Available MCP Tools

List your projects, most recently updated first

Get a project by ID

Create a project (status defaults to backlog)

Update name, description, status, or starred state

Delete a project; rows that reference it are cascade-deleted

List files (id, path, language, updated_at — no content)

Get a file (including content) by ID

Get a file (including content) by project and path; errors if not found

Create a file. content must be a string (may be empty); language is inferred from the extension if omitted

Update a file's content and/or path

Delete a file by ID

Case-insensitive substring search over file contents; up to 10 matching lines per file

Replace the first exact occurrence of search in a file; errors if not found

Apply several write / search_replace / delete edits in order. For search_replace, content is the replacement text. Rollback on failure is best-effort (see note below)

List a file's versions, newest first, including content

Overwrite a file's content with a stored version

List workspaces in a project

Create a workspace

Update name, description, or status (backlog, in_progress, in_review, done, cancelled)

Delete a workspace

List chat messages, oldest first

Insert a chat message (source defaults to chat)

Run the CloudCode AI agent (tools enabled) on one prompt and return { response, tools_used, project_id }

Create an empty chat message to stream into; returns message_id

Replace the message content with the full accumulated text (not a delta)

Write the final content; returns status "complete"

List deployment records, newest first

Insert a deployment record only — does not build or deploy

List env vars (id, key, environment, is_secret, updated_at — values are not returned)

Create or update by key. On update, omitted is_secret / environment reset to false / development

Delete an environment variable

List custom domain records

Insert a custom domain record only — does not configure DNS

Delete a custom domain record

List your terminal sessions for a project

List project memories (max 100), newest first

Create or update a memory, matched by project + category + key

Delete a memory

List error history, newest first, including fixes

Record an error

Mark an error fixed and timestamp it

List session changelog entries, newest first

Add a session changelog entry

List the project activity log, newest first

Add an activity log entry

List project tags

Add a tag

Remove a tag

Get your profile

Update display name, bio, or theme

Submit a fix job (prompt or error report). Generates edits, verifies, pushes to GitHub as a PR or commit, optionally auto-merges, polls CI, re-fixes real CI failures up to max_fix_retries (0–3, default 1), checks the live deploy and auto-reverts broken deploys. Needs a GitHub repo (github_owner/github_repo or the project's linked repo).

Get the full job record: status, progress timeline, CI checks, verification, post-deploy status, result

List recent jobs, newest first

Cancel a queued or running job

Get verification_status, verification report, and progress_steps

Re-run vision regression against the project's latest ready deployment; sets post_deploy_status to live or broken

Revert the job's merge commit on target_branch; sets post_deploy_status to reverted

JSON array of { path, language } for every file in the project

Raw package.json content ("{}" if missing)

Raw README.md content (text/markdown, empty if missing)

Up to 100 project memories, most recently updated first

The 20 most recent error-history entries

Review a file for bugs, style issues, and improvements (embeds the file content)

Ask the model to read the relevant files and fix the error

Ask the model to read the code structure, then implement the feature

Refactor a file for structure, readability, and performance (embeds the file content)

Explain how a file works (embeds the file content)