"I seriously need you to do whatever it is that you do. Like, now."

Integration of a human brain with a computer network in any kind of reasonably safe manner takes time. Specifically, it takes one entire round per rating for software to take hold of a node. So when you first turn your Commlink on and plug your brain into your network, your System and Firewall will be capped at 1 after one round, capped at 2 after 2 rounds, and so on. Thus it takes five entire combat rounds for a character to be able to get full benefit from a Firewall or System rating of 5. When a network crashes, a new network will have to be put up in its place, meaning that there will be a round with no Firewall, a round with a Firewall of 1, a round with a Firewall of 2...

Yugi is in the process of mindcrushing a fool. Unfortunately, it turns out that he has a wicked powerful firewall and a good biofeedback filter in place that makes all of Yugi's cool programs not work. Not to be daunted, Yugi first hits the fool's network with some data attacks, causing it to crash. Then, with the firewall and biofeedback filters gone, Yugi is able to turn his RAS Override taser on him and then begin with blackhammering. I pity the fool.

A character's defense pool in the Matrix can be increased for one initiative pass by a user's Cybercombat skill with a Complex Action. Like full defense in the physical world, a character who is not surprised can abort a future action to go on full defense right now.

The rating of a program limits the number of hits a character can achieve on the test to activate it. A technomancer's CFs are normally of a Rating equal to their Resonance, while a Hacker's programming is of whatever rating they happen to have a copy of. By spending a number of consecutive free actions to augment and optimize the code, the character can make the program operate as if its rating were increased by the number of hits the hacker gets on a Logic + Software test or the number of free actions spent on the attempt whichever is less. A character whose network does not have access to a Program can still attempt to use that Program by enhancing it from negative one. These emulated programs cut an astounding amount of corners and are only good for one use. Grounded as they are in the specific time and place they were formatted in, they will usually not even compile if copied down and run outside their original context. Multiple hacking on the fly attempts are not cumulative, and using a program hacked up from nothing imposes a -2 penalty to the dicepools while using it. A program cannot have its rating increased beyond the network's System rating in this fashion.

Technomancers can do the same thing, though they use their Intuition + Compiling skill rather than their Software skill. Decompiling and Registering forms cannot be threaded or hacked on the fly under any circumstances. Either you know them or you do not.

You are not supposed to be able to turn off biofeedback filtration. But you can. Doing so is unsafe, and illegal in most of the world. Networking security consultants advise you to run additional Biofeedback Filter programming on top of that which is provided by the Firewall of your network. But for those who really don't want the minor delays brought upon by having the system double check to make sure that the signal going to your brain won't fry it – these safety measures can be cut off entirely. With the Biofeedback Filters toggled off, the Firewall provides no defense against B program attacks. On the plus side, the matrix can be pumped into one's brain at BTL levels, increasing the effective Response of the network by 2.

It takes one Complex Action to engage or disengage Biofeedback Filtration, and the Firewall must already be cracked to allow this sort of thing, which requires a Software (4, 1 hour) extended test to get off the ground. One cannot run BTL chips in the manner that they were intended without doing so. If the Firewall has not been previously configured to allow this sort of tomfoolery, the entire Firewall can be disengaged with a number of consecutive Complex Actions equal to the rating of the Firewall. A network which is not running a Firewall at all can gain the same +2 Response from BTL inputs as a network running a Firewall with the Biofeedback Filters disengaged.

A hacker (or technomancer) can switch between AR and VR as a complex action. Note that AR itself is basically impossible to turn off, the best you can hope for is shutting unwanted AR signals out with a firewall or Faraday Cage. While in VR, a character gets 3 IP and the thresholds for piloting remotely are reduced by 1. Also, their Initiative in VR is Reaction + Intuition + Response. Digital speeds are quite fast. However, the character's body is largely inert and physically perceiving anything is made with a -6 penalty. For simplicity, there is no roll involved to engage the sensory and motor cutouts on one's self – with the brain and firewall working together it is always assumed to work.

When two networks/devices/whatever are within Handshake range of one another, a connection can be opened between them with a Simple Action taken by either party, provided that the other party has agreed to allow that sort of thing. Marking or unmarking a source as something which a proposed connection should be allowed with is a Free Action. Terminating a connection while still a Simple Action, is slightly more difficult, as the other party can attempt to fight the attempt. The terminating party makes a Logic + Computer test with a threshold set by the Logic + Hacking test. If distance or background renders the devices outside range, then the connection is terminated no matter how good either user is.

Systems send information back and forth continuously in massive amounts. To tag specific information as worthy of being sent, one merely selects it for that purpose as a Free Action.

Programs have a variable action cost, some of them take Simple or Complex Actions, while others take a long time. When running a program, a character makes a skill check appropriate to the type of program or complex form that it is (for example: an Analysis program requires a Logic + Data Search test while a Decompiling Form requires a Resonance + Decompiling test). Maintenance of a program that can be sustained does not require actions at all, however if a network attempts to sustain more programs than its System rating, all additional programs provide a cumulative -2 penalty on all Matrix related tests. Remember that all Programs use Logic + Skill, while all Complex Forms use Resonance + Skill.

Characters are generally given an opportunity to spot things in the Matrix reactively without even spending actions. However, if a user is sufficiently sure that something is there, whether from simple paranoia or some other piece of information, then the user can spend a Simple Action to make an additional Matrix Perception Test (Intuition + Data Search). If the user already knows what is being looked for, a +2 dicepool modifier is gained.

Networks can be restored to a previous state in which they weren't damaged. To determine how long this takes make a Response + System (10, 1 turn) extended test. This test uses the total System rating rather than the current effective System Rating (which of course, is zero). Note that cascading viruses are the rule in 2071, so larger networks are not any harder to bring down than smaller networks.

The matrix has certain structural weaknesses, which can be exploited by anyone with a basic knowledge of the underlying infrastructure, whether they have a special program or not.

A network can generate the ID of any user whose Public and Private keys are known. Thus, once the asymmetric encryption of an ID has been cracked by a hacker, that hacker can assume that ID by re-encrypting their signifiers with that encryption scheme. A hacker can assume any identity whose codes have been broken with a Complex action. Actually passing one's self off as the impersonated system against actively observing users requires a Matrix Stealth Test.

Mocha Latte has successfully decrypted the Access ID of one of the maintenance workers of a facility and wants to impersonate him. The first thing she does is re-encrypt her Access ID with a single Complex Action. At this point, she can send any e-mails she wants and have them clear as being "from" the maintenance man. This means that any simple authorization system will open doors for her automatically. But she also wants to pass matrix inspection from other occupied networks. Since she is attempting to be unnoticed (in this case by disguising herself), she makes a Matrix Stealth Check (Intuition + Hacking) and gets 4 hits. At this point anyone who observes her in the matrix will need to make 4 hits on their Matrix Perception Check (Intuition + Data Search) in order to spot the discrepancies in internal data transmissions in Mocha Latte's network. And of course, anyone observing her in the real world will find it blitheringly obvious that she's a dwarf rather than a male orkish maintenance worker, but that's an entirely different problem.

Ideally, a network is supposed to broadcast its position constantly in order to make creating connections and transferring data easy and reliable. You can just... not do that. Switching your network into or out of hidden mode requires a complex action. Once hidden, a network cannot be seen by any actor in the Matrix who fails to get a hit on a Matrix Perception check. A hiding user can make a Matrix Stealth check, replacing the threshold to spot them with their hits on that check. A technomancer is always in hidden mode unless they add additional devices to their network and set them to be publicly visible. An orphan brain cannot put itself into hidden mode, naked metahuman brainwaves are simply obvious to even casual users of the matrix.

Ghost is sneaking into an installation. Covered as she is with devices and having a human brain as she does, she would just assume not have her network appear on security scans. While she is making her Infiltration test (Agility + Infiltration), she is also making a Matrix Stealth test (Intuition + Hacking) to essentially infiltrate her network. If her network is spotted but her realspace self is not, then installation defenders may be sent to go find the offending network. If her realspace body is spotted but her network is not, then they may respond to the perceived intrusion as befits their protocols.

Since normally humans and networks go hand in hand, finding just one or the other means "ninja".

An occupied network needs a constant interface between the brainwaves that are operating it and the devices which comprise the peripherals. If a hacker is within handshake range of the device which provides that interface (such as a trode net), an attempt to take the entire network and add it to the hacker's can be made. This requires that the Hacker have Line of Sight, and have a receiver with sufficient power to pick up the signals from the trode net, and a signal rating of sufficient strength to send signal to the trode net (in short: handshake range must be made with literally the interface device rather than the target network as a whole). The Hacker spends a Complex Action and makes a Logic + Electronic Warfare test opposed by the victim making their choice of a Logic + Cybercombat or Logic + Computer test (signal defense applies). If the Hacker gets more hits, the victim is straight locked out of their network and no longer has any matrix protections whatsoever. If the proposed target has a direct wired connection to their network, such as with a datajack or internal Commlink, this will generally not be possible. Note that a technomancer can be within handshake range of wired connections within their biological signal range, meaning that they can attempt to seize a network operated with a datajack, although they are not allowed to use Resonance instead of Logic for this test.

Once two icons share space on the same server, it is child's play to open a connection between them. Since both icons are sending data packets to and from the same data stores, there is essentially no way to prevent another Icon from indirectly sending data packets to the other. With a Simple Action, the Handshake range between two icons connected to the same Server can be reduced to Connection Range. This action itself has a range of Handshake and there is no test involved if one has any Computer or Hacking skill at all.

A VR feed can be a direct SimSense feed from something in the real world. With the same Complex Action that one can replace one's sensory feed with that of a fully immersive matrix representation, one can replace one's sensory feed with that of a drone, a camera, or another device. While jumped in, a character has 3 IP like they were in VR, but they make Physical Perception tests using the sensors of the device they are jumped into. A device which is sending SimSense of a sense that the rigger does not have provides a -2 dicepool penalty on Perception tests. If a character is already in VR, jumping into a drone is a Simple Action.

While jumped into a drone, a rigger can act normally using the drone's physical capabilities instead of their own. So firing a burst from a drone's weapon system would be a Simple Action, and moving the drone would be a free action. The rigger is still in VR, and uses matrix attributes rather than physical ones, and uses Pilot and Gunnery skills rather than Athletics or specific weapon skills.

Sense feeds can be overlain one on top of another, and cycled through at very high speeds. A character can monitor a number of sensory feeds as Arrows, requiring the expenditure of a Matrix Free Action for each sensory feed being used. When cycling through multiple sensory feeds, a character's Perception tests are penalized by -2 dice.

A rigger can give orders to a drone as the same Free Action that one uses to send any other information. If a Drone gets orders it can follow them without having to dither. If a drone does not have relevant orders it has to take actions making Pilot checks.

Merci is watching the video feeds from a couple of drones, which she has named François and Luna. Just being apprised of the visual information from both drones uses up two of her Free Actions each IP. When she notices an enemy with her Intuition + Perception test (at a -2 die penalty for viewing things in windowed mode), she can spend a Free Action to give orders to François and another Free Action to give orders to Luna. Both drones will act normally on their initiative and attack the enemy she chose, but she has spent 4 Free Actions this IP to make sure that happens. Fortunately for her, she has a Response of more than 4 so she still has Free Actions left over.

If drones are left to their own devices to determine what is and is not a valid target, they have to spend a Complex Action figuring that out. It makes a Pilot + Clearsight test to make that happen.