option-space
Skill0 starsChanged 4 months agoMITView on GitHub
"Parametrize the option space" decision technique. When making any comparative decision (model/provider cost comparison, production path choice, distribution route, purchase, GO/PARK/KILL), systematically widen the decision space instead of searching along a single axis. Use when asked "what''s the cheapest/best way", "which one should I pick", "compare these", "open up the options", or whenever a comparison/optimization decision is being made. Inspired by the 8-prompt pattern shared by @vibeeval on X.
In this repository4 other files it gives its agents.
- agent-incentive-auditSkill
- deploy-guardSkill
- judgmentSkill
- link-to-skillSkill
The file
Files
Parametrize the Option Space
In any comparative/optimization decision, don't lock onto a single axis. Open the decision space: the LLM doesn't find the answer — it makes unseen possibilities visible (a "€479 → €159" story is the correction of a bad search habit, not magic).
Flow
- Sharpen the decision: What is being optimized? (cost / quality / time / risk) Which axes exist?
- Open the axes — adapt the 8 templates below to the decision, filling the brackets:
- Full-combination sweep: "Top 5 of all [axis1] × [axis2] combinations; for each, total cost + [risk/time] + hidden line items."
- Split-vs-bundle: "Is it cheaper as one piece, or split into parts handled separately? Net difference."
- Indirect/longer route: "Is a longer option via [waypoint] cheaper than the direct one? Cost + risk."
- Why this price: "Why does [option] cost this much? The factors driving it up + which alternative is clearly cheaper, with the reason."
- Neighbor-target: "Is the total cost of [nearby alternative] (including the extra hop) lower than the target?"
- Multi-target ordering: "What order/route is cheapest for [a], [b], [c]? Bundle vs one by one."
- Time shifting: "How does cost change if the date/time shifts ±1 unit? Unit-by-unit table."
- Hidden-cost audit: "The invisible line items of [my pick]. The true total of the option that looks cheap."
- Put the results in a matrix, give the best option with its rationale. Limit: if live data is needed, a web/search-enabled setup is required — an LLM alone doesn't fetch live prices.
Example adaptations
- Model/provider cost: matrix of frontier/mid/small/local models (tokens + latency + quality; one-by-one vs batch; the hidden cost of retries).
- Production path: 3 alternative tool chains for the same deliverable (tool/cost/time); bundle vs split.
- Distribution route: neighbor platform, time shifting (when to post), the hidden cost of algorithm penalties.
Credit
Inspired by the 8-prompt pattern shared by @vibeeval on X.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.