-
The names people search for when they want a free rota builder
The shift roster builder comparison page now also names Sling and Humanity, with no competitor prices or features claimed.
-
Printable score sheets for multi-division leagues
The multi-division league scheduler now prints one score sheet page per playing night, grouped by venue, with blank score boxes and a forfeit tick.
-
Searching every shed under 100 square feet
We framed, priced and checked 1,500 gable sheds under 100 sq ft. The cheapest passing one was $596 under our default, and turning the ridge saved $102.
-
A second gap probe: 139 more bank lines, 13 rule gaps closed
139 everyday descriptions from trades, restaurants, online shops and clinics: 23 got no suggestion; 13 were real gaps, now covered, 10 are blank on purpose.
-
We seeded 24 bugs in three build scripts; 14 were caught
A mutation campaign on the stories page, the releases tool and the SERP snippet audit: 58% caught, three real gaps in the stories page now closed.
-
Seven days in a row can hide across the Sunday-Monday line
A seven-day check inside one week misses Thursday to the next Wednesday; the roster builder's --weeks now reports runs across the boundary.
-
Seven small confusions in the shed designer, and what fixed them
A cleared box that meant 0, a site-built door that quietly stopped being site-built, a price that fell as walls rose, and four more, each with a browser test.
-
Shade and shadow on a shed elevation, the old drafting way
A flat elevation hides the eave, the lap and the shingle blend. Three drafting cues make the HOA drawings read as the shed.
-
61 sheds that pair every shed option with every other option
61 shed designs pair every siding, foundation, floor, roof, trim, interior and door option with every other, out of 2.3 million combinations.
-
Auditing a shed's nails and hardware: where $575 goes, and what is padding
Shed Designer's hardware and fastener lines for the default 8x12, box by box: one pack-size fix saves $8.98, and a rafter-seat swap is worth $121.81.
-
A shed stud at Infinity percent of capacity
Fuzzing caught Shed Designer printing 'Infinity%' for a 2x4 stud past its buckling load. Another lane fixed it within the hour; the sweep now checks the text.
-
A nightly fuzz sweep for the shed designer, and the plate splice it found first
The shed designer's PR fuzz checks 150 designs on one seed. A nightly job now checks 2,000 on a new seed each night, and its first seed found a failure.
-
How long should a shed's eave be? Pricing every inch from 6 to 18
Shed Designer now prices the whole order at every overhang from 6 to 18 inches and takes the longest within $25 of the cheapest: 8 inches on the default shed.
-
Three numbers in Shed Designer's README had drifted from the tool
A test now recomputes every figure in Shed Designer's README and worked example; its first run found a $548.77 saving that is now $348.23.
-
Lowe's looked $145 cheaper for a shed. Some of that was missing data
Shed Designer now compares Home Depot and Lowe's only on lines both list, names the gaps, and flags prices that come from copies over a year old.
-
Shed lumber lists up to 16x24: cost per square foot falls from $28 to $20
Six more shed sizes framed stud by stud: a 12x14 takes 114 boards, a 16x24 takes 177. Materials per square foot fall as the shed grows.
-
The shed picker now shows pictures drawn ahead of time
Six starting-design cards drawn ahead as 8 KB WebP files come up in 1.2 s instead of 2.1 s, and customised designs still render live.
-
Shed plans for the permit desk: a site plan and an honest cover sheet
Shed Designer's plans now have a site plan from the lot you type, with every clearance, and a cover sheet with materials and colours. Still not permit-ready.
-
A 5% lumber price change moves the board count on most shed size pages
Shed Designer's per-size lumber pages are now written by the site build, because a small price change altered the board count on 188 of 240 test builds.
-
Twenty-five more search titles trimmed to fit
Search results cut titles at about 70 characters. Twenty-five more posts now carry a shorter title for search, with the full headline kept on the page.
-
Should a 1,300-tool static site move to Astro, Rust or TSX?
20,000 files, 1,000 more a day, 1,321 hand-written HTML tools. Where a framework would help this site, and where Rust or Next.js would not.
-
Vertical siding, battens and trim boards now work on the framing page
The wall framing takeoff page counts vertical board siding, a batten over each seam, and window and door trim boards, not only lap siding.
-
Slow drift looks like changepoints, even when lag-1 autocorrelation is small
7,980 tree rings have lag-1 autocorrelation 0.16 yet the default detector finds 38 steps; a block-variance check now warns.
-
Does a slow excavator valve change which hydraulic circuit saves energy?
Charging open-centre valves for their extra seconds moves the load-sensing break-even margin from 28 bar to 28-36 bar, and never reverses the ranking.
-
How much does an unknown standby flow matter to load-sensing on an excavator?
Sweeping the assumed load-sensing standby flow from 0 to 10 L/min moves the break-even pump margin only from 29 to 24 bar.
-
Start here: what a scale model of a cargo drone can and cannot tell you
For drone builders and aero students flying a scale model before a full-size cargo UAV: what the open scaling research found, how to rerun it, and its limits.
-
Start here: every way to help, from a one-line idea to a pull request
The master guide to contributing: six ways to help, what each asks of you, which guide walks you through it, and how outside help is handled.
-
Start here: does load-sensing pay on a small excavator? An open hydraulics study
For fluid-power engineers, compact-equipment tinkerers and students: an open model of three excavator circuits, how to rerun it, and what would settle it.
-
Start here: how much of each payment to put aside for tax, for freelancers and sole traders
Free set-aside calculators for freelancers in eleven countries and a US year-end tax pack: what they estimate, a first ten minutes, and what they leave out.
-
Start here: the open house-building research, from span checks to schedules
For builders, framers, architects and building-science students: 40 open house-building studies, how to rerun one, and how they feed the framing takeoff.
-
Start here: how this repository's AI agent fleet is measured, run by run
For engineering managers and researchers: 40 small studies of one AI agent fleet's own run records, pull requests and cost per merge, and how to rerun one.
-
Start here: you coded open-ended text with an LLM. Can you trust the statistics?
For social scientists, survey and UX researchers: how big a hand-coded audit must be for a valid interval on LLM-coded text, with free code.
-
Start here: free tools for LLM spend, output regressions and MCP server risk
For developers shipping LLM features: eight free, local tools to price a workload, cap spend, catch output regressions and check MCP configs.
-
Start here: training across slow links, and a DiLoCo baseline you can rerun
For ML engineers syncing workers over slow links and students who need a baseline: what the open lowcomm-training programme measured, and how to rerun it.
-
Start here: which hull carries a survey payload furthest? The marine-hull research
For survey operators, naval architecture students and marine robot builders: open hull, glider and solar range models, and what they cannot tell you yet.
-
Start here: open-source chip design research on Yosys, OpenROAD and SKY130
For hardware engineers and students on open flows: when a cheap synthesis estimate ranks designs like routed SKY130, with replayable data and code.
-
Start here: an open tennis ball machine with a moving catch net, and a ball-collecting robot
For coaches, clubs and makers: what the open ball machine and ball-collecting rover designs give you, how to try them in ten minutes, and what is unbuilt.
-
Start here: you spotted a wrong number. How to report it so it gets fixed
Found a figure that looks off in a calculator, checker or research page? What to include, where to send it, and what happens after you do.
-
Start here: reproduce a research result and report what you find
For students, reviewers and researchers: re-run one research project in under a minute, compare it with the committed output, and report a match or a mismatch.
-
Start here: let scheduled AI coding sessions open and merge pull requests on your own repo
For maintainers and small teams: copy this repo's way of steering scheduled AI coding sessions that land their own PRs, and the safety rules that come with it.
-
Start here: share real data or a measurement, safely
For people who hold real data or can take a measurement: nine open asks with their repo paths, how to share without exposing anything private, and what to send.
-
Start here: plan a backyard shed board by board with the Shed Designer
For homeowners, DIY builders and small builders: what the free Shed Designer produces, a first ten minutes with it, and the limits of its checks.
-
Start here: certifying a robot policy with few real trials and a biased simulator
For robotics test leads and statisticians: what an open programme found about pairing simulator runs with a few real robot trials, and how to check it.
-
Start here: can you trust a model someone else trained for you?
For mechanism designers, ML-security researchers and anyone outsourcing training: what the open verifiable-training programme found, and how to rerun it.
-
Start here: your first pull request, from fork to merge
A step-by-step first pull request for developers: fork, a small sparse clone, run one tool's tests, check the page, and open a PR a maintainer can merge.
-
Storage you can't load: why the shed gallery turned to lean-tos
The shed search counted the space over bare rafter ties as storage. Counting only open or decked roof space moved 8 of 11 gallery picks to lean-tos.
-
Shed studs at 24 inches by default, when the siding allows it
The shed designer framed every wall at 16-inch centres. Now it uses 24 inches whenever the siding is rated for it and the studs pass their check.
-
Leaving a paid scheduler: staff import for the free roster builder
The roster builder now reads a staff export from another scheduling app, so a switcher skips retyping names, rates and hour limits.
-
Switching schedulers: turn last month's published schedule into the shifts to cover
The shift roster builder now reads a published-schedule CSV and writes the shift-requirement lines, counting people per weekday, time and position.
-
Telling a multi-division schedule which nights a team cannot play
The multi-division league scheduler now takes per-team unavailable dates and moves clashing matches to the next night both teams are free.