Create and manage Bots
Configure individual bot profiles tailored to specific responsibilities. From code auditors and QA runners to customer support triage agents, each bot maintains its own personality, memory storage, and tool privileges.
Bot configuration schema
Bots can be defined via the web console or saved into your Git repository as declarative manifests:
{
"name": "pr-auditor",
"model": "plannix-3.6",
"system_prompt": "You are a senior security engineer reviewing pull requests.",
"compute_tier": "standard",
"tools": ["git", "semgrep", "github_pr"],
"auto_approve": ["git clone", "git diff"]
}Custom system prompts and personas
Direct your bot with clear operational boundaries. Define formatting rules (e.g. VS Code-style syntax, no emojis, strict markdown tables) and specify domain knowledge libraries.
Memory and state persistence
Choose between three retention models:
| Retention Mode | Persistence | Typical Application |
|---|---|---|
| Ephemeral | Erased when task completes | Isolated code reviews, scratchpad scripts |
| Workspace Bound | Persisted across project turns | Project development assistant, repo context |
| Long-Term Team | Synchronized across all team bots | Knowledge base bot, corporate wiki keeper |
Binding tools and MCP servers
Equip your bot with Model Context Protocol (MCP) servers or internal CLI binaries. Each tool can be configured with discrete execution permissions and audit logging.
Managing bot lifecycles
Start, pause, duplicate, or retire bots with zero downtime. Workspace administrators can export full bot configurations and migrate them between enterprise environments.