assumption-negation
The assumption-negation skill applies reductio ad absurdum reasoning by extracting a core claim, formally negating it, deriving logical consequences through deductive chains, and evaluating whether the negation produces genuine contradiction. Use this skill when testing whether a claim is logically necessary versus contingent, particularly for validating foundational assertions in research or analysis where logical rigor is essential.
git clone --depth 1 https://github.com/yogsoth-ai/de-anthropocentric-research-engine /tmp/assumption-negation && cp -r /tmp/assumption-negation/skills/assumption-negation ~/.claude/skills/assumption-negationSKILL.md
# Assumption Negation ## Tactics - contradiction-derivation - counterexample-heuristics ## Method 1. Extract the core claim or assumption from the artifact 2. Formally negate it (produce ~P from P) 3. Derive logical consequences of ~P through deductive chains 4. Evaluate whether derivation reaches genuine contradiction 5. If contradiction found: original claim survives this test 6. If no contradiction: claim may be contingent, not necessary ## Budget | Size | Negation chains | Max derivation depth | |---|---|---| | S | 3 | 5 steps | | M | 6 | 8 steps | | L | 10 | 12 steps | ## Orchestration 1. Dispatch `claim-negation` to produce formal negation 2. For each negation, dispatch `deductive-chain` to derive consequences 3. Dispatch `contradiction-detection` to evaluate results 4. If no contradiction, dispatch `claim-refinement` for weakened version ## Subagents - claim-negation - deductive-chain - contradiction-detection - claim-refinement
Experiment-specific - summarize the DARE executor's research design into a clean research_result report, forced to write back into the spec file produced by formated-specs.
Experiment-specific - replaces writing-specs, emits DARE's 4-layer call plan as a clean research_graph schema. Last step forces load formated-result.
loss-1 judge - read a sample's full dialogue and decide whether the user simulator semantically enacted its Policy Card. check-blind.
loss-2 judge - pairwise quality comparison across the n rungs within one topic; decide monotonicity and endpoint separation. check-blind, D1-D5 only.
Strategy: 面对异常的最佳解释推理
Remove components one by one, observe system changes to reveal hidden dependencies and generate ideas from structural gaps.
Map system architecture to ablatable units for ablation studies
Design ablation studies to isolate component contributions in ML systems