turnstone console

Cluster /
SAVED SESSIONS
0 selected
WORKSTREAMS
STATE NAME MODEL NODE TASK TOKENS CTX
USERS
USERNAME DISPLAY NAME CREATED ACTIONS
Loading users...
API TOKENS
PREFIX NAME SCOPES CREATED EXPIRES ACTIONS
Select a user to view tokens
CHANNEL LINKS
CHANNEL EXTERNAL ID LINKED ACTIONS
Select a user to view channel links
SCHEDULED TASKS
NAME TYPE SCHEDULE TARGET NEXT RUN STATUS ACTIONS
Loading schedules...
WATCHES
NAME NODE COMMAND INTERVAL POLL CONDITION STATUS ACTIONS
Loading watches...
ROLES
NAME PERMISSIONS ACTIONS
Loading roles...
TOOL POLICIES
NAME PATTERN ACTION PRI STATUS ACTIONS
Loading policies...
PROMPTS
NAME TOOL GATE PRI STATUS ACTIONS
Loading prompt policies...
JUDGE
Loading settings...
Pattern rules for pre-execution intent validation
NAME TIER RISK TOOL REC. SOURCE STATUS ACTIONS
Loading rules...
Regex patterns for post-execution output scanning
NAME CATEGORY RISK FLAG SOURCE STATUS ACTIONS
Loading patterns...

Create Heuristic Rule

fnmatch syntax: bash, write_file, mcp__*
One regex per line. Each pattern is matched against the serialized tool arguments.
Rule confidence between 0.0 and 1.0. Higher values mean the rule is more certain when it matches.

Create Output Guard Pattern

Comma-separated regex compilation flags: IGNORECASE, MULTILINE, DOTALL.

Edit Heuristic Rule

fnmatch syntax: bash, write_file, mcp__*
One regex per line. Each pattern is matched against the serialized tool arguments.
Rule confidence between 0.0 and 1.0. Higher values mean the rule is more certain when it matches.

Edit Output Guard Pattern

Comma-separated regex compilation flags: IGNORECASE, MULTILINE, DOTALL.
SKILLS
CATEGORY NAME RISK ACTIONS
No skills configured
ⓘ Skills are community-published and not vetted by Turnstone.
Review source content before installing.
Search external registries to discover and install skills
USAGE
Loading usage data...
AUDIT LOG
TIME USER ACTION RESOURCE DETAIL
Loading audit log...
MEMORIES
NAME TYPE SCOPE DESCRIPTION UPDATED ACTIONS
Loading…
MODELS
Model definitions used by sessions across the cluster
ALIAS MODEL PROVIDER CTX WINDOW STATUS ACTIONS
Loading...
Per-role model assignments. Empty = use the default model.
Loading…
NODE METADATA
Loading…
SETTINGS docs
Loading…
TLS / CERTIFICATES
DOMAIN SANS ISSUED EXPIRES ACTIONS
Loading…
MCP
NAME TRANSPORT TOOLS RES PROMPTS STATUS ACTIONS
Loading...
ⓘ Servers are community-published via the official MCP Registry and not vetted by Turnstone.
Review source repositories before installing.
Search the official MCP Registry to discover and install servers

Import from GitHub

Paste a link to a repository or SKILL.md file

Create User

Create API Token

Token Created

Copy this token now. It will not be shown again.

Confirm

Link Channel Account

New Schedule

Schedule
min hour day month weekday
Execution

Edit Schedule

Schedule
Execution

Run History

Create Role

Permissions

Edit Role

Permissions

Assign Roles

Create Tool Policy

Edit Tool Policy

Create Prompt

Edit Prompt

Create Skill

Identity

Manifest

Environment requirements like other tools, services, or runtimes the skill expects. Max 500 chars.
Glob patterns gating model-initiated autoload, e.g. **/*.py. Comma-separated. Maps to SKILL.md paths:. Filter consumer pending.
Hides the skill from the user-facing /skill picker. The model can still load it via the skills tool. Maps to SKILL.md user-invocable: false.
Named positional slots substituted as $<name> in the skill body. Comma-separated. Maps to SKILL.md arguments:.
Display string shown next to slash-command autocomplete, e.g. [issue-number]. Maps to SKILL.md argument-hint:.

Deployment

How models discover this skill. Named requires explicit /skill invocation; Default is applied to every session; Search is BM25-discoverable.

Skill Content available: {{model}}, {{ws_id}}, {{node_id}}

Tip: paste a SKILL.md (with --- frontmatter) to auto-fill the form. Parsing SKILL.md…
Used
Runtime Config model, temperature, token limits
JSON array. Each: channel_type + channel_id or user_id
Resources bundled files (scripts, references, assets)

Edit Skill

Identity

Manifest

Environment requirements like other tools, services, or runtimes the skill expects. Max 500 chars.
Glob patterns gating model-initiated autoload, e.g. **/*.py. Comma-separated. Maps to SKILL.md paths:. Filter consumer pending.
Hides the skill from the user-facing /skill picker. The model can still load it via the skills tool. Maps to SKILL.md user-invocable: false.
Named positional slots substituted as $<name> in the skill body. Comma-separated. Maps to SKILL.md arguments:.
Display string shown next to slash-command autocomplete, e.g. [issue-number]. Maps to SKILL.md argument-hint:.

Deployment

How models discover this skill. Named requires explicit /skill invocation; Default is applied to every session; Search is BM25-discoverable.

Skill Content available: {{model}}, {{ws_id}}, {{node_id}}

Used
Runtime Config model, temperature, token limits
JSON array. Each: channel_type + channel_id or user_id
Security Scan
Resources bundled files for this skill

Memory Detail

Add MCP Server

Multitenant Authorization

Import MCP Config

Paste a JSON object with a mcpServers key (Claude Desktop / VS Code / Cursor format). Existing servers with the same name will be skipped.

MCP Server Detail

Install MCP Server

Add Model

Sampling Defaults
Server Compatibility

Delete Coordinators