Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
DeepSeek
Muse+22 moreClaude CodeCursorCodex CLIDeepSeekMuse
Dots
Manus
Grok
Grok Bot+18 agentsAI-Era Software Estimator: Hours You Can Defend
Newer ones often take "minus 40% for AI" off the total.
$9.99
AI-Era Software Estimator: Hours You Can Defend
Example session with this skill installed
Estimate this for my client: add Stripe subscriptions (monthly and yearly), a billing page and invoice emails to our Next.js + Postgres SaaS. Two developers, both mid-level, using Cursor. Give me AI-first hours and a quick range.
- Read your context and instructions
- Compiled the ai-era software estimator
Estimate: Stripe subscriptions for the SaaS app
Mode: AI-first, mid-level team, Cursor for code generation. Overheads shown as workbook factors (QA, bugs, PM) plus explicit deployment and risk rows.
Read: package.json, prisma/schema.prisma, app/(dashboard)/settings/*, existing lib/email.ts. No Stripe code exists yet.
Assumptions
- Stripe Checkout and Customer Portal, not a custom card form
- Two prices (monthly, yearly), one product tier
- Invoice emails reuse the existing email helper
| # | Task | Human hrs (AI-first) | Manual hrs (reference) | Notes |
|---|---|---|---|---|
| 1 | Stripe products, prices, webhook endpoint setup | 2 | 3 | Dashboard config plus env wiring |
| 2 | Checkout session + success/cancel flow | 4 | 8 | Agent scaffolds route and types |
| 3 | Webhooks: subscription created/updated/deleted, invoice paid/failed | 6 | 10 | Idempotency and signature checks reviewed by hand |
| 4 | DB: subscription table + migration | 1.5 | 3 | |
| 5 | Billing page (plan, status, portal link) | 4 | 7 | |
| 6 | Invoice and failed-payment emails | 2 | 4 | |
| 7 | Tests: webhook handlers + checkout happy path | 4 | 8 | |
| 8 | Deployment: env vars, webhook in production, smoke test | 1.5 | 2 | |
| 9 | Risk: webhook ordering and retries | 2 | 2 | Shown as its own line |
| Development subtotal | 27 | 47 | ||
| QA (25% of tasks 1-7) | 5.9 | 10.8 | ||
| Bug fixing (15% of tasks 1-7) | 3.5 | 6.5 | ||
| PM (17.5% of tasks 1-7) | 4.1 | 7.5 | ||
| Total | 40.5 h | 71.8 h |
Range: 35 to 48 active human hours. Agent runtime is not billed and is not included.
Biggest risk: webhook ordering. Stripe can deliver invoice.paid before customer.subscription.created; task 9 covers it explicitly.
To firm up the number: do you need proration when users switch between monthly and yearly? It adds about 4 hours.
Connects securely to your tools. The creator never sees your data.
About this skill
Old estimates assume a person types every line. Newer ones often take "minus 40% for AI" off the total. Neither is easy to defend in front of a client.
This skill makes your agent estimate the way a careful technical lead would: look at what actually has to change in the code, decide per task what an AI agent will do and what a human still has to do, and show the working.
What your agent does
- Reads the real sources. Requirements docs, tickets, screenshots, spreadsheets, an existing estimate or the repository itself. It says exactly what it read and what it couldn't open.
- Traces the implementation path before it groups tasks, so a one-line business title doesn't become a guess.
- Asks only what moves the number. At most five questions, and only when an answer changes scope, architecture or hours. Otherwise it states the assumption and carries on.
- Estimates AI-first. Active human hours are estimated directly per task. Unattended agent runtime and elapsed time are kept separate, so you never sell machine time as labour.
- Keeps the overheads honest. QA, bug fixing, PM, testing, deployment and risk each appear once, as their own visible lines.
- Writes the spreadsheet. With a Google Sheets connector it produces a clean estimate tab with formulas, factors and a single summary, and re-reads it to check the totals.
Modes you can pick
delivery_mode: AI-first, blended or manualdevelopment_mode: AI-assisted coding, agentic SDLC, agentic product, or hybrid- Team level and estimator role, kept separate on purpose
- Depth: a quick range or a full auditable workbook
What's inside
SKILL.mdwith the input contract, operating rules and output format- 9 references: calculation model, AI-first effort calibration, agentic AI estimation, technology routing (.NET, SCA/NetSuite, web, mobile), workbook audit and template, output contract
- A spreadsheet column template (CSV seed) for building the workbook
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Business & Operations.
Hours are estimates, not quotes: they depend on the requirements and access you give it. Writing straight into Google Sheets needs a Google Workspace / Google Cloud MCP connector; without one you get a markdown table or a local .xlsx. It asks before broad repository scans or web research.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
- Software
- Requirements as text, files, ticketsora repository
- Optional
- Google Cloud