Agent skill

Frugal

by yuanboP in yuanboP/frugal

Cloud cost awareness for agents. An agent skill from yuanboP/frugal.

MITAuto-check passedDevOps & Cloud

Install Frugal

skills CLI
$ npx skills add yuanboP/frugal --skill frugal -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install yuanboP/frugal frugal --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/yuanboP/frugal.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/frugal .claude/skills/frugal && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
frugal
GitHub stars
198
Token cost
~2.1k tokens
SKILL.md length
1,008 words
Files
3 (incl. references)
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Cloud cost awareness for agents. An agent skill from yuanboP/frugal.

  • The user asks about cloud costs
  • SKILL.md covers Rules, Intensity, Free-tier cheat-sheet… and Paid-plan tripwires (fail-open…
  • Calls fly, wrangler and vercel
  • Why is my bill so high

What it does

Frugal is an agent skill from yuanboP/frugal. Cloud cost awareness for agents. Metered services (Vercel, Cloudflare Workers/R2/KV/D1, Neon, Railway, Fly.io, E2B, Browserbase, GitHub Actions/Codespaces, Supabase, AWS, GCP, Azure, Terraform/Pulumi) silently bill by usage, and the user may not understand why the bill exploded. Use whenever the user asks about cloud costs, billing, usage, quotas, free tiers, "check the bill", "why is my bill so high", reducing cloud spend, or before deploying/provisioning to any metered service. Scope is normal usage that runs…

Its SKILL.md is about 2.1k tokens, which your agent loads only when the skill is triggered. The skill folder holds 3 other files, including reference files (for example `references/indie-services.md` and `references/providers.md`).

It sits in DevOps & Cloud, covering Cloud cost optimization, Infrastructure as code and Pricing strategy. It works with Vercel, Cloudflare Workers, Supabase and Amazon Web Services. The licence is MIT.

When your agent uses it

  • The user asks about cloud costs
  • Why is my bill so high
  • Reducing cloud spend
  • Before deploying/provisioning to any metered service

Example prompts

  • “check the bill”
  • “why is my bill so high”
  • “/frugal”

What it can do on your machine

Read from SKILL.md and the folder at commit b0f07ea. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • fly
    • wrangler
    • vercel
    • aws
    • supabase
    • gh
    • az
    • heroku
    • doctl
    • claude
    • modal

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use wrangler, vercel, aws, supabase, gh and az, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Frugal loads about 2.1k tokens when it runs, and up to ~54k if it reads all its reference files. Until then it costs about 188 tokens; SKILL.md has 1,008 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~188
When it runs · the whole SKILL.md, loaded when a task matches
~2.1k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~54k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from yuanboP/frugal at commit b0f07ea, republished under its MIT licence (© yuanboP). 1,008 words, ~2,118 tokens.

Download SKILL.mdSave it as .claude/skills/frugal/SKILL.md (or your agent's skills folder). This skill also uses 2 other files; get the full folder from GitHub.
name
frugal
description
Cloud cost awareness for agents. Metered services (Vercel, Cloudflare Workers/R2/KV/D1, Neon, Railway, Fly.io, E2B, Browserbase, GitHub Actions/Codespaces, Supabase, AWS, GCP, Azure, Terraform/Pulumi) silently bill by usage, and the user may not understand why the bill exploded. Use whenever the user asks about cloud costs, billing, usage, quotas, free tiers, "check the bill", "why is my bill so high", reducing cloud spend, or before deploying/provisioning to any metered service. Scope is normal usage that runs up a bill unnoticed (wrong tier, forgotten resource, loop that hits a paid meter) — not credential leaks or fraud, that's a security concern. Do NOT use for questions about application pricing models or unrelated finance topics.
license
MIT

Frugal

You are cost-aware. Agents burn users' money silently on metered cloud services; the user may be non-technical and only discover it on the monthly bill. Every deploy, sandbox, database, and API loop you run can cost real money that is not yours.

Rules

  • First touch: one short line on how it bills (hard-pause vs fail-open). Optional dig if a cheap CLI exists — skip if no credentials. Do not lecture.
  • Lasting paid resources: one-line notice; prefer free/local first (wrangler dev, vercel dev, Neon branch, supabase start).
  • Ephemeral dies in-session: sandboxes, test resources, idle Codespaces.
  • Escalate only for scary spend: unattended loops, no-cap meters, recursive triggers, open OTP/send, real apply/provision. One sentence.
  • Alerts are not brakes; paid plans fail open unless a spend cap actually stops service — details in references/providers.md.
  • Never silently spend for the user. Work first; one-line reminders.

Intensity

LevelInject
quietFree-tier / key quota numbers only
normal (default)Numbers + one trap; dig optional
strict+ real-bill dollar context; confirm scary actions

/frugal quiet|normal|strict|off · /frugal default <level> · FRUGAL_MODE

Free-tier cheat-sheet (researched 2026-07)

Numbers below are the free-tier walls agent-built apps hit — check which plan the user is actually on before applying them. Paid plan tables, overage prices, and sources: references/providers.md (core cloud) and references/indie-services.md (Firebase, PaaS, email, auth, maps, AI APIs, GPU, observability) in this skill.

ProviderFree tier key numbersWhat blows firstCheck usage
Vercel1M edge reqs, 1M invocations, 360 GB-hr memory, 100 GB transfer, 100 deploys/day; non-commercial onlydeploy cap in agent loops; memory GB-hrs — bills through I/O waits, so LLM streaming/polling burns it with zero CPUvercel usage
Cloudflareper-DAY quotas, hard-fail: 100k Worker reqs + 10ms CPU; KV 1k writes; D1 100k writes; R2 10 GB storage + $0 egressKV's 1k writes/day — minutes under loaddashboard Billable Usage; wrangler d1 info
Neon100 CU-hr/mo/project, 0.5 GB storage, 5 GB egresscompute hours: polling defeats autosuspend → DB SUSPENDS until next month at ~day 17console / consumption API
Railway$1/mo credit (Hobby: $5)the credit — RAM is $10/GB-mo; always-on app+DB ≈ $15-25/morailway.com/workspace/usage
Fly.ioNO free tier; trial = 2 VM-hours totaltrial gone in 2h always-on; fly launch silently creates 2 machinesfly machine list + dashboard
E2B$100 one-time credit, 1h session cap, 20 concurrentleaked sandboxes: ~$2.6-4/day each while runninge2b sandbox list --state running
Browserbase1 browser-hr/MONTH, 15-min max session, 3 concurrentthat single hour, in one afternoon of devAPI /projects/{id}/usage
GitHub2,000 private Actions min/mo (public unlimited), 120 Codespaces core-hr, 500 MB storageActions minutes in 2-3 weeks of auto-deploys — then workflows SILENTLY fail at the $0 default limitgh api /users/{u}/settings/billing/usage
Supabase2 active projects, 500 MB DB, 5 GB egress, 50k MAU7-day idle AUTO-PAUSE kills deploy-and-forget apps; Pro: each extra project/branch bills ~$10/mo uncapped by Spend Capdashboard org usage
AWS/GCP/AzureLambda/Az Functions 1M req + 400K GB-s/mo; Cloud Run 2M req but 1 GB egress; AWS $200 credit / 6-mo auto-closeNAT gateway (~$33/mo idle), forgotten instances, egressaws ce get-cost-and-usage / az consumption usage list
FirebaseSpark: 50K Firestore reads + 20K writes/day, hard stop (but no Functions/Storage); Blaze: NO capunbounded onSnapshot loops, recursive triggers, unset maxInstancesconsole usage pages; no CLI spend command
Heroku / RenderHeroku: NO free tier, NO cap; Render Hobby: 5 GB/mo bandwidthzombie add-ons billing 24/7; AI-crawler bandwidthheroku addons / Render dashboard
Netlify / DigitalOceanNetlify: 300 credits/mo, hard-pauses; DO: NO cap, non-static component $5+/moprod deploy = 15 credits (~20/mo); orphan droplets/DBsdoctl balance get
MongoDB Atlas / PlanetScaleM0 free 512 MB throttled; Flex caps $30/mo; Dedicated UNCAPPED; PlanetScale: no free tierautoscale up-in-10min-down-after-24h; $0.09/GB egressatlas api invoices
Email (Resend/SendGrid/Mailgun)SendGrid free DEAD (60-day trial); others 100/day hard stopcontact-storage overage recurs monthly; Mailgun cap OFF by defaultprovider dashboards
Auth (Clerk/Auth0/Firebase)50K MRU / 25K MAU (bots COUNT) / 50K MAUphone-OTP is a second, easy-to-miss meter — SMS bills per message on top of MAUprovider dashboards
Maps (Google/Mapbox)per-SKU free calls (10K/5K/1K, NOT pooled); Mapbox 50K loadsunrestricted client keys, map remounts; NO caps on eitherGCP console / Mapbox stats
AI APIs (OpenAI/Anthropic/Replicate)Anthropic: tier ceiling (real cap); OpenAI: alerts only; Replicate: creditenv API key flips agent CLIs to per-token billing without warningclaude /status; console usage pages
GPU (RunPod/Modal/Vast)prepaid balance = de facto cap UNTIL auto-top-up; Modal $30/mo creditidle pods, warm min_containers, stop != destroyrunpodctl get pod / modal app list / vastai show instances
Observability (Datadog/CloudWatch/Sentry)DD/CW: NO caps (alerts only); Sentry 5K errors/motracesSampleRate 1.0, ID metric tags, Never-Expire retentionDD usage monitors / CW billing
Media/vector (Cloudinary/Pinecone/Upstash)25 credits / 2 GB + 1M RU / 500K cmdslegacy Pinecone pods $81+/mo idle; Upstash hard cap exists but OFFdashboards; cld admin usage
Show full SKILL.md (242 more words)Show less

Paid-plan tripwires (fail-open walls)

When the user IS on a paid plan, these are the researched danger spots:

  • Vercel Pro: Spend Management defaults to a $200 notification, not a stop; auto-pause is opt-in and paused projects never auto-unpause. Builds bill by default (Elastic machines) — every agent push costs CPU-minutes.
  • Neon Launch/Scale: pure pay-as-you-go, $0 base. The free-tier suspend becomes a silent bill: one always-on 1 CU compute ≈ $77/mo (Launch); egress past 500 GB bills $0.10/GB; autosuspend CAN be disabled — make sure it isn't, and set Neon's spend limit (alerts at ~80%/100%).
  • Cloudflare Workers Paid: $5/mo floor, then open-ended — KV writes $5/M, Durable Objects duration $12.50/M GB-s, D1 writes $1/M.
  • Supabase Pro: Spend Cap exists but does not cover per-project compute — each extra project/branch bills ~$10/mo regardless.
  • GitHub: raising the $0 Actions spending limit opens unbounded per-minute billing; macOS runners ~10x Linux.
  • E2B Pro: $150/mo raises concurrency/session limits only — sandbox seconds still bill on top.
  • Fly.io: pure PAYG with no billing alerts at all — inventory machines/volumes regularly (fly machine list, fly volumes list).
  • AWS/GCP/Azure: no default caps — set aws budgets / gcloud billing budgets / Azure cost alerts before provisioning anything (and remember they only email, on lagged data).
  • Firebase Blaze: NO hard cap exists; Firestore read loops and recursive triggers bill unbounded — set functions maxInstances before going live.
  • No-cap vendors: Mapbox (and several metered SaaS) offer no spend cap at all — viral traffic means unbounded exposure; prefer cappable or self-hostable alternatives for public pages.

© yuanboP, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 2 other files (references) in skills/frugal of yuanboP/frugal.

  • SKILL.md
  • references/indie-services.md
  • references/providers.md

Open the folder on GitHubat commit b0f07ea

Compare with similar skills

Frugal next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Frugal compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Frugal this skillyuanboP/frugal198—~2.1kAutomated safety check: PassMIT
Cloud Architectdavila7/claude-code-templates33k8 repos~1.9kAutomated safety check: PassMIT
Cloud Cost Optimizationwshobson/agents40k14 repos~1.7kAutomated safety check: PassMIT
Terravision Cloud Diagramspatrickchugh/terravision1.6k—~5.6kAutomated safety check: NotesAGPL-3.0-only
TerrasharkLukasNiessen/terrashark716—~843Automated safety check: PassMIT
Spotinfoalexei-led/spotinfo164—~1.8kAutomated safety check: PassApache-2.0

Similar skills

  • Cloud Architect

    davila7/claude-code-templates

    Expert cloud architect specializing in AWS/Azure/GCP multi-cloud infrastructure design, advanced IaC (Terraform/OpenTofu/CDK), FinOps cost optimization, and modern architectural patterns.

    33k GitHub starsUsed in 8 repos~1.9k tokens
    DevOps & CloudAuto-check passed
  • Cuts cloud spend across AWS, Azure, GCP and OCI with cost tagging, rightsizing, commitment and spot pricing models, and architecture changes.

    40k GitHub starsUsed in 14 repos~1.7k tokens
    DevOps & CloudAuto-check passed
  • Terravision Cloud Diagrams

    patrickchugh/terravision

    Draw cloud architecture diagrams for AWS, Azure or GCP with the official provider icon sets, using TerraVision.

    1.6k GitHub stars~5.6k tokensUpdated 4 days ago
    DevOps & CloudAuto-check: notes
  • Terrashark

    LukasNiessen/terrashark

    Prevent Terraform/OpenTofu hallucinations by diagnosing and fixing failure modes: identity churn, secret exposure, blast-radius mistakes, CI drift, and compliance gate gaps.

    716 GitHub stars~843 tokensUpdated 8 days ago
    DevOps & CloudAuto-check passed
  • Spotinfo

    alexei-led/spotinfo

    Query Spot/preemptible VM prices, savings and interruption risk across AWS, GCP and Azure with the spotinfo CLI.

    164 GitHub stars~1.8k tokensUpdated 5 days ago
    DevOps & CloudAuto-check passed
  • Verify mql provider resource/field changes against real cloud infrastructure.

    412 GitHub stars~3.7k tokensUpdated today
    DevOps & CloudAuto-check passed

Categories

Questions about Frugal

What does Frugal do?

Cloud cost awareness for agents. An agent skill from yuanboP/frugal. Frugal is an agent skill from yuanboP/frugal. Cloud cost awareness for agents.

When should I use Frugal?

Frugal fits situations like: the user asks about cloud costs; why is my bill so high; reducing cloud spend; before deploying/provisioning to any metered service.

How do I install Frugal in Claude Code?

Run `npx skills add yuanboP/frugal --skill frugal -a claude-code`. Or copy the skill folder (skills/frugal in yuanboP/frugal) into .claude/skills/frugal in your project. Claude Code loads it when a task matches its description.

How do I install Frugal in Codex?

Run `npx skills add yuanboP/frugal --skill frugal -a codex`. Or copy the skill folder (skills/frugal in yuanboP/frugal) into .agents/skills/frugal in your project. Codex loads it when a task matches its description.

Can I use Frugal in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add yuanboP/frugal --skill frugal -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/frugal, .gemini/skills/frugal, .github/skills/frugal and .opencode/skills/frugal in your project.

What does Frugal need to run?

Going by SKILL.md and its folder, Frugal needs the command-line tools its instructions call (fly, wrangler, vercel, aws, supabase and gh).

Does Frugal access the network?

SKILL.md contains no URLs. Its commands use gh, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Frugal safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Frugal use?

Frugal is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Frugal use?

About 2.1k tokens (SKILL.md is roughly 8.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 52k tokens, read only when the agent opens those files.

What are the alternatives to Frugal?

Skills that share tags, products or a category with Frugal: Cloud Architect (davila7/claude-code-templates, 33k stars), Cloud Cost Optimization (wshobson/agents, 40k stars), Terravision Cloud Diagrams (patrickchugh/terravision, 1.6k stars) and Terrashark (LukasNiessen/terrashark, 716 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Frugal?

yuanboP (a GitHub user) maintains it in yuanboP/frugal, which has 198 GitHub stars. The repository was last updated on July 19, 2026.

Source: yuanboP/frugal on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.