Create & configure

What this is for. The Configuration tab is the SQL-level identity of an agent: name, role, model, effort, tools, constraints, and monthly token budget. Workspace files and voice profiles are separate tabs. This is what you fill in when you create someone, and what you change when their job shifts.

When to use it

You are creating an agent and need a name, type, model, and tool list.
You want a coding agent to get read_file / edit_file / grep without depending on a CLI.
A monthly token cap should stop spend, or you need to clear it (0 = unlimited).
Inbound Telegram (or another channel) should reach this agent.
You want the prompt coach to tighten the system prompt — not voice, not project domain.

Typical workflow

Open Agents → click the agent (or Create Agent) — route /agents/:id, tab Configuration.
Fill Name, Role, Tier, Agent Type, Model (or Auto (routing decides)), Tools, Constraints.
Set Monthly Token Budget if you want a cap. Bind a channel on the Channels tab if inbound should land here.
Save Changes. A new conversation assigned to this agent should use this model, tool list, and prompt.

Features

Also shows Token Budget summary and Executing… when a run is active.

Classification

Field	Meaning
Tier	Primary / Team / Specialist (see overview)
Agent Type	Assistant, Engineer, Developer, …

Persona block

Field	Meaning
Name	Display name
Role	Short role line
Description	Longer description
Persona	Persona summary
Goal	What drives decisions (What drives this agent’s decisions)
Backstory	Context that shapes approach (…perspective)
Avatar	Emoji (or image) shown in UI
System Prompt	Agent-level instructions (combined with layered prompts)
Prompt coach	AI coach for the system prompt (operating protocol only — not voice, not project domain) — Prompts

Model & effort

Field	Meaning
Model	Concrete model id, or Auto (routing decides)
Reset to auto	Clear override → routing
Effort	Auto / Low / Medium / High / Max
Effort hint	Higher = deeper reasoning, slower, more expensive
Max Turns	Hard cap on agent loop turns per run

Tools & constraints

Field	Meaning
Tools (comma-separated)	Tool names this agent may call
Capabilities (comma-separated)	Capability tags (e.g. `research, coding`)
Constraints (one per line)	Hard rules (e.g. no destructive ops)

Coding agents (model-agnostic surface)

For implement/fix/review work, grant the first-class file tools so any model (not only Claude Code) can edit without shell:

read_file, write_file, edit_file, grep, glob, git_status, git_diff, run_command, search_indexed, list_search_sources

Tool	Use
`read_file` / `edit_file` / `write_file`	Read and targeted edit under workspace/worktree
`grep` / `glob`	Find symbols and files
`git_status` / `git_diff`	Review helpers (read-only)
`run_command`	Tests/lint (red tier — approval / autonomy)

Existing agents created before 0.8.6 do not auto-pick up new tools — add them here (or re-seed from an updated template). Full catalogue: Tools.

Budget

Field	Meaning
Monthly Token Budget	Cap for the month; `0` = unlimited
Token usage display	Used vs budget on list/detail

Actions

Control	Meaning
Save Changes	Persist configuration

Memories tab (read-only list)

Element	Meaning
Episodic / Working	Memory tier labels
N memories	Count
salience	Importance score
accessed N×	Access count
Empty hint	Fills as the agent works

Channels tab (summary)

Bind channel instances so inbound messages reach this agent. Full field list: Channels overview and agent channels UI:

Control	Meaning
Bind a channel instance	Pick existing Telegram/… instance
Bind to this agent	Attach
Unbind	Detach
Status Connected / Error / Credentials set / Not configured	Instance health
Mode Autonomous	Channel may drive autonomous handling