Loading page
About this tool
Measure cyclomatic and cognitive complexity, Halstead volume and the maintainability index of your code, and get specific refactoring suggestions.
Code Complexity Analyzer measures a pasted function or file on four established metrics — McCabe cyclomatic complexity, nesting-weighted cognitive complexity, Halstead volume and the maintainability index — then names the specific functions that are too long, too nested or take too many parameters. It is for developers deciding what to refactor first and for reviewers who want a number behind "this is hard to follow". Everything runs in your browser; no code is uploaded.
Open Code Complexity Analyzer on AltFTool — it loads instantly in your browser.
The Studio Editor opens with the Complex Processor sample already loaded — type or paste your own code over it, or pick Simple Calculator or Complex Processor from the "Select Sample..." dropdown in the editor bar. Tab inserts two spaces, and the status bar tracks Line N, Column 1.
There is no analyse button: the synchronous scan reruns on its own about 350 ms after you stop typing, then refreshes the result panels when that pass finishes. The default thresholds flag functions longer than 40 lines, nesting deeper than 3, more than 4 parameters, and cyclomatic complexity above 10.
Read the results across the Overview, Functions and Issues tabs — Overview gives the Overall Quality Score ring with Complexity, Readability and Stability alongside Total Lines, Functions, Maintainability and Issues cards, and the Structure panel scrolls the editor to a line when you click a block. Copy puts the editor contents on the clipboard and briefly reads Copied; Reset restores the default sample after a confirmation prompt.
Cyclomatic, cognitive, Halstead volume and maintainability index each catch a different problem, and all four are shown per function.
Smells are reported against the function that has them — long body, deep nesting, too many parameters — with a suggested fix, not a file-level warning.
Parsing and scoring happen in the page, so proprietary source never leaves the machine.
10 or below per function. That is the ceiling Thomas McCabe proposed in his 1976 paper introducing the metric, and it is the default threshold here. The number is one plus one for every decision point — if, else-if, for, while, case, catch, &&, || and each ternary — so a function with no branches scores exactly 1, and each branch you add costs one point.
Cyclomatic counts independent paths and tells you how many test cases you need; every branch costs the same 1 point wherever it sits. Cognitive complexity weights each branch by nesting depth, so an `if` inside two loops costs more than an `if` at the top of the function. A function can be cyclomatically simple and still cognitively awful — that is exactly the case this pairing catches.
With the standard three-term formula MI = 171 − 5.2 × ln(Halstead volume) − 0.23 × total cyclomatic complexity − 16.2 × ln(lines of code), rescaled to 0-100. Higher is better. The three terms mean the index falls as a file grows, as its vocabulary of distinct operators and operands grows, and as its branching grows.
No. Tokenising, parsing, scoring and the suggestions all run as JavaScript in your own tab, and nothing is written to a network request. You can confirm it by opening the network panel and analysing a file — there are no outbound calls.