Default Tenant

ID: default Status: Active

Plugin Configuration

Enable and configure plugins for this tenant. Settings persist when plugins are disabled.

Add Plugin

Add a plugin to this tenant. Multi-instance plugins (e.g., HTTP, MCP) require an instance name.

Email (SMTP) Enabled

Send emails via SMTP

Long-Term Memory Enabled

Save and recall information across sessions (preferences, decisions, facts)

Maximum memory entries per user (0 = unlimited)
Auto-expire memories after N days (0 = never expire)
Automatically include top memories in system prompts
How many memories to include in system prompts
Automatically extract memories from sessions using AI
AI model for memory extraction. Leave empty to auto-select a fast-tier model.

Knowledge Base Enabled

Upload documents to a searchable knowledge base for AI-augmented answers (RAG)

Maximum characters per text chunk for embedding
Character overlap between adjacent chunks for context continuity
Maximum number of chunks returned per search query
Maximum documents in the knowledge base (0 = unlimited)

Task Analytics Enabled

Ask natural language questions about task history, status, and trends

This plugin has no configurable settings.

Slack Enabled

Bidirectional Slack integration — receive messages and send responses

Bot user OAuth token from your Slack app
Found in Slack app Basic Information → App Credentials
For Socket Mode (future)
Slack team ID (e.g., T01234ABC). Used for tenant resolution.
Channel for proactive messages when no target is specified
Comma-separated channel IDs. Leave empty to allow all channels.

Claude Code Enabled

Provides Claude Agent SDK operations (plan generation, code implementation) via sidecar.

API key for Claude Agent SDK (usage-based billing)
Base URL of the Claude Agent SDK sidecar service (e.g. http://localhost:3100)
Claude model to use (e.g. claude-opus-4-6, claude-sonnet-4-6)

GitHub PM Enabled

Manages GitHub issues, PRs, and project status for agentic delivery workflows.

Fine-grained PAT or GitHub App token
GitHub Projects v2 number — tenant-wide default (0 = no board, issues only). Override per solution via 'project_number' in agent.yml.
HMAC secret for validating GitHub webhook payloads
Target branch for pull requests — tenant-wide default. Override per solution via 'pr_base_branch' in agent.yml.
Local path for issue workspaces (worktrees)
Parent directory containing solution repositories (e.g. D:\dev)
Max times Claude can retry after test failures — tenant-wide default (0 or empty = 3). Override per solution via 'max_impl_retries' in agent.yml.
Run AI review of plans before human review gate — tenant-wide default. Override per solution via 'auto_review_enabled' in agent.yml.
AI model for plan generation (sidecar). Leave empty to use the sidecar's default model. Override per solution via 'plan_model' in agent.yml.
AI model for auto-review of plans and PRs (sidecar). Leave empty to use the sidecar's default model. Override per solution via 'review_model' in agent.yml.
AI model for the readiness gate classifier. Leave empty to use the tenant default. Override per solution via 'readiness_model' in agent.yml.
Max times the agent can re-implement from PR review feedback — tenant-wide default (0 or empty = 2). Override per solution via 'max_pr_fix_iterations' in agent.yml.
Squash-merge the PR and delete branch when card moves to Done — tenant-wide default. Override per solution via 'auto_merge_enabled' in agent.yml.
Max Q&A rounds before proceeding with best-effort plan (0 = disabled, default 1). Override per solution via 'max_clarification_rounds' in agent.yml.

GitHub Repository Enabled

Creates pull requests on GitHub repositories.

Fine-grained PAT or GitHub App token. Leave blank to use the GitHubDelivery plugin's token.

AI Model Configuration

Configure AI providers and models for this tenant.

The model used when no specific model is requested.

Anthropic Configured

4 tier(s) enabled

Select which model tiers to make available. The underlying model ID can be changed in appsettings.json without affecting tenant configs.

Azure OpenAI No API Key

0 tier(s) enabled

Azure OpenAI resource endpoint (e.g., https://myinstance.openai.azure.com/)
Select which model tiers to make available. The underlying model ID can be changed in appsettings.json without affecting tenant configs.

OpenAI Configured

4 tier(s) enabled

Select which model tiers to make available. The underlying model ID can be changed in appsettings.json without affecting tenant configs.

Per-Model Overrides

Override temperature and max tokens for specific models. Leave blank to use defaults (temperature: 0.1, max tokens: 8192).

Model
Temp
FC Temp
Max Tokens
Anthropic Balanced (claude-sonnet-4-5-20250929)
Anthropic Best (claude-opus-4-6)
Anthropic Coding (claude-sonnet-4-5-20250929)
Anthropic Fast (claude-haiku-4-5-20251001)
OpenAI Balanced (gpt-5.1)
OpenAI Best (gpt-5.2)
OpenAI Coding (gpt-5.2-codex)
OpenAI Fast (gpt-5-mini)

Task Execution Limits

Control how many AI iterations and concurrent tasks this tenant can run.

Maximum AI-to-tool-call rounds before a task is stopped (1–100).
Maximum background tasks running simultaneously. Additional tasks queue until a slot opens (1–50).

Branding

Customize the app name, logo, colors, and favicon for this tenant. Changes take effect on the next page load.

Shown in the browser title bar and sidebar. Default: "Merri".
Sidebar logo image URL (max height 36 px). Leave blank for the default logo.
Browser tab icon URL (.ico or .png). Leave blank for the default favicon.
Overrides --color-primary: buttons, links, active nav items.
Overrides --color-primary-dark: hover states and darker color variants.
Live Preview
A

Advanced Configuration

Raw JSON editor for the full tenant configuration including AI models, quotas, and features.