name: Tailcall FinOps Framework
description: >-
  FinOps cost visibility and optimization guidance for teams using Tailcall's
  open-source GraphQL runtime and/or ForgeCode commercial product. Tailcall's
  self-hosted runtime shifts infrastructure costs to the operator's cloud bill;
  ForgeCode adds a predictable SaaS subscription cost per developer seat.
version: "1.0"
created: "2026-06-13"
modified: "2026-06-13"
framework: FinOps Framework 1.0

domains:
  - name: Understand
    description: Gain visibility into Tailcall-related cloud and SaaS spend.
    practices:
      - name: Map Tailcall runtime infrastructure costs
        description: >-
          Tag all compute resources (Lambda functions, ECS tasks, EC2 instances,
          Cloudflare Workers) running the Tailcall GraphQL runtime with a shared
          cost tag (e.g., product=tailcall-gateway). This enables accurate
          attribution in AWS Cost Explorer, GCP Billing, or Azure Cost Management.
        effort: low
        tools:
          - AWS Resource Tagging
          - Cloudflare Analytics
          - Datadog APM

      - name: Track upstream API costs amplified by the gateway
        description: >-
          Because Tailcall composes multiple upstream APIs, a single GraphQL
          query can fan out to many upstream HTTP calls. Monitor upstream API
          costs (third-party API fees, egress charges) and correlate them with
          GraphQL operation traffic through Tailcall's access logs.
        effort: medium
        tools:
          - Tailcall access logs (stdout / file sink)
          - AWS Cost and Usage Report
          - Custom dashboards on request counts per upstream

      - name: Track ForgeCode SaaS spend per developer
        description: >-
          ForgeCode is billed per-seat. Maintain an inventory of active
          ForgeCode licenses (Free / Pro $20/mo / Max $100/mo) and align
          seat counts with active engineering headcount monthly.
        effort: low
        tools:
          - ForgeCode billing dashboard (forgecode.dev)
          - Internal HRIS / engineering roster

  - name: Optimize
    description: Reduce waste and right-size Tailcall infrastructure and licensing.
    practices:
      - name: Enable response caching to cut upstream API calls
        description: >-
          Use Tailcall's @cache directive to cache upstream HTTP responses.
          Cached responses avoid redundant upstream calls, reducing both
          latency and per-call upstream API fees. Set maxAge based on data
          freshness requirements per field.
        effort: low
        impact: high
        example: "@cache(maxAge: 300)"

      - name: Enable request batching to reduce N+1 fan-out
        description: >-
          Configure @batch directives on list resolvers to coalesce multiple
          upstream HTTP requests into a single batched call. This directly
          reduces upstream API call volume and associated costs.
        effort: low
        impact: high
        example: "@batch(delay: 5, maxSize: 100)"

      - name: Right-size compute for the Tailcall runtime
        description: >-
          Tailcall is written in Rust and is highly CPU- and memory-efficient.
          Benchmark your p99 latency and throughput on smaller instance types
          (e.g., ARM Graviton on AWS) before provisioning over-sized VMs.
          Use AWS Lambda@Edge or Cloudflare Workers for bursty workloads to
          avoid idle compute costs.
        effort: medium
        impact: medium

      - name: Downgrade ForgeCode seats for low-usage developers
        description: >-
          Review ForgeCode daily request utilization. Developers consistently
          under 50 requests/day may be adequately served by the Free tier,
          saving $20–$100/month per seat.
        effort: low
        impact: medium

  - name: Operate
    description: Ongoing FinOps operational cadence for Tailcall environments.
    practices:
      - name: Monthly infrastructure cost review
        description: >-
          Review tagged Tailcall gateway compute costs monthly. Compare
          against prior month and correlate with GraphQL request volume changes.
          Identify cost-per-request trends.
        cadence: monthly

      - name: Quarterly ForgeCode seat audit
        description: >-
          Audit active ForgeCode subscriptions against engineering roster.
          Remove seats for departed or inactive users. Adjust tier (Pro vs Max)
          based on actual daily usage data from the ForgeCode dashboard.
        cadence: quarterly

      - name: Upstream API spend alerting
        description: >-
          Set budget alerts on upstream API provider spend (e.g., AWS API Gateway,
          third-party REST APIs). Alert when monthly spend exceeds 110% of prior
          month's baseline to catch unexpected query fan-out spikes early.
        cadence: continuous

cost_model:
  tailcall_runtime:
    licensing: free (Apache 2.0 open source)
    infrastructure: operator-paid (cloud compute, egress, Lambda invocations)
    upstream_apis: operator-paid (varies by upstream provider pricing)

  forgecode:
    free: "$0/month — 10–50 AI requests/day"
    pro: "$20/month — 1,000 AI requests/day"
    max: "$100/month — 5,000 AI requests/day"
    billing_model: flat-rate monthly subscription per seat
    url: https://forgecode.dev/blog/graduating-from-early-access-new-pricing-tiers-available/