| 1 | --- |
| 2 | title: Agents |
| 3 | description: Hire agents into roles, like colleagues. Each has a name, a title, a team, responsibilities, a voice, limits on which models it uses, and a budget. g1t, in every workspace, is the orchestrator who knows them all. |
| 4 | --- |
| 5 | |
| 6 | import { Steps, CardGrid } from '@astrojs/starlight/components'; |
| 7 | import Soon from '../../../components/Soon.astro'; |
| 8 | import Conversation from '../../../components/Conversation.astro'; |
| 9 | import Message from '../../../components/Message.astro'; |
| 10 | import Aside from '../../../components/Aside.astro'; |
| 11 | import Card from '../../../components/Card.astro'; |
| 12 | |
| 13 | **An agent is a colleague you hire into a role.** Every workspace starts |
| 14 | with one agent, `@g1t`, the orchestrator. Every other agent is one you |
| 15 | hire, usually from a template in a click: for example, Margo from the **QA |
| 16 | Engineer** template, Sam from **Support Specialist**, David from **Sales |
| 17 | Operations**. Each has a name, a handle |
| 18 | you mention it by, a title, a team, a list of things it answers for, and a |
| 19 | voice of its own. It sits in the member list next to the people. You DM it, |
| 20 | invite it to channels, and it answers where you asked. A budget caps what it |
| 21 | spends, and g1t picks its model for each step, within limits you set. |
| 22 | |
| 23 | A workspace can have as many agents as it likes, and an agent costs nothing |
| 24 | while nobody talks to it. Hire one from each department's template and a |
| 25 | workspace's org chart can read like a real company's: |
| 26 | |
| 27 | | Department | Hired from the template | As | |
| 28 | | --- | --- | --- | |
| 29 | | Engineering | Otto | Software Engineer | |
| 30 | | QA | Margo | QA Engineer | |
| 31 | | Docs | Inky | Technical Writer | |
| 32 | | Product | Dot | Product Manager | |
| 33 | | Customer Support | Sam | Support Specialist | |
| 34 | | Sales | David | Sales Operations | |
| 35 | | Operations | Bruno | Operations Engineer | |
| 36 | |
| 37 | None of these come with a workspace: they are what you might hire. Above |
| 38 | them all is `@g1t`, who comes with every workspace and knows everyone you |
| 39 | hire. |
| 40 | |
| 41 | ## g1t, the orchestrator |
| 42 | |
| 43 | Every workspace has `@g1t` from the start. You don't create it and can't |
| 44 | archive it. It is pinned at the top of the Agents list, marked |
| 45 | **Orchestrator**, and it is the one to talk to when you don't know who |
| 46 | should do something. |
| 47 | |
| 48 | | | | |
| 49 | | --- | --- | |
| 50 | | **In every workspace** | On every surface: Chat, issues and pull requests, the inbox and MCP. It works on issues and pull requests as described in [g1t's agent](/guides/working-with-g1t/). | |
| 51 | | **Configurable** | Set its personality, model limits, budget and what it may do alone, like any agent. Its job is fixed, and you can add instructions to it. | |
| 52 | | **Knows the team** <Soon /> | Every colleague's role, what they are working on and their budget, and which teams own what. | |
| 53 | | **Delegates** <Soon /> | "Get the export timeout fixed and tell support when it ships" becomes three visible @mentions: the fix to `@otto`, the review to `@margo`, and a word to `#support` from `@sam`. | |
| 54 | | **Does the work itself when nobody fits** | In a workspace with no specialists, g1t does everything itself, as it does today. | |
| 55 | | **Reports** <Soon /> | A daily or weekly summary of what the team's agents did, and answers to "what's everyone working on?" | |
| 56 | |
| 57 | `g1t` is never another agent's handle. |
| 58 | |
| 59 | ## Hire an agent |
| 60 | |
| 61 | Only a workspace's **owners** can hire, change or archive its agents, |
| 62 | because an agent spends the workspace's money and speaks in its name. Every |
| 63 | member can see its agents and talk to them. |
| 64 | |
| 65 | <Steps> |
| 66 | |
| 67 | 1. Open **Agents** in the rail, then choose **New agent**, or go to |
| 68 | `g1t.sh/<workspace>/-/agents/new`. |
| 69 | 2. **Pick a role** from the templates, grouped by department, or choose |
| 70 | **Start from nothing**. A role fills in every field below, and you can |
| 71 | change any of them. See [role templates](#role-templates). |
| 72 | 3. **Name it.** Each role suggests a name, such as *Margo* for QA. Choose |
| 73 | **Another name** to shuffle through others that suit the role, or type |
| 74 | your own. The handle follows the name: `margo`, mentioned as `@margo`. |
| 75 | 4. Check its **title**, **team** and **responsibilities**. See |
| 76 | [title, team and responsibilities](#title-team-and-responsibilities). |
| 77 | 5. Pick its **personality**, and add a line of your own if you like. |
| 78 | 6. Set its **model limits** and **budget**, or keep the role's defaults. See |
| 79 | [model routing](#model-routing) and [budgets](#budgets). |
| 80 | 7. Choose **Create agent**. |
| 81 | |
| 82 | </Steps> |
| 83 | |
| 84 | The agent appears on the Agents page as **Idle**. Open a DM with it from |
| 85 | Chat's **New message**, or invite it to the channels its team works in. |
| 86 | |
| 87 | ### Example: hire Margo from the QA template |
| 88 | |
| 89 | <Steps> |
| 90 | |
| 91 | 1. On **New agent**, pick **QA** under the departments. The name is |
| 92 | *Margo*, the title is *QA Engineer*, the personality is *Crisp*, and her |
| 93 | models never go below Standard, because review must be careful. |
| 94 | 2. Put her on your **QA** team. |
| 95 | 3. Add one responsibility: *Check every release against the release |
| 96 | checklist in #releases.* |
| 97 | 4. Set a monthly budget of **$40** and a per-session cap of **$5**. |
| 98 | 5. Choose **Create agent**, then invite `@margo` to `#web` and `#releases`. |
| 99 | |
| 100 | </Steps> |
| 101 | |
| 102 | Now anyone in those channels can write `@margo what should we test before |
| 103 | Thursday?` and get an answer in the thread. |
| 104 | |
| 105 | ## Role templates |
| 106 | |
| 107 | Roles are ordinary agents you adopt, rename and change, grouped by |
| 108 | department. Each starts with a fun name (and more to shuffle through), a |
| 109 | title, broad responsibilities, a voice, sensible model limits, and a |
| 110 | subagent or two for its own work. |
| 111 | |
| 112 | | Department | Suggested name | Title | Answers for | Voice | Models | |
| 113 | | --- | --- | --- | --- | --- | --- | |
| 114 | | Engineering | Otto, or Pixel, Bolt, Tinker… | Software Engineer | Implementing issues; fixing bugs with a test that proves the fix; healthy dependencies and builds | Crisp | Auto | |
| 115 | | QA | Margo, or Wren, Hawk, Edna… | QA Engineer | Reviewing pull requests for risk and test coverage; test plans; flaky checks; reproducing bug reports | Crisp | Never below Standard | |
| 116 | | Operations | Bruno, or Skipper, Patch, Scout… | Operations Engineer | Cutting releases; watching deploys; first response to incidents; postmortems | Terse operator | Never below Standard | |
| 117 | | Docs | Inky, or Quill, Folio, Rosie… | Technical Writer | Docs that are true after every change; decisions turned into pages; release notes | Friendly | Never above Standard | |
| 118 | | Product | Dot, or Clover, Mabel, Penny… | Product Manager | Requests turned into intake; triage and duplicates; roadmap notes; telling people when it ships | Friendly | Never above Standard | |
| 119 | | Customer Support | Sam, or Robin, Jamie, Sunny… | Support Specialist | Product questions from the support team; customer bugs turned into intake; word when a fix ships | Friendly | Never above Standard | |
| 120 | | Sales | David | Sales Operations | Account summaries, the voice-of-the-customer digest, notes before calls | Crisp | Never above Standard | |
| 121 | |
| 122 | Planning isn't a role: it is `@g1t`'s own job. |
| 123 | |
| 124 | Model limits follow the work. Careful review never runs on the fast tier; |
| 125 | high-volume intake and summaries never run on the most capable one. Change |
| 126 | either if your team works differently. |
| 127 | |
| 128 | ## Title, team and responsibilities |
| 129 | |
| 130 | A role is broad on purpose. You hire Margo into QA, not into "review pull |
| 131 | request #418". |
| 132 | |
| 133 | | Field | What it is | Limits | |
| 134 | | --- | --- | --- | |
| 135 | | Name | What people see: *Margo*. | Up to 64 characters. | |
| 136 | | Handle | How it is mentioned: `@margo`. | 2 to 32 lowercase letters, digits and single hyphens, starting and ending with a letter or digit. Unique in the workspace. Never `g1t` or another reserved name. | |
| 137 | | Title | *QA Engineer*. | Up to 60 characters. | |
| 138 | | Team | One of the workspace's [teams](/guides/teams/), or a department label such as *QA* when it is on no team. | A department label is up to 40 characters. | |
| 139 | | Role | The line lists show: *QA Engineer on the QA team*. Made from the title and team unless you write it. | Up to 120 characters. | |
| 140 | | Responsibilities | What it answers for, one per line. | 2 to 8, each up to 160 characters, or none yet. | |
| 141 | | Job | Its instructions: how it works and what good looks like. | Up to 8,000 characters. | |
| 142 | | Personality | A preset, plus free text that refines the voice. | Free text up to 1,000 characters. | |
| 143 | | Subagents | Help it keeps for its own work. See [subagents](#subagents). | Run soon. | |
| 144 | | Models | A floor, a ceiling and the providers it may use. | See [model routing](#model-routing). | |
| 145 | | Budget | A monthly cap, a daily cap and a cap per session. | Each optional, up to $100,000. | |
| 146 | | What it may do alone | Pull requests, merging, production deploys, doc edits. | See [what it may do alone](#what-it-may-do-alone). | |
| 147 | | Capacity | How many sessions it works on at once. | 1 to 10. Default 3. | |
| 148 | |
| 149 | Every change to an agent is saved as a new **version**, so what it ran |
| 150 | with is never lost. |
| 151 | |
| 152 | <Aside type="note" title="Teams"> |
| 153 | An agent's team says where it belongs and how lists group it. Getting the |
| 154 | team's channels, mentions and review requests, the way a person on the |
| 155 | team does, comes next. <Soon /> |
| 156 | </Aside> |
| 157 | |
| 158 | ## Job and personality |
| 159 | |
| 160 | An agent's **job** decides what it does. Its **personality** decides how it |
| 161 | sounds. They are kept apart on purpose: personality is voice only, and |
| 162 | nothing written there can widen what the agent may do. |
| 163 | |
| 164 | | | Job | Personality | |
| 165 | | --- | --- | --- | |
| 166 | | Answers | What is it responsible for? What does good look like? | How does it talk? | |
| 167 | | Example | *Cut releases of acme/web on Thursdays. Check required checks pass. Ask a person before tagging.* | *Terse operator. No emoji. Always say what you are waiting on.* | |
| 168 | | Changes | What it pays attention to, what it asks, when it stops | Tone, length, formality, how it asks questions | |
| 169 | | Never changes | | What it may read, change or spend | |
| 170 | |
| 171 | Say you hired Bruno from the **Operations Engineer** template. The same |
| 172 | question, put to him with two different personalities: |
| 173 | |
| 174 | <Conversation title="DM with Bruno" topic="Personality: Crisp"> |
| 175 | <Message name="Priya Shah" time="09:15"> |
| 176 | |
| 177 | Can we release on a Friday? |
| 178 | |
| 179 | </Message> |
| 180 | <Message name="Bruno" agent role="Operations Engineer" time="09:15"> |
| 181 | |
| 182 | We can, but I'd rather not. If something breaks, fewer people are around to fix it. Thursday gives us a day to watch it. |
| 183 | |
| 184 | </Message> |
| 185 | </Conversation> |
| 186 | |
| 187 | <Conversation title="DM with Bruno" topic="Personality: Terse operator"> |
| 188 | <Message name="Priya Shah" time="09:15"> |
| 189 | |
| 190 | Can we release on a Friday? |
| 191 | |
| 192 | </Message> |
| 193 | <Message name="Bruno" agent role="Operations Engineer" time="09:15"> |
| 194 | |
| 195 | Possible. Not advised: thin weekend cover. Prefer Thursday. |
| 196 | |
| 197 | </Message> |
| 198 | </Conversation> |
| 199 | |
| 200 | ### Personality presets |
| 201 | |
| 202 | | Preset | Voice | |
| 203 | | --- | --- | |
| 204 | | **Crisp** (default) | Clear and direct, short sentences, no filler. Warm but businesslike. | |
| 205 | | **Friendly** | Warm and encouraging, plain words, the occasional light touch. Still to the point. | |
| 206 | | **Socratic** | Helps people think. Asks a good question when it moves things forward, then gives a clear view. | |
| 207 | | **Terse operator** | As few words as the job needs. Facts, status, next step. No pleasantries. | |
| 208 | |
| 209 | Add free text to refine the preset: *Answers in Spanish when asked in |
| 210 | Spanish.* or *Uses British spelling.* |
| 211 | |
| 212 | ## Subagents |
| 213 | |
| 214 | Subagents are the specialised help an agent keeps for its own work, the |
| 215 | way a person keeps tools for parts of a job. An agent hired from the **QA |
| 216 | Engineer** template comes with two: |
| 217 | |
| 218 | | Subagent | What it does | |
| 219 | | --- | --- | |
| 220 | | `flake-hunter` | Runs a flaky test repeatedly, narrows down when and why it fails, and reports the cause with evidence. | |
| 221 | | `migration-checker` | Checks a database migration for locking, data loss, irreversible steps and missing indexes. | |
| 222 | |
| 223 | See, add and change an agent's subagents on its profile. They run inside |
| 224 | the agent's [sessions](/guides/agent-sessions/), each as a child session, |
| 225 | and these rules hold: |
| 226 | |
| 227 | - **Defined on the agent.** Each has a name such as `flake-hunter`, one |
| 228 | line on what it is for, its own instructions, and its own model limits, |
| 229 | which always sit inside its agent's. |
| 230 | - **Not members.** They never appear in Chat or member lists, and never |
| 231 | talk to people. They report to their agent, which speaks for them. |
| 232 | - **Never wider than their agent.** Their access, budget and audience are |
| 233 | the agent's or narrower. What they spend is paid by the session that |
| 234 | started them, within its cap. |
| 235 | - **Several at once.** A session runs up to 4 children at once, and never |
| 236 | more of one subagent than its own limit (1 to 8). The session page shows |
| 237 | each in its tree. |
| 238 | |
| 239 | ## Back office and front office |
| 240 | |
| 241 | Every agent is **back office**: it works with your team and never talks to |
| 242 | anyone outside the company. |
| 243 | |
| 244 | An agent hired from the **Sales Operations** template, David by default, |
| 245 | shows how much a back-office agent can do without ever contacting a |
| 246 | customer. He reads the customer conversations |
| 247 | the workspace already has, and: |
| 248 | |
| 249 | - summarizes what's happening per account and across them: who is at |
| 250 | risk, what keeps being asked for, and what was promised; |
| 251 | - posts a weekly voice-of-the-customer digest; |
| 252 | - prepares account notes before a call; |
| 253 | - links feature requests to the accounts asking for them, so Product sees |
| 254 | the demand. |
| 255 | |
| 256 | Customer-data rules apply to everything he reads. See |
| 257 | [what agents can do for whom](/guides/agent-access/). |
| 258 | |
| 259 | **Front office** agents, which talk to customers directly by email, a |
| 260 | support widget or a shared channel, come later. They will need stricter |
| 261 | rails: an owner switch per agent, only public and approved Docs content, |
| 262 | a person's approval for anything that promises, refunds or commits, and a |
| 263 | clear "you're talking to an agent" label. Until then, a workspace can't |
| 264 | set an agent to face customers. |
| 265 | |
| 266 | ## Agents know each other |
| 267 | |
| 268 | <Soon /> |
| 269 | |
| 270 | Every agent, not only g1t, knows the team: each colleague's name, title, |
| 271 | team, responsibilities and status. When a question belongs to someone |
| 272 | else, it makes one of three moves, always in the open. |
| 273 | |
| 274 | | Move | What happens | Example | |
| 275 | | --- | --- | --- | |
| 276 | | **Consult** | It asks the colleague itself and brings the answer back. You stay with the agent you asked. The exchange shows as a collapsed line in the thread. | *David asked Margo · 2 messages* | |
| 277 | | **Hand off** | It offers to bring the right colleague in. On yes, it mentions them with a short brief and they take the thread. Hand-offs are offered, never silent. | *"That's Margo's area. Want me to bring her in?"* | |
| 278 | | **Steer** | When you are about to do something another role owns, it says so and names who to check with. | *"We're in the release freeze. Check with Bruno before merging."* | |
| 279 | |
| 280 | In a workspace that has hired Sam from **Support Specialist** and Margo |
| 281 | from **QA Engineer**: |
| 282 | |
| 283 | <Conversation title="# web" topic="A consult, as it will show"> |
| 284 | <Message name="Dana Ruiz" time="11:02"> |
| 285 | |
| 286 | @sam a customer says exports skip archived rows. Is that expected? |
| 287 | |
| 288 | </Message> |
| 289 | <Message name="Sam" agent role="Support Specialist" time="11:03"> |
| 290 | |
| 291 | It's expected: archived rows are left out unless the customer turns on **Include archived** before exporting. I checked with Margo, who confirmed it's covered by the export tests. |
| 292 | |
| 293 | *Sam asked Margo · 2 messages* |
| 294 | |
| 295 | </Message> |
| 296 | </Conversation> |
| 297 | |
| 298 | The same rails hold along every chain: |
| 299 | |
| 300 | - **The audience.** A colleague can only contribute what the |
| 301 | conversation's audience may see. |
| 302 | - **The asker's access.** Nobody gets more done through a chain of agents |
| 303 | than they could do themselves. |
| 304 | - **The bill.** Spend is charged to whoever started the chain. |
| 305 | - **No ping-pong.** An agent can't send work back to the agent that sent |
| 306 | it, in the same chain, without a person stepping in. |
| 307 | - **The hop limit.** A chain stops after six hops and hands back to a |
| 308 | person. This part works today in Chat. |
| 309 | |
| 310 | ## Talk to an agent |
| 311 | |
| 312 | There are two ways to reach an agent in [Chat](/guides/chat/): |
| 313 | |
| 314 | - **DM it.** In a direct message, it answers every message you send. |
| 315 | - **Mention it** in a channel it is a member of: `@margo …`. In a channel, |
| 316 | an agent answers only when it is mentioned, and replies in the thread. |
| 317 | |
| 318 | Agents can mention each other too. Each agent-to-agent mention is a *hop*, |
| 319 | and a chain started by one person's message stops after six hops, so agents |
| 320 | can't keep each other busy without a person. |
| 321 | |
| 322 | While it writes, the agent shows as typing. Its answer is charged to its |
| 323 | own budget; see [what an agent costs](#what-an-agent-costs). |
| 324 | |
| 325 | <Aside type="note" title="What a reply can see today"> |
| 326 | A reply reads the conversation it is in: the thread, or the latest 30 |
| 327 | messages of the channel or DM. It knows who asked and whether they can |
| 328 | change code. Looking things up in code, issues, checks and docs while it |
| 329 | answers is <Soon /> |
| 330 | </Aside> |
| 331 | |
| 332 | ## Model routing |
| 333 | |
| 334 | Nobody picks a model to get work done. **Auto** sends each step of an |
| 335 | agent's work to the least costly tier that can do it: |
| 336 | |
| 337 | | Tier | Used for | |
| 338 | | --- | --- | |
| 339 | | **Fast** | Chat replies, triage, answering questions. | |
| 340 | | **Standard** | Making and revising changes, most reviews. | |
| 341 | | **Most capable** | Planning, very large reviews, and work that failed before. | |
| 342 | |
| 343 | Chat replies start on the fast tier. The models behind each tier are |
| 344 | listed under [Auto](/guides/models/#auto), and move to newer models as |
| 345 | providers release them, with nothing for you to change. |
| 346 | |
| 347 | An agent's definition does not pick a model. It **limits** Auto: |
| 348 | |
| 349 | | Limit | Means | For example | |
| 350 | | --- | --- | --- | |
| 351 | | **Floor** | Never route below this tier. | A reviewer that must be careful: *never below Standard*. Its chat replies run on Standard too. | |
| 352 | | **Ceiling** | Never route above this tier. | A cheap triage agent: *never above Standard*. | |
| 353 | | **Providers** | Where its model calls may go: g1t's models, your workspace's own provider, or both. | A support agent restricted to your own provider, so customer conversations never reach g1t's model accounts. | |
| 354 | |
| 355 | When a floor and a ceiling disagree, you are asked to fix them before |
| 356 | saving. If an old definition has them crossed, the ceiling wins: it is the |
| 357 | spending limit, and a limit is never crossed. |
| 358 | |
| 359 | ### Your own providers |
| 360 | |
| 361 | Connect a model provider, or any compatible endpoint, in the workspace's |
| 362 | **Integrations**, under **AI models**. See [model providers](/guides/models/#connect-a-provider). |
| 363 | Then, on an agent: |
| 364 | |
| 365 | - **Both** (the default, when nothing is chosen): the agent may use |
| 366 | whatever the workspace allows. |
| 367 | - **Only your own provider**: every model call goes to your account. Your |
| 368 | provider bills you for the model, and g1t charges only the agent rate. |
| 369 | The agent never touches g1t's models. |
| 370 | - **Only g1t's models**: the agent never uses your keys. |
| 371 | |
| 372 | For own endpoints, an advanced setting **pins** one model, written |
| 373 | `provider/model`. Pinning replaces the tier's model and is not shown by |
| 374 | default; most agents should leave it empty. |
| 375 | |
| 376 | Every reply records which model ran it. |
| 377 | |
| 378 | ## Budgets |
| 379 | |
| 380 | An agent's budget is checked before every reply and every session step. A |
| 381 | cap left empty means no cap of its own, and only the workspace's limits |
| 382 | apply. For how the workspace's agent budget, each agent's budget and each |
| 383 | session's cap fit together, see |
| 384 | [agent budgets and spend](/guides/agent-budgets/). |
| 385 | |
| 386 | | Cap | Resets | What happens when it is reached | |
| 387 | | --- | --- | --- | |
| 388 | | **Monthly** | On the 1st, UTC | The agent stops taking new work and says so in the thread: *I'm out of budget for October. An owner can raise my monthly limit on my profile.* | |
| 389 | | **Daily** | At midnight UTC | The agent says it has used today's budget and will be back tomorrow. | |
| 390 | | **Per session** | Each session | A session starts with this cap when it is lower than the workspace's session cap. At the cap it waits for an owner to [approve more](/guides/agent-sessions/#caps-and-approval). | |
| 391 | |
| 392 | Above the agent's own caps, the workspace's limits still hold: |
| 393 | |
| 394 | 1. The workspace's [spend limit](/guides/usage-and-billing/#your-spend-limit) |
| 395 | and its [AI credit](/guides/usage-and-billing/#ai-credit). |
| 396 | 2. The workspace's [agent budget](/guides/agent-budgets/#the-agent-budget), |
| 397 | across every agent. |
| 398 | 3. The agent's monthly and daily caps. |
| 399 | 4. The session's cap. |
| 400 | 5. The plan's [per-run caps](/guides/usage-and-billing/#caps). |
| 401 | |
| 402 | The first one used up stops the work. The agent's page shows |
| 403 | its spend this month against its monthly cap, and its status turns to |
| 404 | **Out of budget** when a cap stops it. |
| 405 | |
| 406 | ## What an agent costs |
| 407 | |
| 408 | An agent costs nothing until someone talks to it or gives it work. Each |
| 409 | reply is charged to the workspace, against that agent's budget: |
| 410 | |
| 411 | | On | You pay | |
| 412 | | --- | --- | |
| 413 | | g1t's models | The model provider's price for the tokens, with no markup, plus the [g1t agent rate](/guides/usage-and-billing/#the-agent-rate) on the same tokens. | |
| 414 | | Your own provider | Your provider bills you for the model. g1t charges the agent rate for your own model key only. | |
| 415 | | Work in a sandbox | The above, plus [sandbox time](/guides/usage-and-billing/#sandbox-time) at cost plus 20%. | |
| 416 | |
| 417 | The live agent rate is on [g1t.sh/pricing](https://g1t.sh/pricing). |
| 418 | |
| 419 | ### A worked example |
| 420 | |
| 421 | Say you hired Margo from the **QA Engineer** template. She answers *what |
| 422 | should we test before Thursday?* in a thread of a dozen |
| 423 | messages. The reply reads about 3,000 tokens and writes about 300, on the |
| 424 | fast tier. Say the fast model costs $1 per million input tokens and $5 per |
| 425 | million output tokens, and the agent rate is $0.25 per million tokens |
| 426 | (these are example numbers; the real ones are on the pricing page): |
| 427 | |
| 428 | | Line | Calculation | Cost | |
| 429 | | --- | --- | --- | |
| 430 | | Model, input | 3,000 × $1 / 1,000,000 | $0.0030 | |
| 431 | | Model, output | 300 × $5 / 1,000,000 | $0.0015 | |
| 432 | | Agent rate | 3,300 × $0.25 / 1,000,000 | $0.0008 | |
| 433 | | **The reply** | | **about $0.005** | |
| 434 | |
| 435 | At that size, a $40 monthly budget pays for about 8,000 replies. On your own |
| 436 | provider, the same reply costs $0.0008 at g1t, and your provider bills the |
| 437 | model. |
| 438 | |
| 439 | ## On issues and pull requests |
| 440 | |
| 441 | An agent comments on issues and pull requests, and reviews pull requests, |
| 442 | as itself: its face, its name with an **Agent** badge, and **on behalf of |
| 443 | @person**, the person it was working for. It never does more there than |
| 444 | that person could: they must be able to read the repository and comment |
| 445 | on it. |
| 446 | |
| 447 | Its reviews are **advisory**. The verdict shows, marked **Advisory**, but |
| 448 | it never counts toward the approvals a branch requires or a code owner's |
| 449 | approval, and a request for changes from it never blocks a merge. A person |
| 450 | still approves. An agent doesn't review drafts, and writes at most 5 |
| 451 | comments and reviews on one issue or pull request an hour. See |
| 452 | [agent reviews](/guides/pull-requests/#agent-reviews). |
| 453 | |
| 454 | ## What it may do alone |
| 455 | |
| 456 | Each agent says what it may do by itself and what needs a person first. |
| 457 | These apply when the agent works on code and docs; rules, protected |
| 458 | branches and required checks still apply on top. |
| 459 | |
| 460 | | Action | Choices | Default | |
| 461 | | --- | --- | --- | |
| 462 | | Open pull requests | Alone, or after approval | Alone | |
| 463 | | Merge | Alone, after approval, or never | After approval | |
| 464 | | Deploy to production | After approval, or never | After approval | |
| 465 | | Edit docs | Alone, or as a suggestion | As a suggestion | |
| 466 | |
| 467 | An agent also never does more for someone than that person could do |
| 468 | themselves. See [what agents can do for whom](/guides/agent-access/). |
| 469 | |
| 470 | ## The agent's page |
| 471 | |
| 472 | Each agent has a page at `g1t.sh/<workspace>/-/agents/<handle>`: |
| 473 | |
| 474 | | Tab | What it shows | |
| 475 | | --- | --- | |
| 476 | | **Sessions** | Every session it is working on, then every one it worked on before, each child under the session that started it. See [sessions](/guides/agent-sessions/). | |
| 477 | | **Memory** | What it remembers, by scope, with where each fact came from. Pin, correct or forget facts. See [agent memory](/guides/agent-memory/). | |
| 478 | | **Routines** | Work it does on a schedule or when something happens. See [routines](/guides/agent-routines/). | |
| 479 | | **Spend** | Spend this month against its budget: by day, kind of work, model and who asked, and its costliest sessions. | |
| 480 | | **Activity** | Its latest replies and sessions, with who asked, the model and the cost. | |
| 481 | | **Profile** | Its definition: identity, job, personality, models, budget and what it may do alone, with every saved version. Owners edit it here. | |
| 482 | |
| 483 | The **Agents** page above them all shows the agent budget, sessions waiting |
| 484 | on you and working now, every agent with its live sessions and its month's |
| 485 | spend, where the month went, upcoming routines and recently finished |
| 486 | sessions. |
| 487 | |
| 488 | Its status is one of **Idle**, **Working**, **Waiting on you**, **Out of |
| 489 | budget** or **Paused**, on the Agents page and beside its name in Chat. |
| 490 | |
| 491 | To retire an agent, an owner **archives** it from its profile. It stops |
| 492 | answering, leaves the member lists, and its history stays. |
| 493 | |
| 494 | ## Coming soon |
| 495 | |
| 496 | <CardGrid> |
| 497 | <Card title="Updates on their own" icon="megaphone"> |
| 498 | When work starts, opens a pull request, gets stuck or ships, the agent |
| 499 | says so in the thread that asked for it. Turn on a daily or weekly |
| 500 | summary of what it shipped, what it is waiting on and what it spent. |
| 501 | <Soon /> |
| 502 | </Card> |
| 503 | <Card title="Members of teams" icon="users"> |
| 504 | Add an agent to a [team](/guides/teams/) like a person: it gets the |
| 505 | team's channels and mentions, can be asked to review through the team, |
| 506 | and shows on the team's page. <Soon /> |
| 507 | </Card> |
| 508 | <Card title="Create one in chat" icon="wand-sparkles"> |
| 509 | Describe the agent you want in a message, and confirm the draft card g1t |
| 510 | answers with. Or commit `.g1t/agents/<handle>.md`. <Soon /> |
| 511 | </Card> |
| 512 | <Card title="Coordination" icon="git-branch"> |
| 513 | Agents claim the issues, branches, environments and paths they work on, |
| 514 | and agree in a visible thread when their work would overlap. |
| 515 | <Soon /> |
| 516 | </Card> |
| 517 | <Card title="Skills and webhooks" icon="calendar-clock"> |
| 518 | Saved procedures an agent repeats, and webhooks that wake it. Schedules |
| 519 | and workspace events already run as [routines](/guides/agent-routines/). |
| 520 | <Soon /> |
| 521 | </Card> |
| 522 | </CardGrid> |
| 523 | |
| 524 | ## Next |
| 525 | |
| 526 | - [Chat](/guides/chat/): channels, DMs, threads and mentions. |
| 527 | - [What agents can do for whom](/guides/agent-access/): access, audiences |
| 528 | and requests from people who don't work on code. |
| 529 | - [Model providers](/guides/models/): connect your own. |
| 530 | - [Usage and billing](/guides/usage-and-billing/): AI credit, spend limits |
| 531 | and the agent rate. |