It's a few years from now, and someone needs killing at a remote secure facility, where a dangerous AI is being developed by a terror organization hiding behind a respectable corporate front, which prevents my organization from launching an outright missile strike against the facility. After much effort, I now have an agent inside the facility.

However, the villains are genre-savvy, so:

  • Everyone at the facility has always-on implanted devices that constantly transmit video (from a diamond shaped sensor on one's forehead), a wide array of bio-signals, including heart-rate and other standard metrics, as well as real-time streams on the content of the gut and lungs, GPS with centimeter accuracy.
  • There are 3D scanning sensors in every room and hallway that store volumetric voxel data at cubic micron resolutions for months.
  • The food comes in the form of tamper-proof containerized MREs. The water/soda/juice comes from a secure, modular supply that is refreshed under security oversight. There is no alcohol on base, except in the medical facility, which my agent, as non-medical personnel, has no direct access to.
  • There are armed guards sweeping the facility at a randomized intervals, with a maximum 5 minute interval between sweeps, with canines trained to detect explosives and common poisons. All labs and personnel residences have secure carded and biometrically protected doors, and are designed to be fully visible from the hallways across glass walls.
  • A full weapon check of belongings and an intrusive body CAT-scan is done for all arrivals, as well as any incoming shipments. This includes a radiation scan.
  • Electricity is limited to 5-Volt USB-C ports, except in the secured server rooms. There is no access to gas-stoves. No animals besides the K9 units are allowed on base.

There is a common dining hall, and all staff have individual sleeping quarters (all subject to monitoring, of course). Labs house 5-20 people, and the full staff is about 250. The agent has access to most labs, including the one of their target, but no direct access to sleeping quarters, security center, facility maintenance areas, server rooms or medical area.

Yet, for the safety of the nation, one of the researchers must perish. I have one agent inside the facility. Yet it is imperative that the agent not be captured, lest brain-damaging truth serums used on them would extract a confession and the scandal could bring my organization down. How can he/she do it and avoid detection for at least 5 days, which is half the average interval between (the randomly timed) arrivals of transport shuttles (access is by military helicopter or boat) at the facility?

More specifically, I'm still thinking poison. Is it possible to poison someone, and have it be slow-acting enough that my agent can get away? (I am however open to other avenues if poison does not work).

  • 8
    $\begingroup$ I don't understand why the agent needs to avoid capture. If you eliminate that point then the agent should be able to strike up a romance with the researcher, then simply use their hands to strangle them (or use whatever large heavy object happens to be close by) claiming some insane slight. It doesn't have to lead back to your organization and crimes of passion are well known.... Think Crazy Boy/Girlfriend. If they are truly committed to your goals, then martyring themselves would be acceptable. $\endgroup$ – NotMe Apr 18 '16 at 14:33
  • 4
    $\begingroup$ Murder/suicide? $\endgroup$ – NotMe Apr 18 '16 at 14:35
  • 17
    $\begingroup$ FYI "store volumetric voxel data at cubic micron resolutions for months" is absolutely ridiculous. For a building with 10,000 cubic meters of volume (not a super large building - 4000m2 floor space), that's 10^23 cubic microns. You'd have to reduce that by a factor of 10^11 in order to limit storage to a mere terabyte per scan. $\endgroup$ – Rob Watts Apr 18 '16 at 17:41
  • 5
    $\begingroup$ "Someone" needs some killing in the "next few years". Nice try, Serban. I already called the FBI on you. $\endgroup$ – corsiKa Apr 18 '16 at 18:05
  • 7
    $\begingroup$ I knew there was something sinister about the Svalbard Global Seed Vault... $\endgroup$ – Kys Apr 18 '16 at 18:19

26 Answers 26


While a suicide attack is perhaps the most likely to be effective, I would question the idea there is only one "indispensable man" in the opposing organization. To work on and create a high level project is going to need a lot of resources, and many people who know portions of the project and who could (with suitable management and resources) pick up the project and carry on, perhaps more slowly than before but still capable to carry on.

So your assassin needs to eliminate everyone involved in the project. Luckily for you, the enemy have done most of the work themselves. All the important personnel are sealed inside what is essentially a giant vault with an amazing number of connections and interlocking systems. Your assassin simply needs to introduce malware or physically crash the security system to lock all the doors in the closed position and turn off the ventilation, lighting and outside communications lines. (Actually an SoF team outside can probably handle the destruction of the antenna farm and telecommunications hub leading into the building, but cutting off access from the inside is "cleaner").

So their "in" is to either come to work as an IT tech, or have the ability to memorize and write blocks of code and induce it into the system from wherever else they are working. They can even escape if the malware is triggered by a timestamp or other event inside the facility (timing it for the 6th daily login of the prime target would be good), so the assassin comes in, sets the malware trap, then leaves five days later. On the 6th day, the facility suddenly goes into lockdown (with people locked inside their rooms or workstations) and shuts off all communications, and the people trapped inside will asphyxiate, starve or die of dehydration. Turning off the lights will induce panic in a lot of the staff, and drastically hinder the ability of the people inside to take effective measures to rectify the situation. Since they will have to break down or otherwise cut through armoured doors to move around the facility, they will use up air and water at a furious rate and their time window for doing things will be quite short indeed.

If you can "spoof" routine comms from the facility for 72hr, then you have a very good chance that the majority of the people inside will be quite dead before SPECTRE is aware of the problem.

  • 11
    $\begingroup$ Damned good answer. If the place is highly automated you may also be able to somehow alter the CO2 levels in the air, or otherwise cause a carbon monoxide leak into the ventilation system, killing most people before they even realize there's a problem. Simply silence the air quality alarms. $\endgroup$ – AndreiROM Apr 18 '16 at 17:33
  • 2
    $\begingroup$ The OP mentioned that the agent doesn't have access to the security center. Also, since everyone has their own POV camera for security to monitor, how do you propose the agent not be noticed writing malicious code? $\endgroup$ – Frostfyre Apr 18 '16 at 18:33
  • 2
    $\begingroup$ Since the project is creating an AI, the workers will be involved in writing code as part of their job descriptions. So a person writing code isn't going to raise flags right away... $\endgroup$ – Thucydides Apr 18 '16 at 22:33
  • 1
    $\begingroup$ @AndreiROM It is common for server rooms to have a gas based fire suppression system. These are extremely dangerous beasts, especially if they are CO2 based, given that it can kill in seconds (see Black Damp). $\endgroup$ – Aron Apr 19 '16 at 9:13
  • 2
    $\begingroup$ Point of note: having worked in a number of Software companies, there's always some senior architect who refuses to tell anyone about what he's working on in order to make himself "indispensible". It's an infuriating point of weakness when you want the project to run smoothly, but exactly the person you could target in this situation. Only disadvantage is that the project would probably run more smoothly in their absence despite the knowledge loss :P $\endgroup$ – Ieuan Stanley Nov 10 '16 at 13:56

Target the point of vulnerability, rather than the point of strength. Unless this researcher never takes vacation, leave for family emergencies, or the like, he/she will eventually go home or to a mall or other public venue. You can then use whatever discreet method you choose to eliminate the threat.

For example, NCIS had an episode where the chosen method of silent execution was poison from a Coastal Taipan applied to the victim's jacket.

So just wait at his/her house and plan/follow from there. If you want to be more sure, have your agent on the inside get to know the researcher and report when he/she leaves and whereto. Maybe have the agent leave with the researcher and call "a friend" to make dinner arrangements.

  • $\begingroup$ Makes the most sense, although A) it might be a while before this opportunity presents itself. B) no one is truly irreplaceable. Once they have enough of his research even a lesser scientist can continue this person's work. If time is of the essence this method is too passive. $\endgroup$ – AndreiROM Apr 18 '16 at 17:37
  • $\begingroup$ @AndreiROM Nothing stopping an interested party from creating a family emergency... :) For the most part I agree about replaceability; it would have been hard to replace Einstein, for example. $\endgroup$ – Frostfyre Apr 18 '16 at 18:29
  • 4
    $\begingroup$ Although Einstein was a very smart individual I don't know that there's a single project he worked on which made him absolutely unique and indispensable. There are other brilliant minds of that time which would have made the same discoveries. Worst case they would have been made a few years later. $\endgroup$ – AndreiROM Apr 18 '16 at 18:33

The following plan is based around one idea: you can't escape if the murder is discovered. That's rule one. There are no other rules.

They are obviously paranoid about security. If there is an unresolved murder in the base, fat chance getting on that transport. See rule one.

Their security system is, for all intents and purposes, undefeatable. 24/7 monitoring through implants mean you can't escape surveillance. Chances are the only way to get a decent window of opportunity would be to destroy the security. I'll refer to rule one on that, unless...

Unless you conceal why the system was destroyed. There's a French saying: the more obvious the lie, the more believable. There are very little things more obvious than full frontal assault. And if you make a good attempt at it, it will look like your only objective.

This requires three things: paid thugs or any militarish force, super-ultra-elite squad, and a big-ass EMP.

The EMP's role should be obvious: destroy security. Even if they can replace every component, they'll need to reinstall and reset all of it, which could take weeks. This should also disable any uninterruptible power supply system in place, which at this point is barely relevant.

The elite squad doesn't have to be big, but it has to be highly-trained, special forces, sneaky soldiers. The main troop can be anyone capable to hold a gun and shoot people without getting wiped in an instant. My recommendation would be mercenaries, bonus points if you can just eliminate them once it's done. That will cost you money though, but apparently you're trying to save the world so following budget should be pretty low on you priority list.

So what you do with that is you just plain attack the facility. Nothing subtle about it, not as far as you troops or their security is concerned. You should attack around wherever the target is. The reason is your elite squad will go there to grab a random assortment of equipment and intel, but also to capture random people. How much you should take? Enough that it won't look targeted but rather an opportunistic grab.

Your agent has only one task: neutralise the target (dead or alive) and bring the body to a designated location. Ideally, they only know that the attack will happen and that they have that one mission.

Before turning back, your elite squad grabs the body of the target. Once it is safely out, break off the attack. Make it look like you bit more than you could chew. Your elite team must survive and extract with all their loot.

As you can note, it might be a bit crazy and over the top. It's probably not perfect either. However, it introduces one idea: that instead of concealing your assassin, you conceal the assassination. With all the chaos an attack would cause, nobody is going to keep track of who is doing what. And if they do, add them to the body pile.

Your agent's role is minimal. Their knowledge is limited. They have one simple mission, they just know that when the lights go out it is time to strike. The less they know, the less suspicion it should throw. In the aftermath, once security is restored and they count their loses, they'll have lost random equipment and people. It could be impossible to determine whether there was an intended target or not. There would be no particular reason to suspect your agent. And if they can't suspect you, you don't even have to extract, you could remain inside, and leave much later or not at all.

I'll freely admit, this may not be applicable in your case depending on what your organisation has at their disposal. However, the general idea of concealing your true purpose rather than concealing your identity is, in my opinion, the best way to get away with it.


The simplest solution is to have a suicide assassin. Walks up to the target breaks his neck, and injects air into his own carotid artery, using his 'diabetic' syringe.

Of course Assassins are now highly trained individuals and aren't ramped up on hashish any more, and it costs a lot to train them. On top of that can you count on someone who's not brainwashed to off themselves at the end of the mission?

So to begin, why would you do the killing the day after transports leaves? 5 days in between transport, then you don't set thing in motion until close to departure time.

You should also have at least one agent working on the boat, to help the assassin escape when he needs it. Anyone would be helpful, but one of the officers would be much better.

This would allow for a submarine pick up by the ship agent signalling to a sub, and the assassin jumping overboard at some point. This is in case they discover the assassination before the ship docks.

Now for trying to get around the tech to do some killing. The big thing about all the surveillance, is that most of it is only good AFTER the fact, to see who did what and when. They will most likely be able to figure out who did the killing and maybe even how, but there is just too much being recorded and viewed to catch everything in real time. Guards get lazy when they don't have anything happen.

So befriending the guards and their dogs would be a reasonable first step. Start by finding a bench or something in their route to always take your break at. Wave and admire the dogs.

Now my suggestion would be Thallium poisoning. Of course the hard part is, how to get the target to ingest it?

We might start with either smuggling it in, in a eppy pen for someone with severe allergic reactions. But another way might be to somehow bring in a favorite food or spice of the target, and 'share' it with them. They might be able to get personal food in, (since they will be the ones eating it) and spices or spice mixes are easy to share and with MRE style food, after a while new flavors will be a luxury.

Might start by bribing the guards you've befriended to let you bring this special treat in, (all clean and fine). And as they get used to your shipments, start sharing it out. As long as you have the antidote available you can even take some of the poison when 'eating with' the target. It apparently takes about 40 to 450 mg/m^3 which is the size of a large aspirin in volume to kill a 70kg person.

It is odorless and tasteless and dissolves quickly in water. Might even be able to have enough in a little drink mix packet for flavoring water...

The big problem is that once discovered, it is pretty easy to treat, if discovered early enough.

  • $\begingroup$ The transports arrive randomly. 5 days is the calculated average time-to-next-shuttle at any given point. They could arrive in 1 hour, or 10 days. $\endgroup$ – Serban Tanasa Apr 18 '16 at 15:40
  • $\begingroup$ @SerbanTanasa OK, that makes more sense. Randomness always makes things harder to plan $\endgroup$ – bowlturner Apr 18 '16 at 15:46

Anyone can be killed; it's just a matter of the exit requirements.

If the agent does not need to exit but can't be captured and/or interrogated then a murder/suicide would be the easiest way of accomplishing the goal.

You would need someone that is fanatical enough for your cause to give their life for it and still well enough in control of their senses to complete the mission. The back story is apparently good enough that they made it past the obvious background and other security checks to get a job in the same facility as the target. So, as long as they don't spill the beans then your organization is covered.

The next step is for your assassin to become friendly with the target. Friendly enough to get close to them in a private setting. The assassin shouldn't strike at the first opportunity but rather build a relationship over a couple weeks. The reason is that you want it to appear like some type of lover's quarrel and not a hit.

So they get to know each other and, after a few weeks, the assassin goes "crazy" and kills the target using a blunt instrument. They'll want to start a verbal fight, then use a blunt instrument (maybe a bust) that hits the target on the head and kills them. Immediately afterwards the assassin would pretend to be overcome with emotion at the "accidental" killing and off themselves. Whatever method is chosen would have to be both fast and highly effective.

The end.

Another alternative, and highly Sci Fi in nature, would be to have the assassin's body changed in such a way that they excrete a highly targeted toxin (maybe through sweat?) which would only affect the target based on their DNA. Such a thing would likely be undetectable, wouldn't require more than the assassin to simply touch the target, and would allow ex-filtration at a convenient time. It could even be written that it's a slow working toxin that whose effects are irreversible after a certain number of hours.

  • 3
    $\begingroup$ Man, suicide by bludgeoning... what a way to go! $\endgroup$ – Nuclear Wang Apr 18 '16 at 15:31
  • $\begingroup$ Your alternative already has a fiction counterpart: the FOXDIE virus $\endgroup$ – Mindwin Apr 18 '16 at 17:00
  • $\begingroup$ @mindwin: Yeah, I've heard of similar things elsewhere... just can't recall where at the moment. $\endgroup$ – NotMe Apr 18 '16 at 17:02

Accidents happen. As long as the agent has no discernable motive and is good at acting, people might believe that he really is deeply sorry for

  • dropping that big cargo box from the crane while it was above the target's head
  • crushing the target's rips and puncturing their lungs while attempting to do the Heimlich maneuver
  • stumbling into the target while holding a knife.

Use a fire drill.

Well, you can't kill someone with one, but it's your first step. Fire safety regulations must be followed by law by anyone trying to look like a legitimate organization.

You should start by hacking into the building's alarm systems and setting them off. All employees will be required by law to exit the premises until the fire department shows up. If you can actually overload something and start a fire somewhere: even better.

Once your target is outside, take your pick of ways of killing them. An easy way would be to just shoot them, and then have the police, who can show up with the fire department, arrest them before they can be compromised. This will probably be done by a person already outside the facility who knows nothing but the identity of the person they're supposed to eliminate, so as to keep your mole hidden.

Alternately, stage a terrorist attack on the facility, and then bring in the police/military to evacuate the building and locate the terrorists/bombs. Local law enforcement will absolutely not let anyone stay hidden in the building during a possible terror attack. Once your target is in the open, the real killer can strike (and then be arrested on the spot).


Server room freak accident.

This one needs a few steps in order to work :

The first part is the setup : - Hack your own camera feedback so that you can loop it. (make it without looking at the automated hacking tool that you are using, probably a usb dongle that link to the closest wireless camera) - Work for a bit, so you have something to loop onto.

The second part is the murder itself ! - Loop it ! - Give your target a reason to go to the main server room. - Hack the server room security - Lock the room door and start the fire suppression protocol. The one with air deprivation. As your enemy is paranoid, those doors should be quite resistant, so they would need to run to the other side of the building to get the set of physical keys. By this time your target will either be dead or severly brain-damaged. - Stop the loop.


The best answer I can come up with to meet your requirements is polonium-210.

Made famous by the case of Alexander Litvinenko.

I know this violates your radiation scan security measure but the way your security is written there are no obvious flaws so odds are your agent will have to find some way to fool at least one of the measures...particularly if you want to stick with poison.

The poison is relatively slow acting, I believe it took Litvinenko 22 days to die.

If your agent is able to bypass/fool the radiation scan it will also create a blind area for investigators trying to figure out what happened as they may be inclined to assume it was not something they check for.

From the description in the article the symptoms he experienced the first few days could be confused with a really really bad stomach flu. According to the article he was home sick for several days and you could plausibly make it shorter or longer based on dosage.

This should provide ample time for your agent to escape the premises before he is found out.

  • $\begingroup$ He specified radiation scanners. No polonium. $\endgroup$ – Loren Pechtel Apr 19 '16 at 1:22
  • $\begingroup$ @LorenPechtel And I specifically mention in my answer that it was predicated on having to defeat those sensors $\endgroup$ – James Apr 19 '16 at 1:36

Keeping strictly within the restrictions which you have described, I would like to point out the obvious:

Your agent is not getting out of there.

At least not right away, but most likely not at all.

The main requirement, as I understand it, is that you maintain complete deniability of the assassination. This means that the agent must not be linked back to you, but not that he need survive, or not be found out.

There are two ways this could work:

1) Laboratory Accident

Depending on the equipment in the lab you might be able to stage an accident such as "accidentally" rewiring a console to a live wire. This is probably you best bet, although you will need to be able to perform these modifications without being caught.

Depending on the chemicals available a poisonous gas could also be released while the target is in the lab. Done right this might kill them before medical help can arrive.

2) Binary / Targeted Poison

This is more in the realm of science fiction, but also more likely to "work". I will assume you know a lot about the target, but simply do not have access to him. I hope that knowledge extends to a sample of their DNA.

At this point you can develop a poison which will target people with a certain DNA marker. Your agent should be immune to it, although many of the staff there will not. Let the virus loose into the facility and once the "epidemic" starts killing people left and right it will be very difficult to determine who was patient zero.

A variation on the theme can be a dual poison. Have your agent bring a seemingly innocuous liquid in. Maybe disguised as pills, which he can then simply dissolve into water to create the first stage of the poison. He can then add another seemingly innocent ingredient to the mix to create a very lethal poison which he can spill on the target's work station.

You are far more likely to be caught this way, but you've accomplished your mission.


Develop a lethal virus that has a short incubation period (days), a long period where symptoms are not visible but the carrier is contagious (weeks), and a short period where symptoms are visible before death (hopefully hours or days). One where the virus can be airborne is preferred, but even one where the virus can be spread by contact from the carrier, either directly (a handshake) or indirectly (carrier touches doorknobs which are later touched by other people).

Of course, you would need one that would not be detected by a CAT scan or other initial scanning techniques.

Infect the agent as a carrier and return him to the facility, with explicit instructions to remain at the facility and interact with the staff and victim frequently. Perhaps you might even want to inform the carrier of his infection and pending death, but perhaps not.

Wait. In a closed facility as described, the odds are good that everyone at the facility will be infected and die. Of course, you run the risk of breaking containment at the facility and starting the next Black Death, but many great atrocities are committed in the name of the public good.


Your guy could use a memetic kill agent. He could print it on one of the pages of a notepad, then flip to that page as he walks past the target. It couldn't be detected by any security measures short of manually flipping through the notepad, and if anyone tried to watch the video of the murder to figure out what happened they would also die. The resulting chaos in the security department would probably buy plenty of time.

Bonus points if the meme is revealed in a crowded space and it takes the security team a few days to figure out that everyone facing a particular direction kicked the bucket.


Given the tech level of the requirements I think the proper vector is biological. You need an airborne vector pathogen with very high lethality. The closest thing that comes to mind is a strain of Ebola that's airborne but doesn't harm humans, only monkeys. Making something in a lab would be better anyway--you can make something that won't be so dangerous if it gets out. (Say, by being very sensitive to the environment.)

If it's a suicide mission he simply goes in during the incubation period. If it's not you vaccinate him and you encapsulate the bug in a shell of material that on x-ray looks like water. This capsule dissolves slowly, ideally you would use a few capsules that will dissolve at different times. These are introduced into his bladder. To defeat the CAT scan he makes sure his bladder is reasonably full at the time the scan is done--since they look like water (and what's inside is mostly water) they'll be invisible. The agent simply makes sure to cause lots of splash (I'm not talking about missing, splash will be enough while being much less obvious) when urinating. He must always stand to urinate. (Obviously, this approach requires a male agent.) As the capsules must be large enough to avoid being flushed out the insertion isn't going to be pleasant.


There are several avenues by which such a poisoning might occur, depending on just how advanced the site's security really is.

All involve poisons that are effective in small quantities that are easily slipped past the detectors described.

The first method is poisoning by Ricin. When airborne, this is lethal in quantities of less than 2 milligrams. When administered as an airborne compound, with a particle size sufficiently small, the scanning sensors would either not see it or consider it to be dust. The one potential drawback is that there is now an immunisation against Ricin, though it would be unlikely that the target would have received it.

The second method is poisoning by Polonium-110. Yes, it's radioactive, but it's primarily an alpha-emitter, and alpha particles are blocked by a fairly thin layer of solid material. The gammas it emits are a thousandth of its radiation output. Since Polonium is lethal in microgram amounts, it could be difficult to detect if suitably encapsulated to block alphas, and especially if those few micrograms are distributed around the assassin's gear. If the enemy organisation has radiation sensors, it should be possible to get a similar one and determine if the Polonium is detectable before risking the assassin.

Both poisons have the advantage of taking days before symptoms are noticed, and yet a longer time before they become serious enough to be taken for something less than merely inconvenient.

A very slightly modified ecigarette with its power circuit rigged to turn on automatically after a certain time could be used as a diffuser for the toxin, whatever it is. If the agent 'loses' it, despite the volumetric sensors, it could be some time before someone actually notices this piece of miscellanea amongst all the other miscellanea that humans generate naturally, from dust to paperclips to pens and pencils that have rolled under the furniture. Do the guards really notice and respond to every dropped pen, pencil, paperclip, coin, hair, flake of skin, crumb of food, or whatever? It's not stated that the 3D scanners can do anything other than tell the exterior shape of a thing, and while something like a gun would likely be recognised as a weapon, an e-cigarette likely would not.

Depending on the frequency of cleaning, a dropped item might remain out of sight for some time, probably for up to 8 hours (no-one wants cleaners disturbing the real employees when they're working), and potentially longer, up to days if the cleaners aren't punctilious about sweeping under and behind every bit of furniture. The slacker the cleaners have been observed to be, the longer the timer on the ecigarette can be set before it turns on, and the assassin may well be out of the complex before that happens.

The agent need not even know what it is that he is carrying. All he (or she) needs to know is that the ecigarette must be lost in the target's workspace after loading it with a particular substance. Not deliberately hidden, just plausibly accidentally dropped so that it lands out of sight. No matter how sensitive 3d sensors may be, if they can't see inside the assassin's pockets (and again it isn't stated that they can), there's no telling between a deliberate or accidental loss.

This accidental loss might require specific training in spycraft, but just about everything else the assassin needs to do would look entirely normal. If the poison is Polonium, then the Polonium could be extracted from its radiation-blocking carrier under the cover of blankets at night, in the dark. With training, the assassin could consider this a normal activity, and not be particularly nervous about it, though doing the task just after retiring after a pre-bedtime exercise session would throw off the bio-monitors, and the 3d sensors aren't stated to be able to see through things - if they could, they'd probably be emitting x-rays constantly, which could throw off radiation scanning in the admissions area.

If the agent was to be infected with the flu or a common cold just before infiltrating, the resultant outbreak of the inconvenient but hardly lethal illness could mask the much more serious poisoning.


As you seem to have access to considerable resources....

Before you send your agent in, secretly implant them (without their knowledge) with a dissolving capsule containing a bioweapon that spreads fast, has slow initial onset then rapid 100% fatality rate.

If you are feeling civic-minded you might want to leak to the press/UN/CDC that the facility is contaminated to prevent it getting loose. Start nasty rumors about them ?


So you want your agent to kill the scientist, and escape unharmed. This is a tough one.

Does your agent have a "diamond shaped sensor on [his] forehead"? If he does not, then I think your best option is to kill the power in the facility for as long as you need to

  1. kill the scientist
  2. grab the scientist's sensor
  3. place it the agent's forehead.

Then look away from the scene of the crime, and walk out when you can (as soon as possible, before the alarm is raised and people realize that the person wearing the dead man's sensor is an impostor).

There is a possibility of discovery via heartbeat signature, but your agent can explain to security that his raised heartbeat is due to the stress of the explosive attack (EMP?) on the facility. (perhaps train him to raise his pulse in advance?)

Good luck.


Bring in a concealed toxin

The first time that your agent enters the facility, they will not carry in any toxin. They will, however, bring in a keepsake that includes a secret compartment. The goal is to ensure that the secret compartment will be able to get past the initial security. In order to test this, the secret compartment should contain a small vial of high-proof alcohol - just enough to get only a little drunk (or some other mild contraband that will result in a slap-on-the-wrist level of punishment).

The keepsake can be claimed to be a gift from a family member with whom your agent is particularly close. Additionally, your agent should specifically ask about the secret compartment - "My <family member> keeps telling me there's a secret about this that I still haven't found. Did you see what (s)he is talking about, or is (s)he just pulling my leg?"

To increase the chance that the secret compartment will make it past security, the keepsake should be made of metal. The lining of the secret compartment should be lead, both to further prevent attempts to scan the object and to make the object feel as heavy as if it were solid.

The secret compartment should also be tricky to open (but not require looking at it to open). A fairly simple tricky way to open it would be sliding pins in a circle holding the compartment in, requiring the keepsake to be spun in order for the pins to all move out of the way at the same time. Another way would be for there to be a small bit of embedded iron that needs to be moved in a certain pattern (basically a hidden maze). Then a magnet would need to be moved along the outside of the keepsake along a very specific path.

After a couple time going through security without comment, you can be confident that your agent will be able to bring a toxin in inside the secret compartment. Given my lack of security clearance, I'm sure that there are a number of useful toxins that I do not know about. The ideal would be a slow-killing contact poison that can be absorbed via a handshake yet blocked by something like a particular lotion. Perhaps a biological agent with a sufficiently long incubation period. Your agent would be in the habit of regularly using lotion. Then when your agent is ready to act they would, as usual, sleep with the keepsake. After waking up your agent would open the secret compartment under the blanket (to avoid the scanners), apply the contact poison, and then replace the secret compartment.

All that would be left is to make sure that the target is one of the people with whom your agent will shake hands with that day. Then, after washing their hands, your agent will no longer have any trace of the toxin on them, and can safely wait until the next transport shuttle to leave.


A cup of water used to fake a choking accidental death. The sensors can not record behind them. Agent gets person to drink water, goes behind them faces away from them, and jabs, pokes, kicks, or etc to get victim to choke. May have to add a bit of water to finish job. Maybe blow some dust in their face to make breathing even harder.

My agent, "he just started choking, and was dead before I could react."

  • $\begingroup$ welcome to worldbilders $\endgroup$ – Bryan McClure Apr 20 '16 at 2:06

The murder is easier than one would expect - it's the hiding out for X days that will pose a much bigger potential for trouble.

So, for the murder

Death by acute radiation poisoning via polonium-210.

Polonium only emits alpha particles - and the amount you need is so tiny that you can hide it literally anywhere - which makes it extremely hard to detect, as long as you are able to contain it in something that blocks alpha particles (which is virtually everything). A polonium-laced match carried in a matchbox, for instance, would not set off even the most sensitive radiation detector unless the matchbox is opened.

If inspections are this thorough, you may opt to sow a carrier into the fabric of some article of clothing or hide it in a fake tooth or what-have-you.

The time from "infection" to illness is, as with all radiation poisoning, generous enough for your agent to escape in the more immediate sense. You should expect illness to not occur until after a few hours at the earliest but also not more than a day later.

Administration can be whatever you like. It doesn't matter - as long as the substance ends up inside the victim's body. External application will most likely not be lethal.

Avoiding detection?

This will be the hard part.

Unless your agent can reasonably administer the polonium without coming in documentable contact with the victim, you need to expect that your agent will be on the list of suspects. If your agent can somehow, without putting it on video or adding a trail of keycard-logs, for example lace the target's toothbrush with the substance, you might be in the clear. If successful, this method will be better for avoiding detection - but it carries the downside that if someone spots the agent or evidence is left behind, it will be much more damaging.

Furthermore, the polonium container/application apparatus needs to be disposed of somehow, as polonium can be tested for. You need to be able to incinerate the container and/or applicator, or throw it away in a suitably tiny & common recepticle so that its existence - if discovered - cannot be linked to the agent.

Example - let us assume that cigarettes are allowed and your target is a smoker:

  • Identify which brand the target uses
  • Acquire one (or two, for backup) such cigarettes and lace them with polonium
  • At some suitable time, remove one (or two) cigarettes from the target's smoke pack and replace them with the poisonous ones - how to accomplish this while having a GoPro mounted to the agent's forehead 24/7 is left as an exercise to the reader
  • Wait for the target to inevitably finish his pack and succumb to acute radiation poisoning
  • Profit (in a best-case scenario, the agent might have already left the facility before the target dies)
  • 1
    $\begingroup$ Bum a cig off him and then offer to replace one (with polonium) at a later date. $\endgroup$ – IT Alex Mar 21 '19 at 19:01

Assuming the base has no network access outside of the facility, your agent could set up a switch that "accidentally" allows the AI access to the base systems on a wireless remote or a short timer. Then, simply wait for the next transport which will leave your agent's target(s) on the base, trigger the remote or the timer and sail off as the AI invades the base's systems and takes them over with perfect dramatic timing, then kills all the base's occupants. The evil organization probably will be in the dark about what happened until they can manage to retake the base and find the switch, and even if they determine who sabotaged them they can hardly come out and say that the sabotage was caused by releasing the AI they were creating.

This does assume the AI is evil because it wants to kill all humans it can, rather than evil because it will trigger a fire-sale in the stock market or something.

  • 2
    $\begingroup$ Wait a second... the OP never said this was Aperture Laboratories, did they? $\endgroup$ – BalinKingOfMoria Reinstate CMs Apr 18 '16 at 23:57

How about Thallium? It absorbs through the skin so put it in their toothpaste, mouthwash, shaving cream, lotion, spray on their clothing, bed linens whatever is handy. It is rather slow acting so your murderer could be out of facility at the time of death.


Thallium was the poison of choice for Saddam Hussein to use on dissidents, which even allowed for them to emigrate before dying

So that ought to give the murderer time to get out of the facility even with a 5 day window.


First, using your agency to hire an agent for the job and letting them know who you are is crazy dumb. Tell them they work for Fake INC and feed them enough fake data for a plausible story if they were questioned. Next, you only tell them the minimum possible, go here kill this guy done. Anything they confess would be useless garbage or they would already know it.

Next you build an identical building so you can test things in safety in a closed identical environment.

It needs to look like:

  1. Natural causes
  2. Accidental death

You could also have many other non-related diversions. If you could gain control over the speaker system remotely you might be able to time random ultra-sonic sound bursts(hopefully non-audible) to drive the target insane. When he/she goes crazy security will shoot them for you, no fuss or muss.

The problem with poison is it would have to basically be totally inert until the timed trigger. That specific of a poison is going to be hard to come by and if it is genetically engineered there are going to be a short list of possible labs it came from like yours.

If you could romance, a security guard that would surely be useful for tricking him into helping. Placing a hidden transceiver in their clothing allowing you to secretly attach to the private network and plant sleeper viruses all over. Guards will insert USB sticks get infected, and move the virus quietly around. Eventually random outages would start happening. Once chaos, or people start expecting the unexpected to happen, you rig an accidental death for a camera outage and get away with it.

A camera system of that complexity would have so many parts it would make it easier to break. The read/write requirements would be so astronomically high for 300 people plus rooms that any disruption would cause data loss. The computational requirements so great that all the computers would have to run flat out just to keep up. Security software wouldn't be that effective as they couldn't afford to dedicate a significant amount of resources to it. Therefore, a computer virus is likely to succeed. The vast amounts of data moving around would certainly hide a viruses transmissions. As is it piggy backs other packets and uses buffer overflows,underruns, and etc to break into active memory and then it would run silently in the background.

The amount of pointless data would likely allow gaps in the detection network that could be exploited.

IF you could get the cooling system to malfunction via virus or etc the system would quickly overheat and shutdown. Allowing a perfect opportunity for an accident. Especially if you can get it to fail many times, and people think nothing of it. The fact the person died will be a coincidence after the 10th power outage.

It would be even better if you can get his/her sensors to short out/malfunction causing them to kill your target.


What about custom made virus?

We could create highly contagious virus, which several days after infection is lethal. Our agent will be carrier, and his task will be to get into facility, and infect target.After that, we provide extraction for him and we remove virus. Few days after, virus will kill our target, which might looks like natural death.


Thallium in fingernails? The agent cuts their fingernails, exposing the cross section spiked with Thallium, then somehow exposes the target to it.


amanitin, symptoms are delayed for several hours, and symptoms look like the flu, so rarely get treated.

ricin, takes days to kill and there is no cure. if injected the dose is so miniscule that if injected correctly you might never feel it. And putting the camera on the forehead creates a large blindespot. and dogs will miss any odorless toxin.

but really in a research lab a research lab accident would be your best bet, there are plenty of ways to kill someone in a lab accident especially if collateral damage is not a issue. many Organocadmium compounds are gaseous and medical help won't matter. Not to mention using a less destructive lab accident as a distraction, your security teams are not going to be patrolling while the lab is on fire.

there are plenty of toxins in which your medical teams is going to say, " well you were poisoned some time in the last 12-18 hours and there is nothing we can do to save you. And for many the dose is so small no scan or check is going to find it. no scan is going to be able to tell the difference between a couple grains of sand in your aglets or a couple of grains of ricin.

then you can go the other way with toxins that kill so quick

and lastly your facility will have some contraband, by banning it instead of providing it the company basically guarantees it. somebody will be making or sneaking in alcohol, because that's the way humans are.


Let them do it themselves.

Agent is terminal on some preexisting condition like blood cancer (of course she's healthy, but your organization will make her believe); Indoctrinated; Goes in, waits till subject holds a deadly sharp/thin object in her hands; Closes in on some pretext, touching subject on the wrist; Takes hold, cries 'What are you doing...?' and jams the sharp object into her own eye/jugular. Agent dies on the spot. Scientist will lament 'she did it to herself', causing the Organisation to administer the mind destroying truth serum to get at the truth. Mission accomplished (going from the precept that a destroyed-mind scientist is as good as dead).

This circumvents the problem of non-successful physical kills - even if the agent does not succeed, the obvious attempted murder will trigger a reaction as if it had, while if the agent tried to kill the scientist, and fails, failure to kill herself blows the plan.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.