID. Use when you need to permanently remove a conversation from the system.
- Delete Documentdestructive
Tool to delete a document by id. Use when removing a document that is no longer needed.
- Get Conversation
Tool to fetch a conversation by its ID from Cody AI. Use when you need to retrieve details about a specific conversation. Supports optional includes parameter to filter response to list document IDs.
- Get Document
Tool to retrieve a specific document by its identifier from Cody AI. Use when you need to get details about a particular document including its status, content URL, and metadata.
- Get Folder
Tool to retrieve a specific folder by its identifier. Use when you need to get details about a folder.
- Get Message
Tool to fetch a specific message by its ID from Cody AI. Use when you need to retrieve details about a particular message, with optional includes for sources or usage metrics.
- Get Upload Signed URL
Tool to get an AWS S3 signed upload URL for file uploads. Use when you need to obtain a signed URL to upload a file to Cody's storage.
- List Bots
Tool to get all bots with optional keyword filtering. Use when you need to retrieve the list of available bots in a Cody account.
- List Conversations
Tool to get all conversations with optional filtering by bot, keyword, or includes. Use when you need to retrieve conversation history, filter by bot, search by name, or get document associations.
- List Documents
Tool to retrieve all documents from Cody AI account with optional filtering. Use when you need to list documents by folder, conversation, or search by keyword. Returns document details including learning status and content URL.
- List Folders
Tool to retrieve all folders with optional keyword filtering. Use when you need to list or search for folders in the account.
- List Messages
Tool to retrieve a paginated list of messages from Cody, optionally filtered by conversation. Use when you need to list messages, with optional filtering by conversation_id and extra attributes (sources or usage).
- Send Message
Tool to send a message to Cody AI and receive an AI-generated response. Use when you need to send a user message to a conversation and get the AI's reply.
- Send Message for Stream
Tool to send a message to Cody AI and receive a Server-Sent Events (SSE) stream URL for the AI response. Use when you need streaming responses instead of waiting for the complete message. The response contains a stream_url that can be used
- Update Conversation
Tool to update a conversation by its ID including name, bot_id, and document_ids. Use when you need to modify an existing conversation's properties.
- Update Folder
Tool to update a folder by its ID. Use when you need to modify an existing folder's name.
Setup
Setup guide
- 11. Log into your Cody account at cody.ai and navigate to Settings > API Keys. 2. Generate a new API key and copy it to your clipboard. 3. In Switchy, open the Space where you want to use Cody and click Integrations in the sidebar. 4. Find Cody in the list and click Connect. 5. Paste your API key into the prompt and click Authorize. 6. Switchy will confirm the connection and list the bots available in your Cody workspace. 7. To test, type '@Cody create a conversation with [bot name]' in the Space chat, replacing [bot name] with one of your configured bots. 8. If the conversation starts successfully, the integration is working. 9. You can now upload documents, retrieve past conversations, and organize content into folders by @mentioning Cody with the appropriate command.
What teammates see: by default, memories from Cody are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Start Onboarding Chat
@Cody create a conversation with the Onboarding bot. I need to walk a new engineer through our deployment process and code review standards.Open in a Space →
Upload Meeting Notes
@Cody create a document from the text in this message. Title it 'Sprint 12 Standup Notes' and include the blockers and action items we discussed.Open in a Space →
Import External Guide
@Cody create a document from this webpage: https://docs.example.com/api/v2/authentication. We need our support bot to reference their auth flow.Open in a Space →
Retrieve Past Conversation
@Cody get the conversation with ID abc123. I want to see what answers the compliance bot gave Sarah about data retention.Open in a Space →
Organize Knowledge Base
@Cody create a folder called 'Q1 Product Specs'. We'll move all the feature docs into it so the product bot can focus its answers.Open in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Cody pull the last 5 messages from our 'Q1 Planning' conversation and summarize the key decisions we made about the product roadmap.
Based on the Q1 Planning conversation (ID: conv_abc123), the team made three key decisions: (1) prioritize mobile app performance improvements over new features in January, (2) allocate 40% of engineering bandwidth to technical debt reduction, and (3) push the analytics dashboard launch to late February to ensure proper testing. The conversation included 8 participants and spanned 47 messages over 3 days. Sarah from Product emphasized the performance metrics showing 23% drop-off on mobile checkout flows as the primary driver for this prioritization.
This example shows Cody's conversation retrieval paired with AI synthesis. Requires the conversation ID or name to be known beforehand. The MCP fetches raw conversation data; the AI layer interprets and summarizes it. Useful for extracting decisions from long threads without re-reading everything manually.
@Cody create a new document called 'API Migration Guide' from our internal wiki page at https://wiki.company.com/api-v2-migration and add it to the Engineering folder.
Document created successfully. 'API Migration Guide' (ID: doc_xyz789) has been imported from the provided URL and added to the Engineering folder. The document contains approximately 12,500 words covering authentication changes, endpoint deprecations, and code examples. This content is now available in Cody's knowledge base for the team to reference in conversations. Note: The webpage import captured the main content but excluded navigation elements and sidebar widgets.
Demonstrates Cody's webpage-to-document ingestion capability. The source URL must be publicly accessible without authentication—internal wikis behind SSO will fail unless they have a public share link. This is a write action that permanently adds content to your Cody workspace, so verify the URL is correct before running.
@Cody start a conversation with the 'Product Spec Bot' in focus mode limited to documents in the 'Q1 Specs' folder, asking: What are the unresolved questions across all three feature specs?
Conversation started (ID: conv_def456) with Product Spec Bot in focus mode. The bot identified 7 unresolved questions across the three Q1 specs: (1) Mobile push notification frequency limits not defined in the Notifications spec, (2) Data retention policy unclear for archived user content in the Storage spec, (3) Third-party API rate limit handling strategy missing from the Integrations spec, plus 4 additional items related to edge cases and error states. The bot recommends scheduling a 30-minute sync with Engineering and Legal to close these gaps before implementation begins.
Shows Cody's conversation creation with focus mode constraints, which scope the bot's knowledge to specific folders. This prevents the bot from pulling irrelevant context and keeps answers targeted. The conversation persists in your workspace—you can continue the thread or share the ID with teammates to collaborate on resolving the questions.
Use-case deep-dives
When Cody wins for support teams ingesting scattered docs
A 6-person support team fields 200 tickets a week across Zendesk, Notion, and Google Drive. They need one place to query product docs, troubleshooting guides, and past resolutions without switching tabs. Cody's document-creation tools (file upload, webpage import, raw text) let you funnel everything into a single knowledge base, then spin up conversations that search across it. The 100 MB file limit and 21-tool scope mean you can automate the sync from Slack threads or Confluence exports. The trade-off: if your docs change hourly, you're re-uploading constantly—Cody doesn't watch external sources for updates. If your knowledge is stable week-to-week and you want a bot that answers from your own corpus, not the public web, this MCP is the right call.
When Cody fits a 2-week contractor ramp
You hire a freelance designer for a sprint and need them fluent in your brand guidelines, component library, and past campaign briefs by day three. Cody's folder and document tools let you bundle PDFs, Figma export markdown, and style decks into a single bot conversation. The contractor asks questions in natural language instead of hunting through a 40-page brand book. Focus mode keeps the bot from hallucinating answers outside the uploaded docs. The boundary: if your onboarding materials live in a private Notion workspace or behind SSO, you'll export them manually—Cody doesn't pull from authenticated sources. For short-term hires who need a curated knowledge snapshot, not live system access, this MCP delivers fast ramp without IT overhead.
When Cody streamlines leadership briefing assembly
A 3-person exec team runs Friday all-hands and needs a digest of the week's wins, blockers, and metrics from Linear, Slack, and Google Sheets. They create a Cody conversation each Thursday, upload the week's sprint report PDF, paste Slack highlights as text documents, and import the metrics dashboard as a webpage. The bot synthesizes talking points and flags gaps. The 768 KB text limit per document means you can't dump a 5 MB transcript, but most weekly summaries fit comfortably. The trade-off: Cody doesn't auto-refresh the conversation when new data lands—you're manually feeding it each week. If your all-hands cadence is predictable and your sources are exportable, this MCP turns scattered updates into a single briefing thread without custom integrations.
Frequently asked
What does the Cody MCP let me do in Switchy?
It connects your Cody AI workspace so you can create conversations with your bots, upload documents to the knowledge base, and manage folders — all from Switchy's chat interface. You can start a conversation thread, feed it a PDF or webpage, and retrieve the bot's response without switching tabs. Useful when you want Cody's retrieval inside a broader workflow.
Do I need admin access to connect Cody?
You need an API key from your Cody workspace, which typically requires workspace owner or admin permissions to generate. The key grants full read-write access to conversations, documents, and folders, so treat it like a password. If you're on a team plan, confirm with your Cody admin before connecting it to Switchy.
Can the Cody MCP train my bot or change its settings?
No. It can create conversations with existing bots and add documents to the knowledge base, but it can't modify bot configurations, adjust prompts, or retrain models. If you need to change how a bot behaves, you still do that in the Cody web app. The MCP is for runtime use — feeding data in and pulling answers out.
Why use this instead of just opening Cody in a browser?
The MCP makes sense when you're chaining Cody with other tools in Switchy — for example, scraping a webpage with Firecrawl, uploading it to Cody as a document, then asking a bot a question about it. If you only use Cody standalone, the web app is simpler. The MCP is for automation and multi-tool workflows.
Who on the team should connect the Cody integration?
Whoever owns the Cody workspace API key. That person's key determines which bots and documents Switchy can access. If multiple people need to use Cody in Switchy, they can share the same connection or each connect their own workspace. The MCP doesn't consume Cody message credits separately — usage counts against your Cody plan as normal.