Search skmtc…
⌘K
Docs
Sign in
Docs
History
Schema
Generate
Filter operations
AdminGate
2
Agents
62
AI Applications
31
AI Chat
87
get
/ai/applications/{public_id}/blueprint
The workspace's architecture blueprint with the live state of its parts.
put
/ai/applications/{public_id}/blueprint/bindings/{component_id}
Bind a part of the accepted blueprint to the project or service that realises it.
delete
/ai/applications/{public_id}/blueprint/bindings/{component_id}
Remove a part's binding. The project or service itself is untouched.
get
/ai/applications/{public_id}/blueprint/versions
Every version of the workspace's blueprint, newest first.
post
/ai/applications/{public_id}/blueprint/versions
Save a version the user wrote. It replaces the accepted version (and any open proposal) as it is saved.
post
/ai/applications/{public_id}/blueprint/versions/{version}/accept
Accept a proposed version: every chat of the workspace builds against it.
post
/ai/applications/{public_id}/blueprint/versions/{version}/reject
Reject a proposed version. The accepted version, if any, stays.
post
/ai/assistant/conversations
Start a chat with the Temps assistant.
get
/ai/conversations
List the current user's active conversations across all projects, most-recently-active first, annotated with project name/slug.
post
/ai/conversations
Create a private user-owned operator thread without binding its lifetime or authority to a project. Project access is selected per tool call and checked against the user's current role and memberships.
get
/ai/conversations/{public_id}
Load one bounded history page for a private conversation by its owner-facing id. Project/application context is revalidated, but never used as the ownership key.
patch
/ai/conversations/{public_id}
post
/ai/conversations/{public_id}/archive
post
/ai/conversations/{public_id}/attachment-uploads
Start uploading a file in chunks, for clients behind a small request-size limit. Send the chunks in order, then complete.
delete
/ai/conversations/{public_id}/attachment-uploads/{upload_id}
Abandon an upload and free what it reserved.
put
/ai/conversations/{public_id}/attachment-uploads/{upload_id}/chunk
Append the next chunk of an upload.
post
/ai/conversations/{public_id}/attachment-uploads/{upload_id}/complete
Finish an upload once every byte has arrived. The file is stored as a single-request upload would store it, and described the same way.
post
/ai/conversations/{public_id}/attachments
get
/ai/conversations/{public_id}/attachments/{attachment_id}
post
/ai/conversations/{public_id}/compact
Summarize the chat's older messages now. The full transcript stays on screen; only what the model is sent on later turns shrinks.
get
/ai/conversations/{public_id}/diagnostics
Export the bounded, server-redacted diagnostic state Temps currently owns.
get
/ai/conversations/{public_id}/goals
List the goals a chat is watching or has finished with, newest first.
post
/ai/conversations/{public_id}/goals/{goal_id}/cancel
Stop watching a goal. The check never runs again and no fix turn starts.
post
/ai/conversations/{public_id}/goals/{goal_id}/retry
Give a failed or waiting goal a fresh attempt budget and check it again now.
post
/ai/conversations/{public_id}/messages
Submit a turn to a user-owned conversation. The authenticated user's current role and permissions are captured for this turn's tool executor; no project id from the browser is trusted or required.
get
/ai/conversations/{public_id}/operations
The deployments, backups and restores this chat started, newest first, with their current state. Only what the caller may still read is listed: deployments of projects they can access, and backups and restores when they can read backups.
get
/ai/conversations/{public_id}/pending-actions
post
/ai/conversations/{public_id}/permission-mode
post
/ai/conversations/{public_id}/permissions/{permission_id}/resolve
get
/ai/conversations/{public_id}/resources
The projects, databases and services, repositories and sandbox this chat made, with their current state. Projects the caller can no longer read are left out, and so are services attached only to such projects.
post
/ai/conversations/{public_id}/restore
Restore a user-owned archived conversation.
get
/ai/conversations/{public_id}/sandbox
Whether the chat can use a sandbox, and which one it is linked to.
post
/ai/conversations/{public_id}/sandbox
Link a sandbox to the chat, or create one for it. Replaces any earlier link.
delete
/ai/conversations/{public_id}/sandbox
Unlink the chat's sandbox. The sandbox itself is left running.
get
/ai/conversations/{public_id}/sandbox/changes
The uncommitted Git changes of every repository under the sandbox work directory.
get
/ai/conversations/{public_id}/sandbox/changes/diff
Both sides of one changed file: as last committed, and as it is now.
post
/ai/conversations/{public_id}/sandbox/changes/track
Put a folder of the sandbox under Git, so its changes show from now on: a `.gitignore` for dependencies, builds and `.env` files when it has none, then a first commit of what is there. A folder already inside a repository is refused with `409`.
get
/ai/conversations/{public_id}/sandbox/file
One file of the chat's sandbox, for the viewer. Text is cut at 1 MiB; binary files come back without content.
get
/ai/conversations/{public_id}/sandbox/files
List one directory of the chat's sandbox: folders first, then files.
post
/ai/conversations/{public_id}/sandbox/projects
Let the chat's sandbox use a project's database.
delete
/ai/conversations/{public_id}/sandbox/projects/{project_id}
Stop the chat's sandbox using a project's database.
post
/ai/conversations/{public_id}/sandbox/services
Let the chat's sandbox use one service, whether or not it belongs to a project. The service must be one the sandbox can get credentials of its own on; the variables commands receive are the same a deployment gets.
delete
/ai/conversations/{public_id}/sandbox/services/{service_id}
Stop the chat's sandbox using one service. Its data is left as it is.
post
/ai/conversations/{public_id}/sandbox/wake
Start the chat's sandbox if it went to sleep, so the preview and the assistant's next command find it up. Does nothing when it is already running.
get
/ai/conversations/{public_id}/secrets
List the credentials this chat has asked for and not yet used up or lost.
put
/ai/conversations/{public_id}/secrets/{secret_id}
Provide the credential the assistant asked for.
post
/ai/conversations/{public_id}/secrets/{secret_id}/discard
Discard a credential request and anything already typed into it.
post
/ai/conversations/{public_id}/stop
get
/ai/conversations/{public_id}/suggestion-mode
A chat's next-step suggestion setting, with the default it falls back to.
put
/ai/conversations/{public_id}/suggestion-mode
Set a chat's own next-step suggestion mode, or clear it to follow the user's default.
post
/ai/conversations/{public_id}/suggestions
Suggest a few next steps after the chat's latest answer. The first request for an answer makes one small request to the chat's model, with only the last question and answer, and its cost counts toward the chat's spend. The result is saved on the answer, and later requests for it return the saved set at no cost.
get
/ai/conversations/{public_id}/usage
What this conversation has cost so far.
get
/ai/conversations/{public_id}/workspace
The workspace this chat is in, with its other chats.
put
/ai/conversations/{public_id}/workspace
Put this chat in a workspace, creating it when asked. The chats of a workspace share its sandbox, projects and architecture blueprint.
delete
/ai/conversations/{public_id}/workspace
Take this chat out of its workspace. It stops using the workspace's sandbox; the workspace and its other chats are unchanged.
get
/ai/pending-actions/{action_public_id}
post
/ai/pending-actions/{action_public_id}/confirm
post
/ai/pending-actions/{action_public_id}/reject
get
/ai/preferences/suggestions
The user's default for next-step suggestions.
put
/ai/preferences/suggestions
Set the user's default for every chat that does not set its own mode.
post
/ai/preview/authorize
Renew a managed preview only for an already-bound, running workspace. The requested sandbox ID is an object selector, never an authorization grant.
get
/ai/workspace
get
/ai/workspace-activity
Aggregate all active threads for the caller's global workspace and up to 100 requested, visible application workspaces. Thread history is not paged.
get
/ai/workspace/changes
get
/ai/workspace/diff
get
/ai/workspace/directory
get
/ai/workspace/file
get
/ai/workspace/file-limits
Get effective AI workspace file limits
get
/ai/workspace/file/download
Download one global AI workspace file
post
/ai/workspace/files
Upload files into the global AI workspace
post
/ai/workspace/preview-link
Mint a short-lived preview URL for a port in the user's shared global AI workspace. The gateway grant is identical to application previews; the only difference is how the owning sandbox is resolved.
get
/ai/workspace/terminal
Attach an interactive terminal to the user's global workspace
get
/projects/{project_id}/ai/conversations
Find the current user's existing chat for a context (returns `null` if none yet). Conversations are private even between members of the same project.
post
/projects/{project_id}/ai/conversations
Get-or-create the current user's private chat for a context.
get
/projects/{project_id}/ai/conversations/list
List the current user's active conversations for a project, most-recently-active first.
get
/projects/{project_id}/ai/conversations/{public_id}
One bounded conversation-history page (excluding internal context rows).
patch
/projects/{project_id}/ai/conversations/{public_id}
Rename a conversation (set its human-facing title).
post
/projects/{project_id}/ai/conversations/{public_id}/archive
Archive (soft-delete) a conversation.
post
/projects/{project_id}/ai/conversations/{public_id}/messages
Submit a user message and start a server-owned turn. The command returns as soon as the turn is durable; subscribe to the conversation WebSocket for real-time output.
get
/projects/{project_id}/ai/conversations/{public_id}/pending-actions
List all pending actions for a conversation (most-recently-proposed first).
post
/projects/{project_id}/ai/conversations/{public_id}/permission-mode
post
/projects/{project_id}/ai/conversations/{public_id}/permissions/{permission_id}/resolve
post
/projects/{project_id}/ai/conversations/{public_id}/stop
Explicitly cancel the server-owned active turn. Closing or refreshing a browser only detaches its stream; this endpoint is the sole UI cancellation path so execution lifetime is not coupled to connectivity.
get
/projects/{project_id}/ai/pending-actions/{action_public_id}
Get a single pending action by its public id (scoped to the project).
post
/projects/{project_id}/ai/pending-actions/{action_public_id}/confirm
Confirm a proposed AI action: validate permission, atomically claim, execute, persist outcome. The execution uses the CONFIRMING user's auth — never the model's.
post
/projects/{project_id}/ai/pending-actions/{action_public_id}/reject
Reject a proposed AI action (no execution). Status transitions to "rejected".
get
/projects/{project_id}/ai/readiness
Report which AI prerequisites this project satisfies.
AI Gateway
14
AI Gateway Admin
12
AI Gateway Limits
5
AI Gateway Pricing
1
AI Gateway Usage
8
AI Provider Status
3
Alarms
14
Alerts
6
Analytics
43
Analytics Ingest Keys
6
API Keys
9
Audit Logs
2
Authentication
29
Autofix
10
Autofixer
7
Backups
36
Blob
6
Blob Management
6
CI builder
3
CI runner
6
Cloud
8
Compose
3
Containers
8
Cron Monitors
6
Crons
3
Custom Domains
8
Dashboards
5
Deployment Tokens
6
Deployments
25
DNS
1
DNS Providers
14
DNS Records
4
Domains
19
Email Domains
12
Email Providers
9
Email Tracking
7
Email Validation
1
Emails
5
Environments
2
error-alert-rules
5
error-tracking
10
Events
13
External Images
7
External Services
50
External Services - Query
9
Feature Flags
9
feedback
4
Files
1
Funnels
7
GenAI
2
geo
2
Git Provider Connections
6
Git Providers
35
HTTP Checks
12
Imports
5
ingestion-health
1
Insights
1
Internal DNS
2
IP Access Control
6
KV Management
6
KV Store
7
Load Balancer
5
Logs
11
Metrics
17
Nodes
28
Notification Preferences
4
Notification Providers
15
Notification Routes
7
Observability
2
Operations
1
OTel
15
OTel Facets
4
OTel Ingest
6
Performance
6
Platform
6
Postgres Upgrades
7
Presets
2
Preview Gateway
6
Project Groups
22
Project IP Access
8
Projects
47
Proxy Logs
14
Public Repositories
6
releases
1
Repositories
6
Restore
6
Revenue
16
Runtime health
3
Sandboxes
40
Secrets
7
sentry-compat
8
sentry-ingestor
5
Settings
16
Snapshots
5
source-maps
8
Static Bundles
4
Status Page
32
Teams
13
Telemetry
2
Telemetry Logs
1
Telemetry Metrics
4
Templates
7
Traces
8
Traefik Discovery
6
Traffic Delivery
11
Users
13
Visitor Activity
5
Vulnerability Scans
8
Webhook Deliveries
3
Webhooks
6
Workflows
1
get
/projects/{project_id}/dsns
List all DSNs for a project
post
/projects/{project_id}/dsns
Create a new DSN for a project
post
/projects/{project_id}/dsns/get-or-create
Get or create DSN for a project/environment/deployment combination
post
/projects/{project_id}/dsns/{dsn_id}/regenerate
Regenerate DSN keys (rotate keys)
post
/projects/{project_id}/dsns/{dsn_id}/revoke
Revoke (deactivate) a DSN
AI Chat
Copy
Copy Markdown
Copy Markdown
Copying…
Copied
Copy failed
Changes to Set a chat's own next-step suggestion mode, or clear it to follow the user's default.
put
/ai/conversations/{public_id}/suggestion-mode
Oct 20, 2026
79089e9f49ee
Revision changes
info
endpoint added