mcpField
S
spendtron

spendtron

Remote
spendtron.comCloud & Infra0Updated 3w ago
RepositoryVisit website

About spendtron

spendtron is a read-only MCP service for finding costly GitHub Actions workflows and repositories. It ranks actual compute costs and proposes fix diffs verified against run timings.

Key features

  1. 01Rank workflow and repository compute costs
  2. 02Identify costly GitHub Actions workflows
  3. 03Suggest fixes for caching and timeouts
  4. 04Verify suggestions against run timings

Use cases

  1. 01Find workflows driving CI costs
  2. 02Prioritize repositories for CI optimization
  3. 03Review actionable fixes based on run data

MCP support

Exposes the `check_actions_cost` tool to MCP clients. Access uses a GitHub App with read-only permissions and a spendtron API key.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://spendtron.com/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://spendtron.com/mcpAuth required

Tools

1

This tool takes no parameters.

Details

From the project README

spendtron

A read-only tool that finds which GitHub Actions workflow is eating your CI bill, ranks workflows and repos by real compute cost, and suggests pasteable fix diffs (missing cache, missing timeout-minutes) confirmed against real run timing — not estimates.

Delivered as an MCP server — call check_actions_cost from Claude, ChatGPT, or any MCP-compatible client, no dashboard required. Auth is a GitHub App (read-only: Actions + Contents + Metadata) plus a spendtron API key, passed as a bearer token. The cost ranking and one confirmed fix diff are free on every call. Unlocking the full fix list is a one-time $39 payment via Stripe — not a subscription.

The underlying audit logic started as ~/code/ci-audit/actions-teardown.sh (bash + gh api) — lib/auditActions.js is the same algorithm, ported to run per-installation against many customers' orgs instead of one local gh-authenticated user.

Status

Real product, not a waitlist — see HANDOFF.md/FINDINGS.md/CARRD.md in ~/code/ci-audit for the history of how this got here (an earlier waitlist-and-post-to-Reddit attempt correctly got abandoned; the audit logic itself was always real).

The GitHub App must be public (GitHub → spendtron-app → Advanced → Make this GitHub App public) so anyone can hit Connect GitHub. That toggle lives in the GitHub UI, not this repo.

Setup

  1. Copy .env.example to .env.local (or set these in Cloud Run's Secret Manager for prod).
  2. Create a GitHub App at github.com/settings/apps/new — slug spendtron-app, permissions: Actions: Read-only, Contents: Read-only, Metadata: Read-only. Callback URL: <your deployed host>/github/callback. Make the App public before sending customers to /github/install.
  3. Create a Neon Postgres database, run db/schema.sql against it.
  4. Create a one-time Stripe Price (not recurring) for the fix-diff unlock; put its ID in STRIPE_UNLOCK_PRICE_ID. Point a Stripe webhook at /stripe/webhook for checkout.session.completed. (STRIPE_PRICE_ID and the Customer Portal are legacy from an earlier $99/mo model — no longer wired into any live checkout flow.)
  5. Map both spendtron.com and www.spendtron.com to the Cloud Run service (www should CNAME to ghs.googlehosted.com at Squarespace). The app 301s www → apex.
  6. Privacy / terms for Stripe: https://spendtron.com/privacy and https://spendtron.com/terms.
  7. npm install && npm start.

Test

npm test — covers the cost-math port (must match actions-teardown.sh's bash/awk exactly) and API-key auth, against a mocked GitHub client (no live credentials needed for these).

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.