steel-manning-synthesis
$
npx mdskill add yogsoth-ai/de-anthropocentric-research-engine/steel-manning-synthesisThe final integration step — synthesizes all perspective attacks, debate verdicts, and challenge results into a unified assessment. Produces the campaign's final verdict with surviving concerns and recommended modifications.
SKILL.md
.github/skills/steel-manning-synthesisView on GitHub ↗
--- name: steel-manning-synthesis description: Synthesize all attacks and verdicts into a final unified assessment with surviving concerns and recommended modifications. execution: subagent prompt: ./prompt.md input: all_attacks, all_verdicts used-by: [steel-manning] --- # Steel-Manning Synthesis The final integration step — synthesizes all perspective attacks, debate verdicts, and challenge results into a unified assessment. Produces the campaign's final verdict with surviving concerns and recommended modifications. ## Execution Spawns a subagent that receives all attack and verdict data from the campaign and produces the final synthesis. ## Why Subagent - Synthesis requires impartial integration across all prior work - Must weigh competing perspectives without favoring any - Isolation from individual attack/verdict roles ensures balanced final assessment ## HARD-GATE Output must include: - Final verdict: ACCEPT / REJECT / REVISE - Every attack addressed (not ignored) - Surviving concerns with severity ratings - Recommended modifications (if REVISE) - Confidence level in the verdict