Skip to content

g1t/docs/BILLING_OPERATIONS.md

938 lines60,650 bytesCodeBlame

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.

Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily1# Billing operations: costs, margin and prices
2
3How g1t checks what it charges against what Cloudflare charges it, keeps
4prices at cost plus 20%, and tells staff when the margin slips. Internal.
5Code: `services/billing/src/costs.rs` (reading the bill), `margin.rs`
6(reconciliation, drift, alerts), `pricing.rs` (versions, proposals,
Spend caps: a monthly budget for comped workspaces and a daily breaker on what g1t pays7notice), `keeper.rs` (sandbox and Workers for Platforms measurements),
8`budget.rs` (what g1t pays for itself, and its caps; see
9[Spend caps](#spend-caps)).
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily10Page: sudo **Costs & margin** (`/costs`).
11
12Several Cloudflare products g1t runs on are new. Artifacts bills
13"operations" from 2026-10-14 without defining them (see
14[ARTIFACTS.md](ARTIFACTS.md), M1). So nothing here hard-codes a product
15list or a unit: every line Cloudflare bills is kept, and how a line maps
16to what g1t sells is data you change from sudo, without a deploy.
17
18## Data sources
19
20| Source | What | Where it lands |
21| --- | --- | --- |
22| Billable usage, `GET /accounts/{account}/billable-usage?from=&to=` | One row per service per day in FOCUS columns: `ServiceFamilyName`, `ServiceName`, `ChargePeriodStart`, `PricingQuantity`, `ContractedCost` / `BilledCost` / `ListCost`. Every product g1t uses appears once it is used: Workers, Workers for Platforms, D1, KV, R2, Queues, Containers, Durable Objects, Artifacts, Browser Rendering, Workers AI, Vectorize, Cloudflare for SaaS, Email. Inside an included amount the cost is 0. | `cost_lines`, source `billable_usage` |
Costs: Cloudflare's count for a pull request's working copy is shared out to its repository's workspace (repos pull_owners)23| GraphQL `artifactsEventsAdaptiveGroups` | Artifacts' own count by `date`, `eventType` and `repositoryName`. Operations are `create`, `fork`, `push`, `pull`, `delete`; errors (`rateLimited`, `serverError`, …) are kept but not counted. | `cost_lines`, source `artifacts_events`; per workspace (from the store key `<workspace>--<repo>`; a pull request's working copy, `pulls--<id>`, is its repository's workspace's, from repos' `pull_owners`) in `own_counts` as `cloudflare_git` |
Merge branch 'worktree-agent-a633ac0f7f66d419d'24| GraphQL `aiGatewayRequestsAdaptiveGroups`, filtered to `AI_GATEWAY_ID` | What AI Gateway priced g1t's own provider traffic at, by `date`, `provider`, `model` and `wholesale`: `count`, `sum.cost` (dollars), `sum.tokensIn`/`tokensOut`/`cacheReadTokens`/`cacheWriteTokens`. Field names checked against Cloudflare's schema (introspection of `AccountAiGatewayRequestsAdaptiveGroups{Sum,Dimensions,Filter_InputObject}`). An adaptive (sampled) dataset: an estimate, close at g1t's volumes. Only g1t's hosted models go through this gateway: a workspace's own provider is called at its own address, never here. | `cost_lines`, source `ai_gateway`, product `ai_gateway_requests`: per day and model a line `<provider>_<model>` (requests, at the gateway's cost), and at no cost `…__tokens`, `…__cache_read_tokens`, `…__cache_write_tokens`; Cloudflare-billed (unified billing) requests are prefixed `wholesale__`. Mapped to `models` (migration 0036). A re-read day replaces all its gateway lines. |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily25| The ledger | Every charge: its cost at the price book's cost, what it was charged at price, what paid for it. | read, never written |
26| `pending_usage` | Month-end meters (git, storage, scans, embeddings, the cache) as they stand. | snapshotted daily into `pending_days` |
27| `plan_payments` | The plan's $20. | read |
One operation mapping, owned by repos; billing reads it instead of keeping its own28| repos `git_operations` | Operations customers are charged for, per workspace, counted by repos through its `operation_mapping`. | `own_counts` meter `git_operations` |
Costs: Cloudflare's subscriptions read from Cloudflare each day, the estimate only until then; sudo's costs split into Costs & margin and Bill & pricing29| Subscriptions, `GET /accounts/{account}/subscriptions` | What g1t pays each month whatever it uses (Workers Paid, add-ons): each subscription that is paid, trialing or awaiting payment, at its price over its frequency. Not on the billable-usage bill. Read in the daily run with the bill's token; a failure is logged and the last read stays. | `cf_subscriptions` (one row) |
One operation mapping, owned by repos; billing reads it instead of keeping its own30| repos `artifacts_usage` | Every raw meter of the git store (`git.fetch`, `git.receive_pack`, `binding.*`, …) per day and workspace, with repos' `operation_mapping`. | `own_counts` meters `artifacts_<raw meter>`, and `cost_operations` (raw counts × the mapping's `cost_operations`: what g1t expects Cloudflare to bill) |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily31
32A meter's slug is Cloudflare's name lower-cased with words joined by `_`
33and the "(First … included)" note dropped: `Workers for Platforms CPU ms
34(First 60M ms are included)` under `Workers` is product `workers`, meter
35`workers_for_platforms_cpu_ms`. Several rows of the same day and meter
36(regions, tiers) are added together before they are stored.
37
38## Credentials
39
40| Secret on g1t-billing | Permissions | Used for |
41| --- | --- | --- |
Costs: Cloudflare's subscriptions read from Cloudflare each day, the estimate only until then; sudo's costs split into Costs & margin and Bill & pricing42| `CLOUDFLARE_BILLING_TOKEN` (optional) | Account: **Billing Read**, Account: **Account Analytics Read**, for the g1t account only | Reading the bill, the Artifacts events and the subscriptions |
Billing: AI Gateway's analytics are read with the token that can see them, and a gateway that priced nothing is said43| `CLOUDFLARE_USAGE_TOKEN` (exists) | Billing Read, Account Analytics Read, AI Gateway Read | The keeper (settling runs from the gateway's logs); also the bill when `CLOUDFLARE_BILLING_TOKEN` is not set. AI Gateway's analytics are read with this token first and, if that is refused, with the bill's: Cloudflare answers a token without AI Gateway Read with no rows rather than an error, so the bill's token would read as a gateway that priced nothing |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily44
45With neither, the daily run reconciles only what g1t counted itself, and
46the page says the bill cannot be read. Nothing fails. To set the scoped one:
47
481. Cloudflare dashboard → My Profile → API Tokens → Create Token → Custom token.
492. Permissions: Account · Billing · Read; Account · Account Analytics · Read.
503. Account resources: Include · the g1t account. No zone permissions.
514. `cd services/billing && npx wrangler secret put CLOUDFLARE_BILLING_TOKEN`.
sudo: the costs run button is named for what it does, the whole nightly analysis525. In sudo, Costs & margin → **Run the analysis now**.
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily53
54Alerts are emailed through the `EMAIL` binding (Cloudflare Email Sending)
55to `COSTS_ALERT_EMAIL` (`hey@flagon.io`). An empty value sends none.
56
57## Schedule
58
59The daily cron (`17 4 * * *`, `keeper::DAILY`) runs, in order:
60
611. The keeper's measurements (sandbox seconds, app requests and CPU), each
62 a proposal now, not a direct change.
632. `costs_daily`:
Merge branch 'worktree-agent-a633ac0f7f66d419d'64 1. Read the bill, the Artifacts events and AI Gateway's analytics. The first run reads the last
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily65 31 days (GraphQL keeps 31); later runs the last 4, since Cloudflare
66 restates recent days, or back to the last day read after a gap.
67 Lines are upserted on `(day, source, product, meter)`, so a re-read
68 replaces, never adds.
69 2. g1t's own counts for the same days (replaced per day).
70 3. Snapshot `pending_usage` into `pending_days`.
sudo: the costs run button is named for what it does, the whole nightly analysis71 4. Reconcile the last 31 days (further back after a gap) into
72 `margin_days` and `workspace_costs`
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily73 (replaced per day).
74 5. Drift over the last 7 days into `cost_drift`.
75 6. Unit costs over the last 30 days, proposed to the price book.
76 7. Apply price versions whose date has come.
77 8. Open, update and close margin alerts; email new ones.
78 9. Email owners on the plan about rises to come.
79
Costs: Cloudflare's subscriptions read from Cloudflare each day, the estimate only until then; sudo's costs split into Costs & margin and Bill & pricing80**Run the analysis now** at the top of sudo's Costs & margin and Bill & pricing pages
sudo: the costs run button is named for what it does, the whole nightly analysis81(`admin_run_costs`) runs all of step 2 at once, alerts included, with no
82need to wait for 04:17 UTC. Running it twice is safe: every step replaces
83what it wrote.
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily84
85## Reconciliation math
86
87Every Cloudflare line goes to one of g1t's products ("buckets") by
88`cost_map`: the row for its product with the longest matching meter
89prefix, `*` last. A line no row claims goes to `unmapped`.
90
91| Bucket | Cloudflare | Paid for by (`revenue_map`) |
92| --- | --- | --- |
93| `sandboxes` | Containers, Durable Objects compute duration | `sandbox`, `self_hosted`, `builds` |
94| `deployments` | Workers for Platforms | `deployments` |
95| `git` | Artifacts operations (and its events, as counts) | `git` |
96| `repo_storage` | Artifacts storage | `storage` |
97| `actions_cache` | R2 | `cache` |
98| `embeddings` | Workers AI, Vectorize | `context` |
99| `security` | (Workers CPU, under `platform`) | `security` |
100| `domains` | Cloudflare for SaaS | `domains` |
Merge branch 'worktree-agent-a633ac0f7f66d419d'101| `models` | not Cloudflare: AI Gateway's settled cost on the ledger; AI Gateway's own daily total (`ai_gateway_requests`) beside it, to check it | every other task (agent runs) |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily102| `platform` | Workers, D1, KV, Queues, Email, Browser Rendering, other Durable Objects | the plan's price |
103
104For each day and bucket:
105
Costs: a statement that keeps usage sold, running g1t, subscriptions and what was given away (comped, free use, trial, pool) apart, and says who was paid; free use carries its own cost; the run button says it is running106- **Cloudflare cost** = Σ the bucket's lines' cost, as billed: after the
107 included allowances, so a month inside them costs $0 here as on
108 Cloudflare's Billable usage page. `models` uses the ledger's cost of the
109 tokens instead; that is paid to the model providers and is not on
Merge branch 'worktree-agent-a633ac0f7f66d419d'110 Cloudflare's bill. Its "Cloudflare" column is what AI Gateway priced the
111 same traffic at, which drift compares with the ledger (below); it is
112 never added to the cost.
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily113- **Own cost** = Σ the ledger's `cost_micros` for the bucket's keys (the
114 price book's cost when charged), plus month-end deltas. A workspace's own
115 model provider is no cost to g1t.
116- **Value** = what customers were charged at price: `-amount_micros` plus
117 what the plan's included usage, a trial, the open-source pool or g1t paid.
118 g1t's own (comped) workspaces are valued at cost plus the margin.
119- **Cash** = what workspaces paid: `-amount_micros`, and the plan's price.
Merge Stripe Tax, the card fee on card payments, and one free workspace per person120 Never tax or card fees: a payment credits the balance, and
121 `plan_payments`, without them (see [Tax and the card fee](#tax-and-the-card-fee)).
Costs: margin is measured on what was sold; comped workspaces, free periods, the trial and the pools are given away, a budget shown beside it122- **Given away** = the part of the cost that went on usage g1t paid for
Costs: a statement that keeps usage sold, running g1t, subscriptions and what was given away (comped, free use, trial, pool) apart, and says who was paid; free use carries its own cost; the run button says it is running123 itself on purpose, by why:
124 - **comped**: all of a comped workspace's cost, every bucket;
125 - **free use**: a free period's usage, the overruns g1t covered
126 (`ledger.given_micros`), and all of a workspace's cost on a day it had
127 nothing priced (free allowances);
128 - **trial** and **open-source pool**: what `trial_micros` and
Merge branch 'worktree-agent-a633ac0f7f66d419d'129 `oss_micros` paid;
130 - **discount**: what a discount on an account's custom terms took below
131 cost plus the margin (`ledger.discount_micros`, see
132 [Margin floor](#margin-floor)). The usage is valued at its price, so a
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging133 discounted sale never reads as margin lost;
134 - **promotional credit** and **goodwill credit**: what credit staff gave
135 paid for, when it is spent (`given_credit_promotional_micros`,
136 `given_credit_goodwill_micros`, migration 0038). That usage's charge is
137 taken out of cash, so it is never money in. A refund is not here: see
138 [Credits from g1t](#credits-from-g1t).
Costs: a statement that keeps usage sold, running g1t, subscriptions and what was given away (comped, free use, trial, pool) apart, and says who was paid; free use carries its own cost; the run button says it is running139
140 Otherwise a workspace's day is split by those shares of its value at
141 price, and the same shares of each of its buckets' cost are given, its
142 part of running g1t included. The Team plan's included usage is sold:
143 the plan's price paid for it. Stored on `margin_days` (`given_micros`
Merge branch 'worktree-agent-a633ac0f7f66d419d'144 and `given_<why>_micros`, `given_discount_micros` from migration 0036)
145 and `workspace_costs` (`given_micros`). Sudo's Bill & pricing page lists
146 comped, free use, trial and pool by name; the discount part is in the
147 total until the page names it (`givenDiscountMicros`).
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily148- **Month-end meters**: a day's figure is that day's `pending_days`
149 snapshot less the day before's, within a month. Their month-end ledger
150 entries are left out, so nothing is counted twice.
Costs: a statement that keeps usage sold, running g1t, subscriptions and what was given away (comped, free use, trial, pool) apart, and says who was paid; free use carries its own cost; the run button says it is running151- **Product margin** = (value − cost) / value.
152- **Sudo's statement** keeps apart:
153 - **Usage sold**: cash for usage against the cost of the usage buckets
154 less what was given. Its margin is the headline; at cost plus 20% it
155 sits near 16.7%.
156 - **Running g1t**: the plan's price against `platform` less its given
157 share.
Costs: Cloudflare's subscriptions read from Cloudflare each day, the estimate only until then; sudo's costs split into Costs & margin and Bill & pricing158 - **Cloudflare subscriptions**: what Cloudflare lists, a month, over
159 the range (`cf_subscriptions`); until a read has worked,
160 `CLOUDFLARE_FIXED_MONTHLY_MICROS`, an estimate.
Costs: a statement that keeps usage sold, running g1t, subscriptions and what was given away (comped, free use, trial, pool) apart, and says who was paid; free use carries its own cost; the run button says it is running161 - **Not mapped**: billed, charged for by nothing.
162 - **Given away**: by why. A budget, watched under g1t's own spend, never
163 shown as a loss.
164 - **All in**: money in against all of it, with the figure without what
165 was given beside it. **Who g1t paid** splits the cost into Cloudflare
166 and the model providers.
167
168 The overall alert is (Σ cash − (Σ cost − Σ given)) / Σ cash.
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily169- **Quantities**: where a mapping names an `own_meter`, Cloudflare's
170 billed quantity of those lines (or, without one, Artifacts' operation
171 events) against g1t's own count.
172
173**Shared costs to workspaces.** A bucket's cost is shared in proportion
174to, first available: Cloudflare's own per-workspace count
175(`cloudflare_<bucket>`, today the Artifacts events by repository), g1t's
Costs: a statement that keeps usage sold, running g1t, subscriptions and what was given away (comped, free use, trial, pool) apart, and says who was paid; free use carries its own cost; the run button says it is running176own count, what its usage cost (so free use carries its own cost), what
177each was charged for it. `platform`
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily178and `unmapped` are shared by each workspace's share of all usage that
179day. Shares are whole micros that add up to the bill exactly (largest
180remainder).
181
182## Drift (last 7 days)
183
184| Kind | When | What to do |
185| --- | --- | --- |
One operation mapping, owned by repos; billing reads it instead of keeping its own186| Count | g1t's count and Cloudflare's differ by more than the mapping's `drift_percent` (10%) | Find out what Cloudflare counts: compare its events with `own_counts` `artifacts_*` and `cost_operations`. If it counts more (binding reads, `ls-refs`), either change repos' `operation_mapping` so customers are charged for what Cloudflare counts, or leave it and let the per-unit cost rise (below). |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily187| Cost | Cloudflare charged more than `drift_percent` away from the price book's cost of the same usage, with at least `min_daily_cost` | A price is stale: check the proposals. |
Billing: AI Gateway's analytics are read with the token that can see them, and a gateway that priced nothing is said188| Cost, on `models` | What AI Gateway priced g1t's own provider traffic at over the 7 days, against the ledger's model cost for the same days (billed to g1t: comped, free and trial use included, a workspace's own provider not), more than the `ai_gateway_requests` mapping's `drift_percent` (10%) apart, with at least `min_daily_cost`. A ledger with none of the gateway's cost is drift too, and so is a gateway that priced nothing against a ledger with at least `min_daily_cost` of model cost (no percentage): that is not agreement, it is a token that cannot see AI Gateway, or calls that went around it | The gateway higher: model calls g1t paid for and charged no one: runs not settled yet (they catch up within the hour), runs with no session, a run started without a billing ticket, or something else on g1t's gateway. The ledger higher: runs that reached a provider without the gateway. The detail adds why the gateway's own figure may be off: prompt-cache read and write tokens (the gateway prices them at its rates for cache tokens, which can lag the provider's; check against the provider's invoice), requests Cloudflare billed itself (unified billing: on Cloudflare's bill, not a provider's), and models with no price. Days are UTC by when a request ran (gateway) and when a charge was entered (ledger), so a run across midnight shifts a little between days; the 7-day sum absorbs it. |
Merge branch 'worktree-agent-a633ac0f7f66d419d'189| Unpriced | Over the 7 days, a model in AI Gateway's analytics with tokens and $0 cost, or runs settled with `runs.gateway_note` (the gateway could not price all of a run) | The gateway has no price for a model g1t runs: add it in the gateway (custom cost) or route away from it. Until then those runs are charged no less than the sandbox reported (Claude Code's own price table), never $0 silently. |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily190| Leak | Cost of at least `min_daily_cost` and nothing charged for it (never for `platform`), or a meter in `unmapped` | Map the meter (below), or decide it is overhead (`platform`). |
191
192## Prices: versions, proposals, notice
193
194- `price_versions` holds every price ever, never edited. `prices` is the
195 version in force. The daily run applies a version once its
196 `effective_at` has come, and adds the public `price_changes` record.
197 Ledger entries made from the price book carry `price_version` (the
198 version ids, comma-separated), so a past statement is always explained
199 by the prices of its day.
200- Proposals come from the keeper (sandbox seconds, app requests and CPU)
201 and the reconciler (mappings with `scale_to_own`: today git operations).
202 For git operations: Cloudflare's rate per its own operation (the median
203 over charged days of cost ÷ quantity) × (Cloudflare's operations ÷ g1t's)
204 × 1,000. If Cloudflare counts three for each one g1t counts, the per-1,000
205 price triples. At least 1,000 of g1t's operations are needed.
206- Decision (`pricing::decide`): under 2% is noise; more than 4× either way
207 is suspect and waits for staff; within `auto_apply_percent` (25%) it is
208 applied on its own when `auto_apply` is on; anything else waits.
209- Notice: a fall applies at once. A rise applies `notice_days` (14) after
210 the decision, and for a monthly meter (git, storage, cache, domains,
211 embeddings, scans) at the start of the month after that, so no month
212 is charged at two prices. Owners of workspaces on the plan are emailed
213 once per rise (`price_notices`), and the pricing page lists it with
214 "takes effect". Rises are never retroactive; margin protection is for
215 new usage once notice has run.
216- Staff approve or reject in sudo. A rejection needs a note.
217
218## Changing a mapping
219
Costs: Cloudflare's subscriptions read from Cloudflare each day, the estimate only until then; sudo's costs split into Costs & margin and Bill & pricing220In sudo, Costs & margin → Bill & pricing → **Mappings**: Cloudflare's product and meter
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily221prefix (as **Cloudflare's lines** lists them; `*` for the rest of the
222product), g1t's product, and optionally:
223
224- **Price meter**: the price book meter the line measures.
225- **Own meter**: g1t's count of the same units (`own_counts.meter`).
226- **Scale to g1t's count**: price one of g1t's units at as many of
227 Cloudflare's as it took (proposals as above).
228- **Drift threshold**.
229
sudo: the costs run button is named for what it does, the whole nightly analysis230It applies from the next run; **Run the analysis now** applies it at once.
One operation mapping, owned by repos; billing reads it instead of keeping its own231Every change is in the audit log (`cost_mapping`).
232
233## Which raw meters are operations
234
235There is one mapping, and the repos service owns it: `operation_mapping`
236in g1t-repos' database, one row per raw meter with `cost_operations` (how
237many operations Cloudflare bills for it) and `billable_operations` (how
238many the customer is charged for). Change it with repos'
239`set_operation_mapping` RPC (services only), or
240`npx wrangler d1 execute g1t-repos --remote` until sudo has a form. A
241change applies to counts from then on, never to what was counted.
242Billing keeps no mapping of its own: it reads repos' `git_operations`
243(already mapped) for what customers are charged, and `artifacts_usage`
244(raw counts with the mapping) for `cost_operations`. Migration 0023 drops
245the `billable_units` table 0022 made for this, which was never written.
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily246
Models' margin read -14%: usage nothing paid for is valued at price, not $0247**Charged at price** (a product's value) is what each day's usage was paid:
248charged to a card or credit, or drawn from the plan's included usage, a
249trial, a pool or a gift. Usage nothing paid for, as in a free period, is
250valued at price (cost plus the margin), since it was given away at its price
251rather than sold for nothing; so are g1t's own workspaces. Runs on a
252workspace's own model provider have no cost to g1t. Every daily run
253reconciles the whole 31-day window from what is already kept, so a change in
254how a day is valued reaches every day sudo shows.
255
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily256## Alerts runbook
257
258| Alert | Raised when | First steps |
259| --- | --- | --- |
260| Margin under the floor | A product's value against cost under `margin_floor_percent` (10%) for `alert_days` (3) days running, each with at least `min_daily_cost` | Open the product on Costs & margin. Cost up? Check proposals (approve a rise; it waits out the notice). Value down? A mapping or `revenue_map` may have moved. |
Costs: margin is measured on what was sold; comped workspaces, free periods, the trial and the pools are given away, a budget shown beside it261| All of g1t under the floor | The same for money in against the cost of what was sold: every cost less what was given away (comped workspaces, free periods, the trial, the pools), which is a budget watched in budget.rs. While less than $1 a day comes in, it says the dollars, not a percentage | Look at which products moved; check `platform` (it has no revenue of its own and grows with traffic). Before launch, with little paid usage, expect it. |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily262| Leak | Drift of kind leak | Map the meter, or decide it is overhead. |
263| Drift | Count drift | See Drift above. Cloudflare's definitions change in beta: ask them in writing ([ARTIFACTS.md](ARTIFACTS.md), §7). |
Margin alerts measure what is sold, and say dollars when a percentage would mislead264| Costs more than it pays | A workspace's shared cost over 30 days above what its usage was priced at (`value_micros`, whoever paid: card, trial, gift or included usage) × `anomaly_factor`, at least `anomaly_floor`; not comped workspaces | Shown on Reach out as "Costs more than it pays": its usage is priced below what it costs. Abuse (Abuse & fraud page) or a gap in pricing. Not emailed. A trial or gift paying for usage does not raise it. |
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily265
266Alerts close on their own when the condition clears. Open ones are
267emailed again weekly. The red bar on every sudo page shows margin,
268overall and leak alerts.
269
Merge branch 'worktree-agent-a633ac0f7f66d419d'270## Model costs
271
272Every model call g1t pays for is an agent run's (the `claude` CLI in the
Merge the AI Gateway: Anthropic's Messages API on a workspace's tokens273sandbox, `crates/runner`) or a customer's AI Gateway request (a workspace
274token with `models:write` at `models.g1t.sh/anthropic`, `gateway.rs`); the
275only other model is Workers AI's embeddings, which are on Cloudflare's bill
276(`embeddings`). How each reaches the ledger:
Merge branch 'worktree-agent-a633ac0f7f66d419d'277
278| Call | Who pays | Run and session | Ledger cost | Settled to the gateway |
279| --- | --- | --- | --- | --- |
280| Agent run through the model proxy (`services/models`) on g1t's hosted models | g1t | `runs` row; session `ms_…` in `cf-aig-metadata` | On finish, the sandbox's figure (Claude Code's `total_cost_usd`, at its own price table, cache tokens included) | Yes, every 15 minutes |
281| Agent run straight to the gateway (no `MODELS_URL`) | g1t | `runs` row; session `rs_…` in `cf-aig-metadata` (`services/runner` `gatewaySession`) | As above | Yes |
282| Agent run with no gateway (`AI_GATEWAY_ID` empty, self-hosting) | g1t's key | `runs` row, no session | The sandbox's figure | No: nothing to settle against |
Merge branch 'model-routing'283| Agent run on a workspace's own provider | The workspace | `runs` row, `billed_to = 'workspace'`, session `ms_…` (the proxy counts its tokens by it) | No model cost (none to g1t); the agent rate on `agent_tokens_own`, line `<run>/agent-own` | No; never on g1t's gateway. Closed by the cron's `settle_own_runs`, which charges tokens counted late |
Merge the AI Gateway: Anthropic's Messages API on a workspace's tokens284| AI Gateway request on g1t's key | The workspace, from included usage and AI credit (`gateway_admit` refuses at $0, over the limit, or with no plan) | No run; `gateway_requests` row; session `gw_<token>_<YYYYMMDDHH>` in `cf-aig-metadata` (`task: gateway`) | Per request: its tokens at `gateway_models` (the table, migration 0045: list price per million by kind) plus the `gateway_models` meter's markup (0 in beta); a `usage` line with task `gateway`, reference the request id, quantity 1 | No, not yet: the session tag is there for it. **Cost** drift on `models` compares the gateway's total with these lines too. Requests billed other than by tokens (fast mode, `inference_geo`, fallbacks, server tools, containers) are refused by the proxy |
285| AI Gateway request on a workspace's own key | The workspace's provider | `gateway_requests` row, `own_key = 1` | None | No; never on g1t's gateway |
Merge branch 'worktree-agent-a633ac0f7f66d419d'286| A sandbox that died before reporting | g1t | as its route | Charged from the gateway when settled | Yes |
287| Embeddings (indexing) | g1t | none (Workers AI) | Month-end `context` meter | No: Cloudflare's bill, `embeddings` bucket |
288| Embeddings (queries, search and agent context) | g1t | none | None: not charged, by design | No: in Cloudflare's `embeddings` line, shared out |
289
290**Settling.** A run's charge is corrected to what AI Gateway priced its
291session's requests at (`settled_cost` in `keeper.rs`). The gateway's
292figure is trusted in full: it is not held to the $100 cap on a sandbox's
293own report. It is never taken below what the sandbox reported when it
294cannot be the whole cost: a request with tokens and no cost (a model the
295gateway has no price for) or more logs than are read (2,000). Such a run
296keeps `runs.gateway_note`, its correction says why, and it raises the
Merge the AI Gateway: Anthropic's Messages API on a workspace's tokens297**Unpriced** drift. For agent runs g1t keeps no token rates of its own:
298the first figure is Claude Code's, the final one the gateway's. AI Gateway
299requests are the exception: they are charged from `gateway_models` (one
300row per model offered), which has to follow the provider's price list by
301hand until they are settled like runs.
Merge branch 'worktree-agent-a633ac0f7f66d419d'302
303**The daily total.** AI Gateway's analytics for the day (above) against
304the ledger's model cost is the check that nothing slips past: a model call
305with no run, or a run never settled, shows as **Cost** drift on `models`.
306The gateway's per-request `cost` is its estimate from its own price list:
307it can be off for prompt-cache tokens, for requests Cloudflare bills
308itself, and for models it has no price for. The drift's detail says when
309any of those were in the window; the provider's invoice is the last word.
310
311### Margin floor
312
313A sold charge is cost × (1 + `MARGIN_PERCENT`), rounded up (`margin_on`;
314`charge_micros` for a sandbox's own report). Terms change it only as
315follows (`Terms::discounted`, `Billing::charged`):
316
317- **Standard**: charged in full.
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging318- **A 100% discount** (what was "comped"; see [Discounts](#discounts)),
319 `FREE_WHILE_BUILDING`, the plan's included usage, the trial,
Merge branch 'worktree-agent-a633ac0f7f66d419d'320 the open-source pool, and overruns g1t covers: given, and counted by why
321 (above).
322- **Custom, with a discount**: the discount comes off, and what it took
323 below cost plus the margin is written on the entry as
324 `ledger.discount_micros` and counted as given (**discount**), so the sale
325 is valued at its price and charged plus given is never under cost plus the
326 margin. On a settlement correction it moves with the charge (less than
327 nothing when the charge comes down).
328- **Goodwill credits** (overages) are separate, given by staff on purpose:
329 their margin part first, the cost only up to the cap, each audited.
330
331Every usage path goes through this: `finish_run`, settling, sandbox time,
332features and builds (`charge_feature`), and the month-end meters.
333
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging334## Discounts
335
336An account's terms are standard, or custom: a **discount** from 1 to 100%,
337a limit of its own, or both, with a reason (the terms' note) and an
338optional end date. What used to be "comped" is a **100% discount**
339(`Terms::full_discount`; migration `0039_discounts_not_comped.sql` moved
340every `comped` row to `custom` at 100%, and code reads a leftover `comped`
341row as 100%). In SQL, `sales::FULL_DISCOUNT_SQL`.
342
343- **Charging.** Every charge records what the discount took off it
344 (`ledger.discount_micros`), 100% included: the entry is charged nothing
345 and the discount is its whole price. Migration 0039 backfilled the
346 discount on a 100%-discounted workspace's earlier entries that were
347 charged nothing and paid by nothing, at cost plus 20%.
348- **What a 100% discount still does as "comped" did.** The plan is on
349 without its price (`PlanKind::Internal`), trust is `internal` (no limit
350 on unpaid usage), nothing is invoiced or closed, and g1t's own spend on
351 it is held to the monthly budget (the terms' limit, at cost; see
352 [Spend caps](#spend-caps)).
353- **The statement and Usage.** The customer sees every usage line at its
354 price (`StatementLine.price_micros`: charged, plus what paid for it, plus
355 the discount), the discount per day or project and in the totals
356 (`StatementTotals.price_micros`, `discount_micros`, `discount_percent`),
357 and the CSV has price and discount columns. The Usage page measures at
358 price for a discounted account (`Usage.discount_micros`,
359 `discount_percent`).
360- **Margin.** A 100% discount's usage is given away as before, in the
361 bucket still named `comped` (`given_comped_micros`); sudo calls it
362 **100% discounts**. A partial discount's part below cost plus the margin
363 is `given_discount_micros` (**partial discounts**). Both are kept apart
364 from margin on what was sold.
365- **sudo.** The workspace's **Terms** form takes a discount (None, 25%,
366 50%, 100%, or Custom, a whole percent; the custom field shows by CSS
367 alone), a limit, an end date and the reason. Badges and filters say
368 *100% discount* or *N% off*. Each change is audited (`terms`), such as
369 `standard → 100% discount, monthly budget $150.00: g1t's own`.
370
371## Credits from g1t
372
373Staff give a workspace credit from sudo; the code is
374`services/billing/src/grants.rs`, the tables `credit_grants` and
375`ledger.credit_kind` (migration `0038_staff_credits.sql`).
376
377### Giving credit
378
379sudo → the workspace (or an enterprise, choosing one of its workspaces) →
380**Give credit**:
381
3821. **Amount**: $10, $20, $25, $50, $100, or **Custom** (up to $10,000).
383 Up to $100 it is one step; over $100, type the workspace's slug as well.
3842. **Kind**: promotional (a welcome, a referral, an event), goodwill (an
385 apology), or refund (money back for something that went wrong: say what
386 it refunds and, optionally, the day).
3873. **Expires**: never, 30, 90 or 365 days, or the end of a chosen day
388 (UTC). A refund never expires.
3894. **Note**: required. It is on the statement and in the owners' email.
390
391The form needs no JavaScript: the fields for one choice (the custom amount,
392a refund's details, the expiry date) show by CSS alone, and all show where
393`:has()` is not supported. `admin_credit` checks everything again.
394
395A grant is a `crd_…` ledger line (kind `top_up`, so never a payment) with
396`credit_kind`, and a `credit_grants` row. The balance rises at once. The
397owners are emailed through identity's `notify_owners` (the same path as
398limit notices). It is audited as `credit`. The Overages queue's one-click
399goodwill credit is a grant too, of kind goodwill.
400
401The inbox is not told: its items are threads on a repository, built from
402events, and a credit is a workspace's. That needs a workspace-level inbox
403thread first.
404
405### How it is spent
406
407Credit is spent before anything prepaid, the soonest-expiring grant first
408(never-expiring last, then the oldest). Given while the workspace owes, it
409pays what is owed first, the most recent usage first. What each grant paid
410for is never stored: `grants::replay` works it out from the ledger in order,
411so the charge paths do not know about credit and the answer is always what
412the ledger says. A charge that comes down (a settled run) gives back to
413the grant that paid last, while it can still be spent.
414
415### Expiry and revoking
416
417- **Expiry.** The daily run (`expire_credits`, before the reconciliation)
418 closes grants past `expires_at` and enters what was left as a negative
419 `crd…_expired` line; audited as `credit_expired`. A grant past its expiry
420 pays for nothing even before the run.
421- **Revoke.** sudo → the workspace's **Credits** (or **Credits & refunds**)
422 → **Revoke unused**, with why (`admin_revoke_credit`): what is left, as a
423 `crd…_revoked` line, audited as `credit_revoked`. What was spent stays
424 spent.
425
426Neither takes the balance below zero: at most the balance, if a refunded
427payment left less there than the credit.
428
429### How margin treats them
430
431| Kind | When spent | On the day it was given |
432| --- | --- | --- |
433| Promotional | The usage is valued at its price, its charge comes out of cash and is given (`given_credit_promotional_micros`) | Nothing |
434| Goodwill | The same, as `given_credit_goodwill_micros` | Nothing |
435| Refund | Paid for: cash, as any usage | Its amount (less what was revoked) comes off cash on the day it refunds, shared over that day's paid usage |
436
437A refund gives back money already collected, so counting it as given would
438make it look like a budget g1t chose to spend. Taking it off cash for the
439day it refunds says that day's sale was worth less, and counting what it
440later pays for as cash keeps money in equal to what was collected. The
441refund's day is clamped to the last 30 days, the days the reconciliation
442recomputes; an older one lands on the oldest. Refunds never expire, so
443cash taken back is never stranded.
444
445What credit paid of a month-end meter (storage, git, scans) is its own
446row on the day it was charged, since those meters are reconciled from
447snapshots. Sudo's Costs & margin lists promotional and goodwill credit
448under **Given away**, and below the statement the range's credits given,
449spent, and refunded. **Credits & refunds** (`/credits`, `admin_credits`)
450lists every grant (by kind, month, staff and workspace) and the last 12
451months by kind: given, spent, expired, revoked.
452
Usage, Billing settings and prepaid AI credit; fixes from the UX audit453### Purchased and scoped credit (prepaid AI)
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging454
Usage, Billing settings and prepaid AI credit; fixes from the UX audit455`credit_grants` also has `scope` (`all`, or `models`: model usage only,
456`grants::is_model_usage`, which includes the agent rate) and `source`
457(`staff`, `purchase`, `promo_code`, `upgrade`), and `CreditKind::Purchased`.
458Spending takes credit scoped to models first, then the soonest-expiring. A
459grant scoped to models given while the workspace owes pays only what models
460owed, never other usage (`replay`). Purchased credit is money paid in: its
461ledger line is a payment (Stripe's id, never `crd…`, `credit_kind`
462`purchased`, statement kind *AI credit*), and the usage it pays for stays
463money in, never given. Staff cannot give it (`admin_credit` refuses the
464kind). Code: `services/billing/src/ai.rs`; migration `0040_ai_credit.sql`.
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging465
Usage, Billing settings and prepaid AI credit; fixes from the UX audit466- **Buying.** `buy_ai_credit` opens Stripe Checkout (payment mode, $10 to
467 $1,000, a second line *Card processing fee* when the `card_fee` cost
468 setting is on, `setup_future_usage=off_session`), recorded in `checkouts`
469 with `feature = 'ai_credit'`, `amount_cents` the credit and `fee_cents`
470 the fee. The credit is entered by whichever comes first, the person coming
471 back (`confirm_ai_credit`, `?ai_credit=cs_…`) or
472 `checkout.session.completed`: both claim the row `open → paid`, the
473 grant's id is the session's id (`INSERT OR IGNORE`) and the ledger's
474 reference is unique, so a payment is credited exactly once. Expires 365
475 days after purchase (the daily `expire_credits`).
476- **Owed.** AI credit props up the balance but is money only for models, so
477 what is owed is `max(0, AI credit left − balance)` (`owed_with`; at a
478 month's close, `models_left_before` the month's start).
479- **Auto-reload.** `ai_reload` (settings; off by default) and `ai_reloads`
480 (one row per attempt). Each cron run (and a run that would be refused)
481 calls `reload_now`: below the threshold, it charges the customer's default
482 payment method off-session for the target less the balance (whole
483 dollars, at least $10, within the month's maximum), with the idempotency
484 key `reload/<workspace>/<YYYY-MM>/<n>` (a retry after a crash is the same
485 PaymentIntent), and grants purchased credit with the PaymentIntent's id. A
486 decline or a payment needing the person turns auto-reload off
487 (`failed_at`, `error`), emails the owners and audits `ai_reload_failed`.
488- **At $0.** `start_run` on g1t's models refuses with `payment_required`
489 when the workspace is on the paid plan (not a 100% discount, not an
490 enterprise), its included usage is used, and AI credit is $0 or less.
491- **Upgrade credit.** The first time a plan subscription is recorded active
492 (`features::record`), $5 of promotional credit scoped to models, id
493 `crd_upgrade_<workspace>`, expiring in a year: given, never revenue. Never
494 for a workspace with a 100% discount.
495
496### The agent rate and models' markup
497
498Price-book meters (migration 0040, each with versions and a public change):
499`agent_models` (per provider dollar; markup 20% until 2026-10-08, then 0,
500a fall applied at once), `agent_tokens` ($0 until 2026-10-22, then $0.25 a
501million tokens: a rise, after the 14 days' notice, emailed to owners on the
502plan by `tell_owners_of_rises`), `gateway_models` (markup 0 during beta),
503`card_fee_percent` (29,000 micros per dollar) and `card_fee_fixed`
504(300,000). Changing any is a price-book change, never a deploy. `finish_run`
505and `settle` charge models at `agent_models`' markup; `charge_agent_rate`
Merge branch 'model-routing'506charges the weighted tokens of the run's session since it was last charged
507(`runs.agent_tokens`, the weighted tokens charged so far, claimed with a
508compare-and-set), on a line `<run>/agent` (later `<run>/agent/<tokens>`),
509with `quantity` the weighted tokens. What it counts is the more of what
510`token_usage` holds for the session and what the sandbox reported with its
511cost (`finish_run`'s `tokens`, from Claude Code's closing `usage`), each
512weighted by kind. **Card fee switch:** sudo → Costs → Guardrails → *Card fee
Merge Stripe Tax, the card fee on card payments, and one free workspace per person513on card payments* (`cost_settings.card_fee`, `on`/`off`, on by default). It
514covers every card payment now, not only AI credit: see
515[Tax and the card fee](#tax-and-the-card-fee).
Usage, Billing settings and prepaid AI credit; fixes from the UX audit516
Merge branch 'model-routing'517**On a workspace's own model key** (migration `0041_agent_rate_own_key.sql`):
518the run keeps its model session (`runs.session_id`, `ms_…`) so the proxy's
519counts reach it, and the agent rate is charged at `agent_tokens_own` ($0 until
5202026-10-22, then $0.25 a million, a rise from nothing with its notice), on
521`<run>/agent-own` (later `<run>/agent-own/<tokens>`), `billed_to = 'g1t'`
522(g1t's own charge: it counts toward limits and spend), named *Agent rate,
523your own model key* on Usage and the statement. The model is never charged.
524`settle_runs` skips these runs (nothing on g1t's gateway); the cron's
525`settle_own_runs` closes them 5 minutes after they finish (3 hours after
526they start, for a sandbox that never reported) and charges tokens counted
527late. Runs from before have no session and are never charged the rate.
528
529### The agent rate's token weights
530
531How much each kind of token counts toward the agent rate, on g1t's models
532and own keys alike, is four price-book meters (migration
533`0042_agent_rate_weights.sql`): `agent_token_weight_input`, `_output`,
534`_cache_read` and `_cache_write`, each a weight in millionths in
Billing: cache reads count a tenth toward the agent rate535`cost_micros` (1,000,000 counts a token once). They started at 1, which is
536what the rate always counted; since 2026-10-08 cache reads count a tenth
537(100,000; migration `0044_cache_reads_count_a_tenth.sql`), as model providers
538price them. Input, output and cache writes count once. A cached agent run reads most of its context from
Merge branch 'model-routing'539cache (about 90% of its tokens on a typical Sonnet implement run), so the
540cache-read weight is the lever: at 1 the rate adds about 44% to such a run's
541model cost; at 0.1, far less.
542
Billing: cache reads count a tenth toward the agent rate543To change a weight (cache reads went to a tenth this way):
Merge branch 'model-routing'544
5451. Add the version, effective at once (a lower weight is a fall):
546
547 ```sql
548 INSERT INTO price_versions (id, meter, version, cost_micros, markup_percent, effective_at, reason, created_by, created_at)
549 VALUES ('pv_agent_token_weight_cache_read_2', 'agent_token_weight_cache_read', 2, 100000, 0,
550 '2026-10-08T00:00:00Z', 'Cache reads count a tenth toward the agent rate', 'staff', '2026-10-08T00:00:00Z');
551 ```
552
553 in a migration, or with `npx wrangler d1 execute g1t-billing --remote`
554 until sudo has a form.
5552. The daily run applies it once `effective_at` has come and writes the
556 public `price_changes` record (**Run the analysis now** applies it at
557 once). Raising a weight later is a rise: give it an `effective_at` 14
558 days out, and owners on the plan are emailed.
5593. Check `/pricing`: the agent rate's row lists the weights, and Usage's
560 agent-rate lines name them.
561
562`weighted` (`ai.rs`) rounds down to a whole token; a weight is never below 0.
563
Usage, Billing settings and prepaid AI credit; fixes from the UX audit564### Budgets
565
566The owners' spend limit is the budget. `limits.alert_levels` (comma
567separated, default every level), `limits.pause_at_limit` (default 1; off,
568100% is a warning, never a stop; the trust ceiling still stops work) and
569`limits.budget_webhook` (an https address, not g1t's; posted once per alert
570level a month by `warn_limits`). `set_budget` sets them, with `keepLimit`
571to leave the limit itself alone. A free workspace (trust `new`) has no
572spend limit to set: `set_spend_limit` and `set_budget` say so.
573
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging574### Earlier credits
575
576Migration 0038 makes every earlier `Credit from g1t:` line a goodwill
577grant with no expiry: the old form asked for "a refund or goodwill" with
578no way to tell them apart, and goodwill never reads as money in.
579
Mission control shows model usage, yours and the workspace's: tokens, cost, active days, cache share, each day, and the mix580## Token usage
581
582The model proxy (`services/models`) reads Anthropic's `usage` from every
583`/v1/messages` answer, streamed or whole, on g1t's models and on a
584workspace's own provider alike (OpenAI-shaped providers are translated
585first). Count-tokens requests are not answers and are skipped. After the
586answer, it calls `record_tokens`, which adds input, output, cache reads and
587cache writes to one row per day, workspace, person, session and model in
588`token_usage` (migration `0030_token_usage.sql`). The person is who the run
589was for, from the model session's `requested_by`; never g1t's agent. A
590report that fails is dropped and never affects the answer.
591
592`token_usage` reads a window (42 days by default, 366 at most) for the
593workspace or one person: totals, every day's tokens and the active days,
594with `costMicros` the window's run charges from the ledger, measured as
Merge branch 'model-routing'595`usage` measures them. Usage's report also lists tokens by model
596(`UsageReport.models`). The agent rate is charged on these counts (above);
597a run's model is still priced from AI Gateway's logs, never from them.
Mission control shows model usage, yours and the workspace's: tokens, cost, active days, cache share, each day, and the mix598
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily599## Tables (migration `0022_costs_and_margin.sql`)
600
One operation mapping, owned by repos; billing reads it instead of keeping its own601`cost_lines`, `cost_map`, `revenue_map`, `own_counts`,
Git storage hardened, pages in tens of milliseconds, honest security alerts, and costs reconciled daily602`pending_days`, `margin_days`, `workspace_costs`, `cost_drift`,
603`margin_alerts`, `price_versions` (seeded with every current price as
604version 1), `price_proposals`, `price_notices`, `cost_settings` (the
605guardrails, seeded), the `actions_cache` price, and `ledger.price_version`.
606Every create is `IF NOT EXISTS` and every seed `INSERT OR IGNORE`; the one
One operation mapping, owned by repos; billing reads it instead of keeping its own607`ALTER` is applied once by D1's migration tracking. Migration
608`0023_one_operation_mapping.sql` drops `billable_units` (see above).
Merge branch 'worktree-agent-a633ac0f7f66d419d'609Migration `0036_model_costs_in_full.sql` adds `ledger.discount_micros`,
610`margin_days.given_discount_micros`, `runs.gateway_note` and the
Billing: credits with a kind and expiry, discounts instead of comped, and safer charging611`ai_gateway_requests` → `models` mapping. Migration
612`0038_staff_credits.sql` adds `credit_grants`, `ledger.credit_kind` and
613`margin_days.given_credit_{promotional,goodwill}_micros`, and backfills
614earlier credits (see [Credits from g1t](#credits-from-g1t)).
Spend caps: a monthly budget for comped workspaces and a daily breaker on what g1t pays615
616## Spend caps
617
618Two caps keep what g1t pays for itself bounded while billing takes no
619real money. Both are measured at **cost** (what Cloudflare and the model
620providers charge g1t), never at price. Code: `services/billing/src/budget.rs`.
621Page: sudo **Costs & margin** → **g1t's own spend** (`/costs#spend`).
622
623### What counts as g1t's own spend
624
625Every charge that settles (an agent run's model cost from `finish_run` or
626AI Gateway's settlement, sandbox time from `record_sandbox`, a build from
627`charge_feature`) is split by what paid for it and g1t's part is added to
628`g1t_spend` (day, bucket, billing account):
629
630| Bucket | What |
631| --- | --- |
632| `comped` | All of a comped account's work (flagon-io) |
633| `trial` | The trial credit's share |
634| `oss` | The open-source pool's share |
635| `given` | A free workspace's overrun past its last bit of trial |
636| `unpaid` | Charged, but with no real money behind it: Stripe's test key, or `FREE_WHILE_BUILDING` |
637
638The plan's included usage and on-demand charges count as revenue only
639with live payments; in test mode they are `unpaid`. A workspace's own
640model provider costs g1t nothing and is not counted. Month-end meters
641(git, storage, scans, embeddings, the cache) are not counted here; the
642daily reconciliation covers them. Migration `0024_spend_caps.sql`
643backfills the current month from the ledger.
644
645### Caps
646
647| Cap | Variable (g1t-billing) | Default | At the cap |
648| --- | --- | --- | --- |
649| A comped account's monthly budget | `COMPED_MONTHLY_CEILING_MICROS`, or the account's own **Limit** in its terms | $150 a month | New work on the account (agents, checks, workflows, builds) is refused with "<name>'s monthly budget for g1t's own agents is used up … Staff can raise it in sudo". Runs already going finish; the per-run cap still applies to them. It lifts when staff raise the budget or the month turns (UTC). |
650| The daily breaker | `PLATFORM_DAILY_SPEND_CAP_MICROS` | $75 a day (UTC) | New agent runs on g1t's hosted models that g1t would pay for are refused until 00:00 UTC. Not paused: agents on the workspace's own model provider, checks and builds, and workspaces paying with live payments on the plan (not given by staff) or an enterprise contract. In test mode that exemption covers no one. |
651
Costs: Cloudflare's subscriptions read from Cloudflare each day, the estimate only until then; sudo's costs split into Costs & margin and Bill & pricing652`0` turns either off. Cloudflare's subscriptions are read from Cloudflare
653each day (`cf_subscriptions`) and shown on the page only;
654`CLOUDFLARE_FIXED_MONTHLY_MICROS` ($30) stands in until a read works.
Spend caps: a monthly budget for comped workspaces and a daily breaker on what g1t pays655
656The checks are cheap: `reserve` reads today's total (one indexed sum) and,
657for a comped account, its month's comped rows. Refusals come back as
658`paused`, which the compute gate honours for every plan, internal and
659enterprise included (`packages/contracts/src/compute.ts`). The runner tells
660billing whether an agent run is on hosted models (`hostedModel` on
661`reserve`); a caller that does not say is treated as hosted.
662
663### Alerts
664
665All to `COSTS_ALERT_EMAIL` (`hey@flagon.io`), through the `EMAIL` binding:
666
667- **Comped budget**: at 50, 75, 90 and 100%, once each per account and month
668 (`budget_alerts`), checked every 15 minutes. A jump past several levels
669 sends only the highest.
670- **Breaker**: at once, from the charge that trips it; if that email fails,
671 the 15-minute cron sends it (`spend_breaker.told_at`).
672
673While the breaker is open or a comped budget is used up, every sudo page
674shows a red **Spend cap** bar.
675
676### Raising and lifting
677
678- **Raise a comped budget**: sudo → the workspace → Billing → **Terms**, set
679 **Limit $** to the new monthly budget (blank goes back to the default),
680 with a note. It applies to the next start; nothing to deploy. The change
681 is in the account's audit log.
682- **Lift the breaker for today**: sudo → Costs & margin → **g1t's own
683 spend** → **Lift for today**, with why (`admin_lift_breaker`; audit action
684 `breaker_lifted`). It resets by itself at 00:00 UTC.
685- **Change a default**: edit the variable in `services/billing/wrangler.jsonc`
686 and deploy g1t-billing.
Billing keeps Stripe's view itself: the saved card on the account, missed events replayed every 15 minutes, and the endpoint kept687
sudo: reset a test workspace's billing so it starts again as a new customer; refused on a live Stripe key, for comped workspaces and for an enterprise's688## Resetting a test workspace
689
690sudo → the workspace → **Reset billing (testing)** (`admin_reset_billing`)
691returns a workspace used for testing to how a new customer starts. It
692deletes the workspace's rows from every billing table: ledger and balance,
693plan and plan payments, limits and limit requests, trial grant, invoices,
694holds, card checks, alerts sent, price notices, month-end snapshots and
sudo: the billing reset no longer names sandbox_months (dropped in 0015), checked against the migrations by a test; a failed reset says why instead of an error page695closes, storage meters, token usage, spikes, sales records and
sudo: reset a test workspace's billing so it starts again as a new customer; refused on a live Stripe key, for comped workspaces and for an enterprise's696notes, `workspace_costs`, its workspace margin alert and its own billing
697account. It keeps `own_counts` (what Cloudflare's bill is compared with)
698and the audit log, which records the reset with the note and the number of
699rows. The workspace, its members and its repositories are identity's and
700repos' and stay.
701
702Billing refuses it while `STRIPE_SECRET_KEY` is a live key, for comped
sudo: a billing reset runs the costs analysis again so every figure is fresh; every submit button shows it is working (CSS only); no margin percentage on less than a cent sold703workspaces, and for a workspace an enterprise pays for. It then runs the
704costs analysis again (as **Run the analysis now** does), so the margin
705figures drop the workspace's past usage at once; if that run does not
706finish, the page says so and the button does it.
sudo: reset a test workspace's billing so it starts again as a new customer; refused on a live Stripe key, for comped workspaces and for an enterprise's707
Billing keeps Stripe's view itself: the saved card on the account, missed events replayed every 15 minutes, and the endpoint kept708## Stripe
709
710Billing keeps what it needs from Stripe so reads never wait on it, and
711hears of changes three ways (`webhooks.rs`, `stripe_sync.rs`).
712
Usage, Billing settings and prepaid AI credit; fixes from the UX audit713**API version.** Every request sends `Stripe-Version: 2025-02-24.acacia`
714(`stripe::STRIPE_VERSION`), the version billing's field reads are written
715for; without it Stripe answers at the account's default. Webhook events
716come at the destination's own version: billing reads an invoice's
717subscription from `subscription` or `parent.subscription_details.subscription`.
718Raising the version is a code change: read Stripe's upgrade notes for every
719field billing reads.
720
721**Failures.** No Stripe failure reaches a page as a 500: each payment page
722(`page_opened`), the portal, confirmations and plan changes turn it into
723`stripe::friendly` (Stripe's own message, never the request or a key), and
724log the full error with the workspace. Every payment page is recorded
725through one insert (`CHECKOUT_INSERT`), checked against the migrations by
726`every_checkout_insert_fills_the_table`, and has an idempotency key
727(`page/<purpose>/<workspace>/…/<10-minute bucket>`).
728
Stripe's webhook secret is a Worker secret, STRIPE_WEBHOOK_SECRET, from a destination made in Stripe's dashboard729**The webhook.** A destination made in Stripe's dashboard (Developers →
730Webhooks → Add destination) with the endpoint URL
731`https://api.g1t.sh/stripe/webhook`, in the mode of billing's key (at
732launch, make one in live mode and put its secret). Its signing secret
733(`whsec_…`: the destination, Signing secret, Reveal) is the billing
734Worker's secret:
735
736```sh
737cd services/billing && npx wrangler secret put STRIPE_WEBHOOK_SECRET
738```
739
740Without it every event is refused with 400. After rolling the secret in
741Stripe, put the new one; during the roll Stripe signs with both, so there
742is no gap. The event list need not be exact: billing adds any event it
743handles that the destination does not send (daily, or **Fix destination**
744in sudo → Stripe), and enables it again if Stripe disabled it. It never
745changes the secret. sudo → Stripe shows whether the secret is set, the
746destination and its status, missing events, and the latest events.
747Events are claimed once each in `stripe_events`; a handler that fails
748forgets its claim, and Stripe retries.
Billing keeps Stripe's view itself: the saved card on the account, missed events replayed every 15 minutes, and the endpoint kept749
750**What is kept, and how it stays current**
751
752| Kept | Where | Refreshed by |
753| --- | --- | --- |
754| Saved card (brand, last 4, expiry) | `accounts.card_*`, `card_synced_at` | `customer.updated`, `payment_method.*`, `setup_intent.succeeded`; forgotten when g1t sets a default card, so the next read asks once; the daily pass for cards older than 7 days |
755| Plans | `subscriptions` | `customer.subscription.*`, `invoice.*`; a read asks Stripe at most hourly once a period is over; the daily pass for rows older than a day |
756| Payments, refunds, disputes | ledger, `checkouts`, invoices | their events |
757
758**Every cron run (every 15 minutes)** replays missed events: Stripe's event
759list from an hour before `stripe_sync.through`, oldest first, through the
760same once-only claim. `through` moves to 5 minutes before now when all were
761handled, back to the first failure otherwise, and stays when more than
7621,000 events were listed. Claims stuck at `handling` for 10 minutes are
763dropped so the replay retries them. The first run reads 3 days back.
764
Stripe's webhook secret is a Worker secret, STRIPE_WEBHOOK_SECRET, from a destination made in Stripe's dashboard765**Daily** (`keeper::DAILY`): the destination at billing's address is
766enabled again if Stripe disabled it and given any missing event, audited as
767`stripe`/`webhook`; then up to 25 stale cards and 25 stale plans are read
768again.
Billing keeps Stripe's view itself: the saved card on the account, missed events replayed every 15 minutes, and the endpoint kept769
770**What still calls Stripe on a request**: starting a payment page, a plan
771or a card check; opening the billing portal; settling a page the person
772came back from; renaming a workspace (the customer's name). Nothing a page
773view reads.
Teams and CODEOWNERS, labels and milestones, dependency updates, the security suite, and a clearer top bar774
Merge Stripe Tax, the card fee on card payments, and one free workspace per person775## Tax and the card fee
776
777Owner decision 2026-10-08. Code: `services/billing/src/tax.rs` (what is
778kept, the address hold), `stripe.rs` (every request's tax fields),
779`invoices.rs`, `webhooks.rs`, `ai.rs`; migration
780`0043_tax_and_card_fees.sql`.
781
782### What Stripe is asked
783
784Every price excludes tax (`tax_behavior=exclusive`) and carries tax code
785`txcd_10103001` (software as a service, business use; the card fee too,
786since a fee for paying for a sale follows the sale). Fields are valid for
787`2025-02-24.acacia`.
788
789| Request | Tax | Card fee |
790| --- | --- | --- |
791| Checkout, plan or Security (`subscription_fields`) | `automatic_tax[enabled]`, `billing_address_collection=required`, `tax_id_collection[enabled]`, with a customer `customer_update[address]=auto` and `[name]=auto`; the subscription keeps automatic tax for every renewal | A second recurring line, *Card processing fee* |
792| Subscription on a saved card (`saved_subscription_fields`) | `automatic_tax[enabled]`; a customer Stripe Tax cannot place fails, and the person is sent to Checkout, which asks for the address | `items[1]`, the `card_fee` product |
793| Checkout, prepay (`prepay_fields`) | As above | By card only; none by bank transfer |
794| Checkout, AI credit (`credit_fields`) | As above | Its own line |
795| Checkout, card check (`card_check_fields`) | Setup mode: nothing charged, nothing taxed. `billing_address_collection=required`, and the card's address is copied onto a customer with none (`fill_address`) | — |
796| Auto-reload (`charge_saved`) | `POST /tax/calculations` first (credit and fee as lines), the PaymentIntent for the total, then `POST /tax/transactions/create_from_calculation` with the PaymentIntent as reference; a refund reverses its share (`create_reversal`, `mode=partial`) | In the calculation and the amount |
797| Workspace invoice, month close and threshold (`invoice_workspace`) | `automatic_tax[enabled]` on the draft; each item `tax_behavior`, `tax_code` | An item *Card processing fee*, when the default payment method is a card |
798| Enterprise invoice (`invoice_enterprise`) | The same | Never |
799| Products (`Stripe::product`) | Made with `tax_code`; one found without it is given it | The `card_fee` product, `metadata[g1t]=card_fee` |
800
801### What is kept
802
803A payment credits the balance with what it paid for, never its tax or fee:
804prepay credits the page's `amount_subtotal` less the fee line
805(`credit_prepayment`), a workspace invoice `amount_paid − tax − fee`
806(`credit_invoice`), AI credit its credit amount, and `plan_payments` the
807plan's invoice less its tax and *Card processing fee* lines
808(`stripe::invoice_split`). Each payment's tax and fee are rows in
809`tax_and_fees` (`<reference>/tax`, `<reference>/card_fee`; the enterprise's
810account id in `workspace` for its invoices), with the PaymentIntent, so a
811refund (`charge.refunded`) gives back the balance, tax and fee in
812proportion (`tax::refund_split`, negative rows under `refund/<charge>/…`).
813`workspace_invoices.fee_micros` and `tax_micros` sit beside each usage
814invoice.
815
816- **Statement.** *Tax* and *Card processing fees* are their own lines per
817 day (`StatementLine.passed_micros`), never in `charged_micros`; totals
818 `tax_micros`, `card_fee_micros`. The CSV has a row a day for each.
819- **Margin.** Cash never holds them, so margin is untouched. sudo → Costs
820 shows **Tax collected** and **Card fees passed on** for the range
821 (`OverallMargin.tax_collected_micros`, `card_fees_micros`). Tax is owed to
822 the authorities: file it from Stripe Tax's reports, never from g1t's.
823
824### No address
825
826Stripe Tax needs a country (in the US a ZIP code, in Canada a postal code
827or province: `stripe::address_places_customer`). Before a workspace
828invoice is drafted, g1t checks the customer; without an address, or when
829Stripe leaves the draft at `requires_location_inputs` or refuses with
830`customer_tax_location_invalid`, nothing is charged:
831`accounts.tax_address_needed_at` is set, the owners are emailed once
832(`notify_owners`), and Billing shows **Add a billing address**. Saving
833Invoice details with an address Stripe Tax can use clears it, as does a
834charge that goes through. Work is not stopped for it; the limits still
835apply. Auto-reload without an address fails like a declined card (turned
836off, owners told). An enterprise's invoice is not sent without an address:
837sudo → the enterprise → Invoices → **Billing address** (`admin_enterprise_address`,
838with its tax ID; audited `billing_address`).
839
840### The card fee
841
842`card_fee_cents` grosses Stripe's fee up so the amount paid for is left
843after it: `(amount + 30¢) / (1 − 2.9%)`, rounded up; $0.91 on $20, $1.06
844on $25. It is worked out on the amount before tax, so Stripe's fee on the
845tax itself (a few cents) is g1t's. It is shown before paying: the plan card
846and pricing page (`Plan.card_fee_cents`), AI credit (*Card processing fee
847$1.06, plus tax where it applies*), Prepay. Never on a bank transfer or an
848enterprise's (`send_invoice`) invoice. Meters stay at cost + 20% and models
849at the provider's price plus the agent rate (price versions in migration
8500040); the fee is passed through, not margin.
851
852### Tax-exempt customers and tax IDs
853
854g1t never sets `tax_exempt`. For a customer who sends an exemption
855certificate, set it in Stripe's dashboard (Customers → the customer → Tax
856status: Exempt, or Reverse charge); Billing then says so. Tax IDs come
857from Checkout (`tax_id_collection`) or Invoice details (`set_billing_details`,
858validated against Stripe's types in `details::TAX_ID_TYPES`); Stripe checks
859EU VAT numbers and Stripe Tax applies a reverse charge where it should.
860Billing shows Stripe's `verification.status`.
861
862### In Stripe's dashboard (not done by g1t)
863
8641. **Settings → Tax → Get started**: turn on Stripe Tax in live and test
865 mode.
8662. **Origin address**: Flagon, Inc.'s head office address.
8673. **Default tax code**: Software as a service, business use
868 (`txcd_10103001`); **default tax behavior**: exclusive.
8694. **Registrations**: add each jurisdiction where Flagon is registered to
870 collect (its home state at least; then states as thresholds are
871 crossed, which Stripe Tax's monitoring flags; the EU's OSS and the UK
872 if selling there). Stripe collects only where a registration exists.
8735. **Customer portal**: tax ID and address updates are already allowed
874 (`portal_configuration`).
8756. Refund an invoice through a **credit note**, so its tax is reversed in
876 Stripe Tax.
877
878## Free workspaces
879
880Owner decision 2026-10-08: one free workspace per person, and a free
881workspace adds no one. Billing answers one question, `free_workspaces`
882(`credits.rs`): which of the given workspaces are on no paid plan
883(`plan_kind_for` is `Free`). The plan, an enterprise's terms and a 100%
884discount (flagon-io) count as paid; with payments off nothing is free.
885Identity asks it (`services/identity/src/paid.rs`) and refuses with
886`payment_required`:
887
888| Where | Refused when |
889| --- | --- |
890| `create_workspace` | The person owns any free workspace. Several from before are kept (grandfathered); none can be added until each is paid for or deleted. |
891| `add_member`, `invite_member` | The workspace is free. |
892| `add_collaborator` | Someone outside a free workspace (a username who is not a member, or an address). Members' roles are fine. |
893| `accept_invite`, `respond_repo_invitation` | The invite's workspace (or repository's) is free now: it waits. A sign-up with such an invite makes the account without joining. |
894
895`@g1t` is never counted as someone added. If billing cannot be asked, the
896change is refused for now ("try again"), never let through. The site says
897so first (New workspace, People, a repository's Access, from the same RPC);
898the API and MCP pass identity's refusal on as `402`.
899
Teams and CODEOWNERS, labels and milestones, dependency updates, the security suite, and a clearer top bar900## The Security and quality activation
901
902A second monthly subscription a workspace can hold beside the plan
903(`Feature::Security`, `feature = 'security'` in `subscriptions`). It turns
904on the security suite's paid features for the workspace's private
905repositories: custom secret patterns, validity checks, delegated bypass,
906code scanning, dependency review and the security overview. Public
907repositories have them free; secret scanning, push protection,
908vulnerability alerts and security updates are free everywhere.
909
910- **Price.** The price book's `security_activation` meter (unit
911 `workspace-month`, `source` `list`, markup 0): 10,000,000 micros, $10,
912 from migration `0037_security_activation.sql` with its first
913 `price_versions` row and a public `price_changes` record. `plan()` and the
914 `prices` RPC read it (`features::security_plan_at`); if the price book
915 cannot be read, $10. Nothing in the web app hard-codes it. A change is a
916 new price version, like any other: noticed on the pricing page and
917 applied from its `effective_at` to new subscriptions. Subscriptions
918 already running keep the amount Stripe has until they are changed in
919 Stripe.
920- **Stripe.** Its own subscription and its own product, tagged
921 `metadata[g1t]=security` (the plan's is `plan`). Started from the Billing
922 page with `subscribe` (`feature: security`) on the checked card, or
923 through Checkout; ended with `cancel_subscription` (`feature: security`)
924 at the period's end. Its invoices count in `plan_payments` like the
925 plan's, as paid revenue.
926- **Who has it.** `has_feature(workspace, security)`: on with an active
927 subscription, with comped terms or as an enterprise's workspace, or when
928 Stripe is not configured. The plan's allowance (`allowances.plan`) does
929 not include it. Refusals are `PaymentRequired` with the price from the
930 price book and the Billing page's address.
931- **Where it is checked.** The security service, on each paid call for a
932 private repository (`suite::entitled`) and before using custom patterns
933 in a push (`patterns_for`). When billing cannot be reached it is taken
934 as off: a paid feature waits rather than running unpaid.
935- **Fixes.** "Fix with g1t" runs g1t's agent, charged as agent usage, never
936 to the activation.
937- **Sales figures.** MRR in sudo counts `feature = 'plan'` only; the
938 activation's subscriptions are not in it yet.

This file's history is long; its oldest lines are credited to the oldest commit read.