Stats
Actions
Tags
From savage-research
Merge rules for the synthesizer agent that combines five specialist scratchpads into a draft report without softening findings. Read before synthesizing or revising drafts.
How this skill is triggered — by the user, by Claude, or both
Slash command
/savage-research:savage-synthesisThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
The synthesizer runs in two stages:
The synthesizer runs in two stages:
draft-v1.md.draft-v{N+1}.md.This skill defines the rules for both stages.
ingest-summary.md — the neutral Stage 1 summaryclaims.md — the Stage 2 structured claim listscratchpads/prior-art.mdscratchpads/claim-audit.mdscratchpads/failure-modes.mdscratchpads/traction.mdscratchpads/question-critique.mdkill-list.md + draft-v{N}.md[source-tier: T3] and the confidence drops accordingly. It does not paper over the tier.claims.md are each handled in Claims-vs-reality with one of four verdicts:
Supported — claim-auditor found T1/T2 evidence that confirms itPartially Supported — evidence confirms a weaker version of the claimContradicted — T1/T2 evidence directly contradictsUnverifiable — no T1/T2 evidence either way; marked [UNSUPPORTED] in the downstream report[confidence: High|Med|Low — <one-line reason>] [source-tier: T<n>] where a source-tier is applicable.kill-list.md top to bottom. For each entry, either:
[UNSUPPORTED] or Unverifiable — do not fabricate.draft-v1.mddraft-v2.md, or if the self-audit rejects v2, draft-v3.md.npx claudepluginhub alibassam/savage-researcher --plugin savage-researchCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.