Pick any line to see why it is the way it is: the commit, the pull request and issue it came from, and what the agent was thinking.
| Chat and workspace agents: channels, DMs and named agents you talk to | 1 | { |
| 2 | "$schema": "../../node_modules/wrangler/config-schema.json", | |
| 3 | "name": "g1t-agents", | |
| 4 | "account_id": "1e6f2cffa3f445920836e8ebe446bb58", | |
| 5 | "compatibility_date": "2026-09-26", | |
| 6 | "main": "./src/index.ts", | |
| 7 | // Reached only through service bindings: the site, the API and the chat | |
| 8 | // service. Chat hands it messages to answer (`deliver`). | |
| 9 | "workers_dev": false, | |
| 10 | "placement": { "mode": "off" }, | |
| 11 | "d1_databases": [ | |
| 12 | { | |
| 13 | "binding": "DB", | |
| 14 | "database_name": "g1t-agents", | |
| 15 | // npx wrangler d1 create g1t-agents, and its id here. | |
| Chat and agents: their D1 databases | 16 | "database_id": "454dd730-b6cb-42a3-885f-a02f9adbfce0", |
| Chat and workspace agents: channels, DMs and named agents you talk to | 17 | "migrations_dir": "migrations" |
| 18 | } | |
| 19 | ], | |
| 20 | // Each agent's desk: one object per agent, which queues what it is | |
| 21 | // handed and works at most its capacity of replies at once (src/desk.ts). | |
| 22 | "durable_objects": { | |
| 23 | "bindings": [{ "name": "DESKS", "class_name": "Desk" }] | |
| 24 | }, | |
| 25 | "migrations": [{ "tag": "v1", "new_sqlite_classes": ["Desk"] }], | |
| 26 | "services": [ | |
| 27 | // Reading the conversation, typing, and posting replies as the agent. | |
| 28 | { "binding": "CHAT", "service": "g1t-chat" }, | |
| Merge the workspace shell: navigation and phone shell, g1t as orchestrator, agents in roles with audience-checked reads, reactions and custom emoji, live notifications and browser push, the homepage tour (agents 0002, chat 0002) | 29 | // Workspaces by slug; for read tools, the audience's people with |
| 30 | // their access (users_for_audience), members and teams. | |
| Chat and workspace agents: channels, DMs and named agents you talk to | 31 | { "binding": "IDENTITY", "service": "g1t-identity" }, |
| Merge the workspace shell: navigation and phone shell, g1t as orchestrator, agents in roles with audience-checked reads, reactions and custom emoji, live notifications and browser push, the homepage tour (agents 0002, chat 0002) | 32 | // Read tools, always after the audience check (src/audience.ts): |
| 33 | // files, issues and pull requests, and code search. | |
| 34 | { "binding": "REPOS", "service": "g1t-repos" }, | |
| 35 | { "binding": "WORK", "service": "g1t-work" }, | |
| 36 | { "binding": "SEARCH", "service": "g1t-search" }, | |
| Chat and workspace agents: channels, DMs and named agents you talk to | 37 | // The compute gate (reserve, settle), runs (start_run, finish_run), |
| 38 | // the price book and staff's model defaults. | |
| 39 | { "binding": "BILLING", "service": "g1t-billing" }, | |
| 40 | // The workspace's own model provider, and a model session for each | |
| 41 | // reply, as runs get. | |
| 42 | { "binding": "INTEGRATIONS", "service": "g1t-integrations" }, | |
| 43 | // The audit log: agents created, changed and archived. | |
| 44 | { "binding": "EVENTS", "service": "g1t-events" }, | |
| 45 | // The model proxy, which replies' model requests go through with a | |
| 46 | // model session's token, as sandboxes' do: it holds the keys, g1t's or | |
| 47 | // the workspace's own, and counts what each answer cost. A binding, | |
| 48 | // not its address: same-zone Worker-to-Worker fetches can be refused. | |
| Agents work in sessions: bounded, visible, steerable work spun off from chat, with subagents and colleagues in a tree paid by its root; memory with sources and scopes; routines; a workspace budget for every agent; agents file issues for whoever asked | 49 | { "binding": "MODELS", "service": "g1t-models" }, |
| The artifacts service is services/artifacts, the Worker g1t-artifacts, bound as ARTIFACTS by the API, the site and the agents; its live rooms move to it with a Durable Object transfer from g1t-docs-service, and its database, bucket, indexes and queue keep their names. The git store's binding and settings are GITSTORE, its ops scripts gitstore-*, and workflow run artifacts keep their compatible API under run_artifacts modules. The deploy tool puts a Worker that has never deployed before the Workers in its stage that bind to it, and the deploy guide gives the cutover runbook. | 50 | // Artifacts (services/artifacts): agents search, read, write and suggest |
| 51 | // edits to artifacts, for the person they act for (src/ports.ts folioPorts). | |
| 52 | { "binding": "ARTIFACTS", "service": "g1t-artifacts" }, | |
| Agents work in sessions: bounded, visible, steerable work spun off from chat, with subagents and colleagues in a tree paid by its root; memory with sources and scopes; routines; a workspace budget for every agent; agents file issues for whoever asked | 53 | // Notifications: a session waiting for an owner to approve more spend. |
| 54 | { "binding": "NOTIFY", "service": "g1t-notify" } | |
| Chat and workspace agents: channels, DMs and named agents you talk to | 55 | ], |
| Agents work in sessions: bounded, visible, steerable work spun off from chat, with subagents and colleagues in a tree paid by its root; memory with sources and scopes; routines; a workspace budget for every agent; agents file issues for whoever asked | 56 | // Every five minutes: routines that are due, and session steps a desk |
| 57 | // lost (src/routines.ts, src/sessions.ts). | |
| 58 | "triggers": { "crons": ["*/5 * * * *"] }, | |
| Routines run when something happens: a pull request ready for review or merged, checks or a deploy failing, an issue opened | 59 | // Events routines run on: pull requests ready for review or merged, |
| 60 | // checks and deploys failing, issues opened (src/triggers.ts). Create it | |
| 61 | // once: npx wrangler queues create g1t-events-agents | |
| 62 | "queues": { | |
| 63 | "consumers": [{ "queue": "g1t-events-agents", "max_batch_size": 20, "max_batch_timeout": 2, "max_retries": 3, "dead_letter_queue": "g1t-events-dlq" }] | |
| 64 | }, | |
| Chat and workspace agents: channels, DMs and named agents you talk to | 65 | "vars": { |
| 66 | // Who may use g1t's hosted models while billing takes no real money: | |
| 67 | // the same list as the runner's (services/runner/wrangler.jsonc, and | |
| 68 | // src/hosted.ts there, which this service uses). Keep the two equal. | |
| 69 | "HOSTED_AGENT_WORKSPACES": "flagon-io", | |
| 70 | // The runner's routing policy, the same JSON as its AGENT_ROUTING: | |
| 71 | // the model behind each tier. Staff's defaults in sudo (billing's | |
| 72 | // model_defaults) go on top, read once a minute, as in the runner. | |
| 73 | // Chat replies start on the small tier; an agent's floor and ceiling | |
| 74 | // move them. Keep it equal to the runner's. | |
| 75 | "AGENT_ROUTING": "{\"tiers\":{\"small\":{\"modelName\":\"Claude Haiku 5.5\",\"model\":\"claude-haiku-5-5\",\"price\":{\"input\":0.1,\"output\":0.5,\"cacheRead\":0.01,\"cacheWrite\":0.125}},\"large\":{\"modelName\":\"Claude Sonnet 5.5\",\"model\":\"claude-sonnet-5-5\",\"price\":{\"input\":2,\"output\":10,\"cacheRead\":0.1,\"cacheWrite\":2.5}},\"frontier\":{\"modelName\":\"Claude Opus 5.5\",\"model\":\"claude-opus-5-5\",\"price\":{\"input\":4,\"output\":20,\"cacheRead\":0.2,\"cacheWrite\":5}}},\"tasks\":{\"implement\":\"large\",\"revise\":\"large\",\"answer\":\"small\",\"review\":\"change\",\"update\":\"small\",\"plan\":\"small\"},\"effort\":{\"plan\":\"high\",\"answer\":\"medium\",\"update\":\"low\"},\"smallChange\":{\"files\":10,\"lines\":200},\"largeChange\":{\"files\":60,\"lines\":3000},\"largeLabels\":[\"security\"],\"frontierLabels\":[\"architecture\"],\"smallLabels\":[\"documentation\",\"docs\",\"typo\"],\"frontierAfter\":2,\"learning\":{\"window\":20,\"minRuns\":5,\"stepDownAt\":0.9,\"stepUpAt\":0.5}}", | |
| 76 | // The model proxy's address, used only without the MODELS binding | |
| 77 | // (an installation that runs its proxy elsewhere). | |
| Every agent starts with g1t's foundational skills, playbooks for documents, research, data, code, communication, and files and media, each using only tools the agent already has and saying plainly what isn't possible yet: agents make PDFs, Word documents, spreadsheets, CSV and Markdown files with make_file, kept with a doc in Artifacts and served from g1tusercontent.com; an agent's Skills tab shows what each skill can do and with which tools, and owners turn a skill off per agent; the agent skills guide says how. | 78 | "MODELS_URL": "https://models.g1t.sh", |
| 79 | // Where files kept with artifacts are served, for the links to files agents make (src/ports.ts). | |
| 80 | "USERCONTENT_URL": "https://g1tusercontent.com" | |
| Chat and workspace agents: channels, DMs and named agents you talk to | 81 | }, |
| 82 | // Every log kept while replies are new: a failed one must be traceable. | |
| 83 | "observability": { "enabled": true, "head_sampling_rate": 1 } | |
| 84 | } |
This file's history is long; its oldest lines are credited to the oldest commit read.