It is not like the agents will know if it is agonising or not.
You just tell them it is painless. It is not like they are going to test it on themselves.
If someone somehow ends up whitnessing a usage and they complain you tell them “this is very concerning, we are going to investigate. Maybe a bad batch? Or enemy action?” and then you classify the whole thing for 150 years.
If there is such a widespread leak that you can’t do the above (for example one of your agents swallows poison direct on camera broadcast publicly) then you say to your agents: “Turns out it is not working as painlessly as we thought. Obviously we are not monsters, we couldn’t test it on live subjects. Discontinue old formulation, new painless formulation is being distributed.” And then you give them new pills with the same or nearly the same poison and visibly new packaging/concealment. But that is a last resort, only if there is too much grumbling amomg the rank and file.
- the actual use of poison pills is very rare. If our agents need to pop them all the time we suck and our organisation is hemourhaging valuable resources and we need to improve our trade craft.
- things done in missions are murky, especially murky when the mission went so horribly wrong our agents commited suicide. Most often the agency notices that something went this wrong by an agent failing to report in. It is not like the agent who died a painfull/painless death will complain about it at the “illegal undercover agents anual ball”. Information spreads slowly by default under these circumstances and through unreliable narrators.
- the business of inteligence agencies is to keep information compartmented and on a need to know basis. They have processes with teeth to enforce this. Keeping information available to one part of the organisation away from other parts of the organisation is their bread and butter. Keeping the truth about a possibly painfull poison pill is not a special case there, you can use the same aparatus to keep this info locked down you use to keep other information within the agency.
In either case, the actual information on whether it is painless is immaterial to the choice. The only thing that can factor into an agent's decision is the information that he learns about the suicide agent, and the agency has a lot of control over that information.
It would be pretty easy to tell a plausible lie, though, especially if you can think of a plausible less-painful poison to pretend you used. Tell them it's something moderately painful, so you can still tell them, "look, it's still better than torture". Just don't remind them about the edge-case scenarios where it really has to be fast and reliable, and they probably won't question it too hard...
"I work for an agency that routinely gains people's trust in order to betray it, but I trust them and they would never betray me" seems to be the default belief for this lot, so maybe it isn't so far fetched.
You just tell them it is painless. It is not like they are going to test it on themselves.
If someone somehow ends up whitnessing a usage and they complain you tell them “this is very concerning, we are going to investigate. Maybe a bad batch? Or enemy action?” and then you classify the whole thing for 150 years.
If there is such a widespread leak that you can’t do the above (for example one of your agents swallows poison direct on camera broadcast publicly) then you say to your agents: “Turns out it is not working as painlessly as we thought. Obviously we are not monsters, we couldn’t test it on live subjects. Discontinue old formulation, new painless formulation is being distributed.” And then you give them new pills with the same or nearly the same poison and visibly new packaging/concealment. But that is a last resort, only if there is too much grumbling amomg the rank and file.