Fun fact: I’ve done this too!
I used #1 and #2 to start a D&D campaign. After we brainstormed the setting and a rough map I told the players: “you come together at the Inn, as usual. Drinks come and go and each of you hears a different rumour around that picks your interest. What is it?”. Each one said something simple that was related to their character, each one a nice simple thing I could have spun into a good adventure, but the last one just asked the barman for adventure and got me suggesting they could go into an ancient battleground to try and find magical weapons. Except that later in the campaign I took all their suggestions and built on what could have happened in all these places that they didn’t go. On one of them a small demonic posession became a demon stronghold that blocked the capital and was about to open a portal to bring reinforcements from hell.
Also at some point I asked a player what do his character do. Jokingly he said “Ok, I kick a Tarrasque”. I didn’t laugh and reply “all right, you do. It’s a really small one that goes flying, falls against a tree and then runs away”. Everyone laughed. A couple of sessions later they discovered an imprisioned group of tieflings summoned the devourer of planes to get their revenge, except that they failed the invocation due to anti-magical measures present on their eternal prision. When they were released along with the dragon’s orb of power (the item that kept them imprisioned and got cut in half by one of the PCs cursed sword) I tell the player: “At that point, distracted by the battle, you see a small shadow jump at half of the orb of power… yes, it’s the Tarrasque you kicked weeks ago, now growing to full size and looking at you reaaally pissed”
I made an end-session mechanic where players could try to improve their Bonds with other characters. I gave Bonds a durability score like if they were items, so, In the middle of the session, whenever they rolled a fumble they could chose to fail foward and take damage to their bonds. At the end of the session they could play scenes to try and fix these by roleplaying, so we got some downtime scenes where either their character got warming bonding moments or drama further ensued.
The instant character-relevant secrets might be easier to introduce as a blank Bond that can be either created by the player or suggested by the GM when it makes sense. Like, whenever players choose to fail foward they could put instead points into this “dark secret” bond and once it reaches a value (4 or 6, nothing too high so it acts sooner) their secret is discovered. Either at the start of the game or just when it’s discovered, the player can talk it with the GM and decide what it is.
I’ve also been working on another procedure to create/discover mysteries at the same time. I’ll try to sketch what I have so far:
First step can be prepped by the GM or done with the player’s help, it doesn’t change things too much either way as long as players keep what they wrote in secret. The GM/the group fills 10 cards with simple generic clues: “a corpse”, “somebody dissapears”, “something important is missing”, “tracks of blood”, “a tomb”, etc. All cards are mixed into a single pile. The GM draws and shows everyone the first card to set a starting point for the mistery and then takes 4 cards.
On each scene a card from the pile is revealed, though players have to explain what makes this clue particular, interesting or relevant and write it down on the card. If they can’t link it yet to the case, the GM can reveal one of their own cards as another clue, or save it for later if she can’t make the connection. If the GM uses all her cards, it’s game over and the GM narrates a bad ending: the mystery isn’t completely resolved, the problem continues, the culprit escapes, etc. If the players manage to connect 5 cards with the case and explain it, the mystery is solved and a positive outcome is narrated.
If players need to use any of the GM clues to solve the mystery, the GM can introduce a negative twist or consequence for each one used.
This is totally untested but if you can see any potential pitfalls please let me know. Perhaps the clues can be predefined with a random table like Fiasco’s sets? Still needs a lot of work.