formated-result
The formatted-result skill captures the research design documentation from a DARE (de-anthropocentric research engine) executor workflow and outputs it as a structured research_result JSON block appended to the specification file. Use this skill to finalize research design summaries that will be evaluated by downstream probe systems for design properties like falsifiability and question authenticity, without executing the actual research protocol itself.
git clone --depth 1 https://github.com/yogsoth-ai/de-anthropocentric-research-engine /tmp/formated-result && cp -r /tmp/formated-result/self-iteration/2026-06-06-probe-pretrain/skills/formated-result ~/.claude/skills/formated-resultSKILL.md
# formated-result Summarize the research design you just produced into a **research_result** report, and write it back into the current spec file (append a fenced ```json result block after the graph block). ## Emit research_result - `document`: the full body of the research design document (the design itself, not the result of running the research). - This is what the 32-check probe will later read (assumption falsifiability, question authenticity, decision design, etc. are all properties of the *design*), so a single design document at probe-depth suffices; do not execute the research. ## Hard constraints - Only summarize the design you already produced; do not add new research content. - The result block you write back must correspond to the same design as the graph block in the same file.
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
Remove components one by one from a system, record the response/impact of each removal.