You have a code change. But what proves it works—and what should the next session do? Opus 5.5 Coding Workflow & Cost Kit gives you a small, practical system for answering both questions. Define the task, review a plan, run meaningful checks, inspect the diff and leave a useful handoff. When you use the API, estimate token charges with visible formulas and editable rates. Built by JRDEVHUB for developers who already read code, run tests and work with AI coding tools. It is a focused companion to The Coding Agent Lab and useful on its own. What you can do with it Turn a vague request into observable acceptance criteria and a bounded implementation plan. Keep project context and session handoffs concise enough to reuse. Investigate bugs with reproducing inputs and review AI-generated code against evidence. Estimate standard Claude API token charges without confusing them with Pro or Max limits. Run a complete Swift cooldown exercise, compare the flawed starter with the final code, and inspect the same regression tests before and after. What is included A 13-page designed PDF guide covering setup, planning, one-session execution, context, handoffs, testing, code review and cost accounting. Five editable Markdown workflow templates, plus a CLAUDE.md starter. An editable .xlsx calculator with Calculator, Worked Example and Instructions sheets. Five token categories, manual rate inputs, formulas and invalid-input warnings. The example data is fictional. Two standalone Swift packages: an intentionally flawed starter and its corrected version, each with the same 12-test suite. Includes replayable instructions, a filled handoff and actual local test logs. START_HERE, dated sources, license and a complete file manifest. Exact files in the ZIP START_HERE.md 01_WORKFLOW_GUIDE.pdf 02_API_COST_CALCULATOR.xlsx SOURCES_AND_LIMITS.md LICENSE.txt FILE_MANIFEST.txt templates/01_PROJECT_BRIEF.md templates/02_IMPLEMENTATION_PLAN.md templates/03_SESSION_HANDOFF.md templates/04_BUG_INVESTIGATION.md templates/05_FINAL_CODE_REVIEW.md templates/CLAUDE.md example/WALKTHROUGH.md example/VERIFICATION.md example/Starter/Package.swift example/Starter/Sources/DashCooldown/DashCooldown.swift example/Starter/Tests/DashCooldownTests/DashCooldownTests.swift example/Final/Package.swift example/Final/Sources/DashCooldown/DashCooldown.swift example/Final/Tests/DashCooldownTests/DashCooldownTests.swift example/evidence/starter-tests.txt example/evidence/final-tests.txt example/evidence/boundary-check.txt example/evidence/rejected-attempt-check.txt Requirements A PDF reader, text editor, and Excel or LibreOffice for the workbook. Swift 5.9+ to run the example; verified locally on Apple Swift 6.2.3 and macOS 26.0.1. On macOS, use a selected Xcode command-line toolchain. The packages have no external dependencies. To follow the Claude Code workflow, bring your own installation and account with access to the selected model. Claude subscriptions, API credits and model access are not included. Templates can be adapted to other languages; the runnable example is Swift domain logic, not a complete game or UI app. FAQ Do I need The Coding Agent Lab? No. This kit stands alone. If you own the Lab, use its specialist briefs within this kit's planning or review steps. The Coding Agent Lab is a separate product. Does the calculator predict my Pro or Max limit? No. It estimates first-party Claude API token charges from manual inputs. It does not connect to your account, predict subscription capacity or enforce a budget. Which prices does it use? Standard global synchronous Opus 5.5 API rates checked on 28 September 2026. Rates are editable. Fast/Batch modifiers, regional premiums, partner rates, tool charges, taxes and subscription fees are outside the displayed total. Was the example an Opus benchmark? No. It is newly authored teaching code made with AI assistance. Local Swift tests were executed, but no Opus cost or performance result was reproduced. The final package passed 12 tests; game UI integration was not tested. Can I use the files commercially? Yes, in your own projects under the included license. The Swift example is MIT licensed. Do not redistribute or resell the reusable kit itself. This purchase does not include automatic updates or consulting. Will it guarantee lower bills or correct AI code? No. It provides working documents, arithmetic and review checkpoints. Your usage, code and verification determine the result. Independent product by JRDEVHUB. Not affiliated with or endorsed by Anthropic. Download the kit, open START_HERE.md and choose one task to work through.