Smartcooky beat me to it with the facts. NTRS is still massively broken after the decommissioning of CASI and the subsequent breakage of many links to technical reports.
*cracks knuckles*
This is my special area of (somewhat dated) expertise. My work on the Delta family of launch vehicles (chiefly the Delta III but also backward toward the Delta II and forward toward the Delta IV) was the payload-bearing structures and their relevant coupled-load analysis and separation sequence analysis. The half-dozen or so payload-attach fittings (PAF) for those vehicles make special use of pyrotechnical and ingenious mechanical solutions.
Was the staging anomaly on Skylab 1 the failure of a single pyrotechnic device? Yes. Normally we design these assemblies with redundant pyros so that the failure of any one device doesn't fail the system. Here they didn't. A single device was meant to sever all the straps holding the structures together. Any mechanical or thermal damage to that single device meant that the detonation front would be interrupted at the point of damage, which is what occurred on that flight. Systemic redundancy was limited to multiple detonation initiations of the single device, implemented electronically but not mechanically. Well, sort of. The electrical connectors that pass the "fire" command (both primary and redundant) to the device separate after less than a centimeter of separation has occurred between the interstage and the second stage, and only at the point where the connectors are mounted on the skirt. This means the interstage can partially separate while defeating the electrical redundancy. This is considered a component-level design factor, not a system-level design factor.
In the sense that this was dependent on electrical properties of the stage design and staging sequence design, this was a system-level design failure. The designers should have foreseen that mechanical damage to the single pyro -- which had a massive mechanical extent in the overall vehicle design -- could result in partial separation of the interstage, which could defeat the electrical-only redundancy program. A better design using the existing hardware would have initiated the circumferential pyro detonation simultaneously at both ends. This would have meant that a single discontinuity would have been perfectly sustainable; the interstage joint would have to be hit two or more times to result in partial pyro detonation.
The system failure argument incorporating the micrometeoroid shield is valid but weak. Clearly there was a failure to communicate design constraints between design teams, and that's what I did for the Delta LV team back in the day. It's weak because specific causes of debris damage aren't important to LV stage designers. You design launch vehicles (especially those with cryogenic propellants) to tolerate impact damage from any source, including from unknown sources. Literally anything forward of the point in question can break off and slam into the launch vehicle with supersonic force.
And yes, Challenger. The STS design was fully head-in-sand about this. By the early 1970s it was already known that launch vehicles were likely to suffer minor impact damage during boost. Modern launch vehicles had already been adapted to this fact. The Saturn V, by the early 1970s, was no longer a modern launch vehicle in this respect. It could be made resilient in the face of this particular failure by the modification I outlined above. But as this was the last flight of the Saturn V, that wasn't really a consideration. STS vacillated about this, with the knowledge that the orbiter TPS was especially vulnerable to impact damage and the naive expectation that debris shedding could be controlled with additional engineering.
Nope.
Payload attachment strategies today typically use clampband designs. The payload mates to the circular adapter structure in such a way that the shared circular "lip" is a V-shaped arrangement that accepts the V-shaped concave channel of a strong band going around the circumference. At 0-degree and 180-degree radials, the band is held in place and tensioned by studs that are severed by pyro cutters. Only one of the two cutters has to succeed in order for the clampband to release, and since they are at diametrically opposed points on the vehicle, the chances of them both being damaged by impacts is minimized.
Full-race linear shaped charges these days are permitted only in a few places, such as payload fairing separation. Since the payload fairing is the most forward structure, it isn't likely to be damaged mechanically. Gemini's "angry alligator" showed the problems with clampband methods for separating payload fairings, but a few designs still use(d) it (e.g., SpaceX Falcon 1).
But back to stage joinery. Rocket stage joints have to accept bending-moment loads on the order of 80,000 lbf-in. This requires them to be especially robust in tension around the perimeter of stage. This further requires separation devices also to be located near the vehicle skin and be susceptible to debris impact damage. Newer interstage joint designs make more extensive use of fuze pins and other simply-frangible structures to ensure both sufficient strength to withstand flight loads and reliability at separation.
