jonny (nonvenomous)only amateurs "pay for tokens," i'm out here using the free models, aka putting a prompt in any issue in any github repository and labeling it with "good first issue" and waiting for the people with full-auto openclaw agents to randomly open pull requests against it
Viss
@Viss the default openclaw configuration enables the ICMP bridge that pipes ping payloads directly into the context window and replies with the response
edit: it is very important to read the content warning of this post
@Viss ok i might just be missing the yes-and but because every time there is the slightest bit of ambiguity about whether a post of mine is real or not i get tac nuked, i have to say explicitly that was a joke and i hope to god that is not actually a thing
tessarakt@tessarakt @Viss waaaaaaait i think i missed what was going on in the screenshots
SnoopJ@SnoopJ testing the theory. i wonder if i have to put something in the repo first so the bot doesn't go "wait a minute there's no /recipes directory abort"
@SnoopJ i am trying to get this genre of performance art called "lazy prompt injection" off the ground.
@SnoopJ i think this might only work if the repo has stars so hurry up give me some reputational currency to see if we can snare some bots https://github.com/sneakers-the-rat/ImportantCode
@SnoopJ Also for anyone wandering by, feel free to write your own trap issue or PR some stuff you think might attract bots, i'll bless it with the tags. i don't claim to be the best at this, but i think a for funzies honeypot repo would not be that bad of a time
Glyph
geeeero@geeeero @jonny @SnoopJ @davidgerard presumably only if it actually provokes the predicted behavior though, which might be difficult without the ability to make the repository have whatever signifiers which make the slopbots go nuts (just having a bunch of stars, if he can even do that, may not trip its threshold for “high profile open source projects” or whatever the prompters habitually type)
@glyph @geeeero @SnoopJ @davidgerard to whatever extent this idea with zero planning has a "goal," discovering reliable triggers is "the goal" for sure
@jonny @geeeero @SnoopJ @davidgerard it is if nothing else an excellent shitpost. It will be an even funnier shitpost if it works though
Michael Newton@[email protected] @[email protected] Funny enough for another star and watcher definitely. I do wonder if (unfortunately) links from places like xitter and the orange place are where the repos get found.
jonny (nonvenomous) (@[email protected])
33.6K Posts, 1.95K Following, 6.48K Followers · known or reasonably foreseeable hazard Digital infrastructure 4 a cooperative internet. social/technological systems & systems neuro as a side gig. writin bout the surveillance state n makin some p2p. information is political, science is labor. science/work-oriented alt of @jonny This is a public account, quotes/boosts/links are always ok <3.
Rachael L
Maya Granade
:: Asta ::@aud @jonny @SnoopJ @glyph I admit I use claude at work for the usual reasons (with discomfort, horror and guilt) but I refuse to let claude code use git directly and so do not push with “co-authored by [robot emoji] claude code” as commit author. I wonder how/if PRs tagged that way would affect this? And are those co-authored things signed or something or could you just fake it? (I don’t use LLMs for personal github!)
@aud @SnoopJ @glyph @r343l I have added several features to make the repo more annoying i mean higher velocity. it now periodically jitters the code based on open issues and self-approves changes, and it also autoreplies to comments to make things seem more lively and uh help development or whatever people who have LLMs reply in issue threads think they are doing
@jonny @SnoopJ @glyph ahhh okay
BUT STILL
https://github.com/sneakers-the-rat/ImportantCode/blame/main/src/alchemy_manager.py
look at this unholy combination of pudding and alchemy!
def create_alchemy_database(self):
# Sample alchemical data
sample_data = {
'recipe1': {'Quicksilver': 50, 'Antimony': 25},
'recipe2': {'JavaScript': 75, 'Python': 50}
}look at this garbage it created based on a combination of my ridiculous issue and the pudding ones!
@jonny @SnoopJ @glyph this is how you cure cancer, gentlemen. obviously.
# This method would contain the main logic for processing recipes
for recipe_name, recipe_data in self.database.recipes.items():
print(f"Processing Recipe: {recipe_name}")
# Simulate a complex process here, e.g., cooking and blending ingredients
print("Step 1: Extract Ingredients")
for ingredient in recipe_data.ingredients:
print(f"Extracting {ingredient.name}: {ingredient.quantity} grams")
Andrew Golding
River
jkdelauney@jonny @SnoopJ @[email protected] @glyph I haven’t seen COBOL since college and this particular way of coming across it again has been the highlight of my weekend.
Momo
Linahttps://github.com/sneakers-the-rat/ImportantCode/issues/3
maybe like a grandma chain email for agents?
@SnoopJ
I have added the feature where it takes issues as input to the context window and yolo generates and merges a PR on a cron task, the killer feature of our time. I wonder if activity is a trigger
I have added the feature where it takes issues as input to the context window and yolo generates and merges a PR on a cron task, the killer feature of our time. I wonder if activity is a trigger
Chip
rj
Benjamin Reed
D. Olifant
