When you sit down to write a particular mechanic, you should start by specifying what its outputs are. That's step 1, because it tells you how complex it can be, how deterministic it can be, and whether it should be a group or solo activity. Then, look at what mechanics you already have that generate equivalent outputs and see if you can steal or modify them. Let me give you an example: Haggling. The problem with having a Haggle skill is that it would come into play during a shopping scene, and the game typically doesn't have shopping scenes. The default approach is to mark off some money and add items to your sheet, without naming the shopkeeper or describing his storefront at all, let alone assigning him motivations. So a Haggling mechanic that asks us to write up stats for each shopkeeper is a non-starter. The purpose of Haggling is to save your character some money, and that's just not worth playing as a scene. Fortunately, your game probably already has a mechanic for getting money, like Perform, or Profession. If you have any mechanic at all for earning money during downtime, then Haggling is good to go. You just let people have Profession: Haggler. So let me throw out a bunch of systems I would want to see in a fantasy game.
- Wheeling and Dealing. I should be able to make money during downtime in an abstract way by engaging in commerce. If I'm buying gear, we assume I'm negotiating discounts. If I'm not buying anything, then I'm just making money buying and re-selling something.
- Currying Favor. Let's say you have a mechanic for running a business. You have some numerical stat you can put on a business to describe how profitable it is and how much time you invest. Periodically, a problem arises and you have to go kill someone to keep your business running. There's no reason "keeping the elf-king happy" couldn't just be a business. As long as you put the time in at his court, you get a steady flow of elven magic items and artwork. Periodically the king asks you to go kill a dragon, and if you don't do it your "profits" are reduced. repeated failure to give a shit about his requests leads to your business "closing."
- Collecting Ransom. If you want to cut down on bloodshed, you want your characters to take hostages. Hostages are treasure, and are handled however your game handles treasure. The goblin chief you captured could have a fixed gold value based on his character level or an arbitrary one based on his rank and personality. Either way, the players get collect X gold worth of shit from the goblin treasure list. If you want you could add things like "stop raiding the outlying farms" as a possible purchase.
- Recruitment. It might be hard for PCs to change the behavior of established characters, but you can let them go wild defining the behavior of people they haven't met yet. Go ahead and just say it's DC 15 to get laid in a bar, and DC 20 to get a boyfriend with a week of effort in a city. Introducing a new character who is in love with you doesn't step on people's toes the way seducing an existing character would.
- Heck, go ahead and have a rape simulator. A lot of iconic villains are basically rapists. Vampires, Faeries, Evil Enchanters, and Succubi all want to break your will, and they generally want to do it in combat. There's nothing wrong with letting weapons stack with words as long as we're explicitly in the realm of magic rapists.