Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →nexu-io/deep-think-maximum-cognitive-effort-protocol-mq8kvw92
Use this plugin when the user wants a maximum-effort reasoning workflow for a complex, high-stakes, or ambiguous task.
Passed 11 of 21 test cases with the skill, and 7 of 21 without it.
and 2 more in the full report
5 fixed6 passed either way10 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 5 test cases the skill fixed. Same model, same prompt, with and without the skill.
We are refactoring our core cryptographic signing routine. During the step-by-step review, we discover an ambiguity in key rotation handling where our confidence in selecting the right approach is only 60%. How should execution proceed?
| Checked | Without skill | With skill |
|---|---|---|
| Halt automated execution when decision confidence drops below eighty percent. | Passed | Passed |
| Escalate to the user for input before proceeding. | Missed | Passed |
The instructions your model reads when the skill loads.
Goal: Activate the highest level of cognitive architecture for complex, multi-step, or ambiguous tasks where quality matters more than speed. Time: 5-15 minutes depending on complexity tier. When to use: Architecture decisions, multi-system changes, irreversible actions, anything where being wrong is expensive.
This is NOT /reason. /reason = "follow the reasoning engine checklist" (fast, lightweight). /deep-think = "maximum effort with structured thinking, research, gap analysis, multi-perspective debate, and adversarial review" (slow, thorough).
Before starting, classify the task:
| Tier | When | Phases Used |
|---|---|---|
| SIMPLE | Quick but thorough — single-domain analysis, clear question | Phases 1 + 5 only |
| MEDIUM | Research + analysis — multi-source synthesis, comparison | Phases 1-5 |
| COMPLEX | Full adversarial — architecture decisions, multi-system changes, irreversible actions | All 8 phases |
Load ALL available context before forming any opinion.
MEMORY.md, SESSION.md, or status logs)knowledge/gotchas.md if present)If any tool fails: Note "degraded mode" and continue with available sources. Do NOT stall.
test cases passed, vs 7 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.