Roadmap

@weckr/sdk on npm. Wraps OpenAI, Anthropic, Gemini, Kimi in two lines.

weckr-sdk on PyPI. Same interface, full Python support.

app.useweckr.com. Cost, revenue, and margin in one view.

gpt-5.6, gpt-5.4-mini, and the older gpt-4o family. Token counts captured server side.

claude-opus-4-8, claude-sonnet-4-6, claude-haiku-4-5.

gemini-3.6-flash, plus the gemini-2.5 and 1.5 families.

Moonshot AI Kimi (K2, K2.5, K2.6, K3). OpenAI compatible, detected by base URL.

SDK repo passes the OpenSSF Best Practices criteria, publishes a Scorecard, runs CodeQL and Dependabot, pins actions by SHA.

See exactly which users cost more than they pay. Per LLM call. Real time.

Know which features in your app are the most expensive to run.

Amber banner on the dashboard the moment a user goes negative.

Fire-and-forget logging. Your LLM calls are never blocked by Weckr.

Monthly spend limits per plan. Block the call or downgrade automatically.

When a user crosses their cap, silently swap to a cheaper model in the same provider.

Spots features paying for a heavy model on simple tasks. Estimates the saving.

From real usage data, suggests what to charge per plan for healthy margins.

When a user burns 50k tokens in 5 minutes, Weckr fires an alert in seconds. Catches runaway agents before the bill arrives.

A Slack ping the moment the velocity threshold trips. Per-project webhook, no global config.

Same trigger as Slack, delivered via email through Resend. HTML payload with the metric + a link to the user.

Set a per-user margin floor. The moment a user tips below it, Weckr alerts you by Slack and email, with a cooldown so you never get spammed.

Every Monday, a per-project email with last week’s AI cost, your most expensive users, and the margin trend.

Set caps, alert thresholds, Slack webhook, and email recipient straight from the dashboard. No code changes.

Track multiple products under one account. Switch projects in one click.

npx -y @weckr/mcp. Ask Claude directly which users are costing you money.

Full clickable dashboard with realistic seeded data at /demo.

Interactive tool at /calculator to see if your pricing actually clears costs.

Four skills (integration, model pricing, cost estimator, margin audit) installable as one plugin, so Claude Code knows real prices and the SDK.

useweckr.com/pricing.json, the same table the API prices with, plus a weekly watcher that opens a pull request when a provider changes rates.

Ten real product features run against OpenAI, Anthropic, Gemini and Kimi, priced by the production code path, published with raw data.

useweckr.com/status, built from real recorded checks of the API and database, no synthetic green.

Light or dark for the dashboard and the site, remembered per browser.

Boundary-wrap pattern documented for LangChain users. Wrap the underlying OpenAI client; chains stay untouched.

Same boundary-wrap pattern. Loop detection works on every CrewAI agent that goes through the wrapped client.

Two MIT starters with auth, an LLM endpoint and Weckr wired: clone, deploy, see a margin row.

Stripe Checkout with a 7-day free trial, then $49/month. Cancel anytime from the customer portal.

Branded emails on trial start, subscription activation, and cancellation, so a billing change never goes silent.

Not just by user. Get pinged when a specific feature spikes.

Connect your own Stripe account so Weckr imports your real plan prices automatically. Today you set plan revenue through the SDK or dashboard.

Send cost and margin events to your own systems in real time.

Drop-in CallbackHandler so chains and agents log cost without manual boundary-wrapping. Current pattern stays supported.

Automatic cost tracking for LlamaIndex pipelines.

An alternative integration. Point your baseURL at a Weckr proxy and get cost tracking with a one line config change. For teams that prefer a proxy over an SDK.

Emit Weckr cost and margin data as standard OpenTelemetry spans. Plug into the observability stack your team already runs.

First-class support for streamed LLM responses. Chat UIs and real-time apps.

Full support for Azure-hosted OpenAI deployments.

Mistral models. Built for EU founders who need EU-hosted inference.

Cohere Command models.

Invite co-founders and team members to your Weckr dashboard.

Download all your cost and margin data as CSV or JSON anytime.

Run the Weckr backend on your own infrastructure. For security-sensitive teams.

A badge for your README showing live monthly AI spend. Open accounting for AI SaaS.