Simple mitre box to nicely cut OpenBeam (or, possibly, other extrusions).
The idea for the screws doubling as guide pegs comes from Terence Tam's mitre box (not on Thingiverse), however that was too elaborate (esp. the need for a hold-down clamp). I found a simpler one here (see sources), but wanted to clamp it down. Hence this, which uses regular M3 screws as guide pegs, and < $5 ratcheting clamps. I was able to get almost perfectly repeatable cuts; YMMV though.
Print box and base pieces and glue them together. I used ABS at .3mm with 60% infill (probably overkill?) and acetone-based glue.
For guide pegs I used 45mm long partially-threaded M3 screws; the smooth part is just long enough! Best to clean holes with a drill bit (by hand, of course) before inserting screws (holes are intentionally tight).
You may want to add some tape (electrical, duct, whatever -- as long as it provides some friction) to the bottom (surfaces that make contact with beam and with work surface).
I used a standard .025" thick, 24TPI bi-metal blade. Terence Tam suggests 32TPI, but I didn't have one around. I found the results quite acceptable, esp. if you go nice and slow.
The included STLs are for OpenBeam and 2" ratchet clamps (such as these, for example). If you make a box to cut at an angle, you may want to use T-clamps, or clamp the beam using it's slots (or increase the mitre box length, if your printer can handle it), otherwise blade may hit clamps. Haven't tried that option.
Also, if you plan to adjust the blade guide gap: if you go narrower, you'll probably need to grind off a little bit of the screw heads so they clear each other. The screw diameter is also adjustable, but you'll have to download the SCAD file.
You may be able to get away without the M3 screw pegs (and the base) if you have a super-steady hand. My original design did not include these, and I got up to ~0.8mm off-perpendicular on some cuts (also slightly damaging the box). Multiply this by nine beams (Kossel horizontals), and I got an overall length variation in the range of 1.1-1.2mm. This took two hours of slow filing to rectify (which I could have done without! :) ), and the worst cut was still ~0.3mm off.
The sample cuts in the pic are the verticals, for which I used the current version. The results are pretty good, very repeatable (using the first piece to measure the others) with almost negligible variation (esp. for a hand cut!), and with a nice finish after just a couple of minutes of very light filing with a flat bastard file.