atomise
Atom of Thoughts (AoT) reasoning - decompose complex problems into atomic units with confidence tracking and backtracking. For genuinely complex reasoning, not everyday questions. Triggers on: atomise, complex reasoning, decompose problem, structured thinking, verify hypothesis.
85
Quality
87%
Does it follow best practices?
Impact
73%
2.28xAverage score across 3 eval scenarios
Passed
No known issues
Evaluation results
83%
68%Race Condition in Distributed Cache Invalidation
AoT output structure and atom fields
Phase 0 restatement
33%
100%
Premises extracted
0%
62%
Approach sketch
0%
50%
DECOMPOSE step present
16%
100%
Atom structure fields
0%
100%
Correct atom types
0%
100%
Hypothesis confidence cap
0%
25%
Confidence propagation
0%
62%
VERIFY step present
0%
71%
CONTRACT step present
0%
100%
KEY CHAIN present
0%
100%
ANSWER section
33%
100%
CONFIDENCE with explanation
66%
100%
RISKS section
100%
100%
Without context: $0.2378 · 1m 26s · 13 turns · 20 in / 3,970 out tokens
With context: $0.4718 · 2m 29s · 21 turns · 24 in / 7,509 out tokens
61%
14%Authentication Flow Security Assessment
Confidence adjustment and backtracking
Multiple hypotheses decomposed
100%
100%
Hypothesis confidence cap
12%
100%
Verification steps present
62%
100%
Partial verification reduces confidence
33%
33%
Refuted path shows major reduction
55%
0%
Backtracking demonstrated
0%
0%
No forced conclusion
88%
100%
Adversarial verification style
87%
100%
CONTRACT after backtrack
0%
0%
ATOMS table present
0%
57%
RISKS section present
71%
100%
ANSWER and CONFIDENCE present
60%
70%
Without context: $0.2310 · 1m 33s · 11 turns · 16 in / 4,313 out tokens
With context: $0.6984 · 4m 14s · 20 turns · 27 in / 13,682 out tokens
75%
41%Event Sourcing vs CRUD for a Financial Ledger System
Depth modes and domain verification style
Deep atom tree (7+ levels or high atom count)
25%
100%
Confidence threshold for conclusion >=0.85
100%
100%
Tradeoff analysis in verification
100%
100%
Feasibility checks present
87%
100%
Contract summaries <= 2 sentences
0%
0%
Multiple CONTRACT sections
0%
0%
Confidence is heuristic disclaimer
0%
55%
EVALUATE step documented
0%
62%
Confidence propagation from parents
12%
75%
KEY CHAIN present
0%
100%
RISKS / invalidation section
100%
100%
Atom types correctly labeled
0%
100%
Atoms table with all fields
0%
100%
Without context: $0.2859 · 1m 59s · 13 turns · 18 in / 5,466 out tokens
With context: $0.6850 · 3m 52s · 22 turns · 27 in / 12,695 out tokens
- Repository
- NeverSight/skills_feed
- Commit
5342bca
- Evaluated
- Agent
- Claude Code
- Model
- Claude Sonnet 4.6
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.