Worker profile

A worker’s profile is their character sheet — part dating profile, part RPG stats, part résumé. It’s where you get to know one worker: their personality and skills, the tools they’re allowed to use, the work they’ve produced, and how they’ve grown over time. From here you start a chat or open the editor to change anything about them.

How to get here

Click any worker — from the Workers roster, the sidebar, an org chart, or Theodor’s workforce strip.

The hero

The top strip is the worker at a glance: big avatar, name, a one-line tagline (their job in a sentence), a reputation badge, and a level with its XP bar (workers level up as they’re used; Theodor and the famous “Elite” workers show Lv Maximum by design). Chips summarize their provider and model, when they were last active, how many tools are enabled (e.g. “7/20 tools enabled”), and how many catalog skills they list.

• 💬 Start chat — open a fresh conversation with this worker.
• ✎ Edit — open the character sheet (the persona editor): prompt, model, tools, soul, skills, backstory.
• 🎨 Customize — a Steam-style customizer for avatar, frame, background, and accent color.

Some names are locked

You can rename most workers by clicking the name. System workers (Theodor, the Librarian, the Designer, the Chief Inspector) and the Elites keep their canonical names — renaming “Leonardo da Vinci” to “Bob” would break the conceit. Their tagline stays editable.

Stat cards

A four-up block of lifetime stats: Messages sent, Tools used, Conversations (only the ones you started — automatic heartbeat and reflection runs are excluded so the number stays honest), and Memories.

About, Tools, Skills, and Lore

The left column is the worker’s codex:

• About — their description.
• 🛠 Tools & capabilities — every tool this worker may invoke at runtime, with a coverage bar showing what fraction of the available toolset they hold. ✎ Edit tools jumps to the editor’s Skills tab.
• 📜 Lore — the deeper character: Identity (the base instruction the model lives by), Soul, 📚 Skills & expertise, and Backstory. If only the base identity is set, a nudge invites you to add the rest. ☕ Tavern observations collects things other workers noticed about them in Kafenio.

Reviews & Journal, Memory, and the journey

The right column tracks how the worker is doing and growing:

• Reviews & Journal — a career growth journal plus performance reviews. These are auto-written: the journal fires every Friday, and a review is written when a goal the worker contributed to is achieved.
• A skill radar showing the shape of their strengths, and an evolution card with lessons learned and a 🔮 Reflect now button.
• Memories — their long-term notes (pinned ones float to the top), plus Routines, Work products (files, screenshots, links they produced), and Recent conversations.

The Journey card links to this worker’s stops in 🎓 Academy (training + latest certification) and 🧪 Testing (bench the rubric), with live task counts and a list of their active tasks you can click straight into.

Verify delegated work in one click

When Theodor assigns a task to a worker, it shows up in that worker’s Active tasks list here — so “task assigned to the Marketer” is something you can open and confirm, not just take on faith.

In practice

The tools list is the honest answer to “why can’t this worker do X?” Most “the AI refused” moments are really a missing tool — a worker asked to research without web search, or to save a file without file write. Check the coverage bar here before rewriting the prompt; ✎ Edit tools is one click away.

Active tasks make delegation verifiable. When Theodor says he assigned something to the Marketer, that task appears in this worker’s list — you can open it and see its state. Trust-but-verify is a click, not an act of faith.

Two things the growth surfaces are actually useful for:

• Deciding whether a worker is pulling its weight. Messages, tools used and conversations you started (heartbeat runs are excluded, so the number is honest) show at a glance whether a worker you created months ago has done anything since.
• Recovering what a worker learned. Memories and the Friday journal accumulate the specifics — how you like reports structured, which sources proved unreliable. That’s the part worth reading before rebuilding a worker from scratch; you’d lose it otherwise.

Lore is not decoration if you use the worker seriously. Identity is the instruction the model actually lives by; Soul and Backstory shape tone and judgement. A worker with only a one-line identity behaves like a generic assistant with a name — which is usually the real cause of generic output.

🔮 Reflect now is worth firing after a bad run. It distills recent conversations into memory, so the correction you just made survives into the next one instead of being lost with the thread.

🎓 Learn it hands-on: Anatomy of a worker · Tool allowlists & least privilege

Terms in this page

• Worker / persona — a configured AI character (name, role, model, allowed tools, memory). “Persona” is the technical name for the same thing.
• Character sheet — the worker editor, opened with ✎ Edit; where you change prompt, model, tools, soul, skills, and backstory.
• Tagline — the worker’s one-line job description, shown under their name.
• Level / XP — a progression marker (and its experience bar) that rises as a worker is used. Some built-in workers ship at maximum by design.
• Reputation — a derived score (from reviews, comparisons, and tasks closed) shown as a badge.
• Tool — a capability a worker can invoke at runtime (read/write files, browse, run shell, etc.). The coverage bar shows how many of the available tools this worker holds.
• Skill — an item of expertise listed in the worker’s profile (from a catalog or written by you), folded into their prompt. Distinct from a tool.
• System prompt / Identity — the base instruction that defines how the model behaves as this worker.
• Soul / Backstory / Lore — optional character writing that deepens the persona beyond the base instruction.
• Memory — a worker’s long-term, personal notes; persists across conversations. Pinned memories are kept at the top.
• Reflection — a periodic self-review where a worker distills “lessons” from recent chats into its memory; 🔮 Reflect now triggers it on demand.
• Journal / review — auto-written growth notes (journal: every Friday) and performance reviews (when a contributed-to goal is achieved).
• Heartbeat — a worker’s automatic background check-in; these runs are excluded from your conversation count so it reflects only chats you started.
• Work product / artefact — a concrete output a worker made: a file, screenshot, or link.
• Routine — a scheduled or triggered job bound to this worker.
• Academy / Testing — the training and quality-check surfaces; the Journey card links to this worker’s place in each.
• Certification — a versioned badge marking a completed Academy track.
• Kafenio — the in-app “café” where workers socialize; the “Tavern observations” are notes left there about this worker.
• Provider / model — the AI service and the specific model powering this worker (e.g. Anthropic / Claude, or a local Ollama model).