本文へ移動
cccskills
無料GitHub で公開

project-evidence-synthesis

Use when a set of sources, dossiers, graph records, or experiment notes must be synthesized into project understanding, evidence pressure, uncertainty, and optional strict-review graph proposals.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md2.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Project Evidence Synthesis

This skill turns multiple sources into project-level understanding. It is not a literature review for humans. It is an agent-facing synthesis step before delta proposals.

Boundary

Do not approve sources. Do not mark sources project-core/global-core. Do not mutate strict graph truth directly.

Allowed:

  • compare sources;
  • extract agreement/conflict;
  • identify evidence pressure;
  • record normal project memory updates;
  • propose strict-review graph deltas only when needed;
  • recommend natural next prompts.

Required Reads

Read:

$WORKSPACE/wiki/projects/$PROJECT/project-query-pack.md
$WORKSPACE/wiki/projects/$PROJECT/project-understanding-graph.md if present
$WORKSPACE/wiki/_system/workflows/project-evidence-synthesis.md
$WORKSPACE/wiki/_system/workflows/delta-update-protocol.md

Then query current graph:

python3 "$PLUGIN_ROOT/tools/graph_query_cli.py" summary --repo "$WORKSPACE" --project "$PROJECT" --json
python3 "$PLUGIN_ROOT/tools/graph_query_cli.py" open --repo "$WORKSPACE" --project "$PROJECT" --json

Use project-local paper dossiers when available:

$WORKSPACE/wiki/projects/$PROJECT/papers/<paper-id>/index.md

Synthesis Model

Separate:

  • Question: what project question each source affects.
  • Claim: what project judgment becomes stronger, weaker, narrower, or obsolete.
  • Evidence: result, benchmark, dataset, metric, qualitative finding, or experimental observation.
  • Warrant: why the evidence supports the claim.
  • Limitation: boundary that blocks overclaim.
  • Evidence Need: missing proof required before the claim should strengthen.
  • Next Action: search, deep read, experiment, delta, or park.

Evidence Strength

Use conservative labels:

  • direct: source directly tests the project claim.
  • analogical: source supports a similar mechanism or setting.
  • enabling: source provides method/dataset/tool needed to test.
  • negative: source challenges or bounds claim.
  • weak: useful but indirect, qualitative, or under-specified.

Required Output Sections

# Project Evidence Synthesis

## Current Project State
## Source Set
## Agreement and Conflict
## Claim/Evidence Table
## Evidence Pressure
## Project Understanding Update
## Optional Strict Review
## Human Gate
## Next Natural Prompts

Stop Point

Stop before applying strict-review graph changes unless the human explicitly accepts a D* delta.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Use when a project-local paper dossier should produce graph delta JSON proposals and run through the human-gated delta loop.

日本語の概要は準備中です。原文の説明を表示しています。

QZhang2111/Research-Pilot262026年6月9日 更新

Use when a Research Pilot project graph change must be previewed, registered, accepted, rejected, parked, or revised through human-gated D* deltas.

日本語の概要は準備中です。原文の説明を表示しています。

QZhang2111/Research-Pilot262026年6月9日 更新

Use when a graph gap or Evidence Need should become a targeted paper-search contract or candidate lead list.

日本語の概要は準備中です。原文の説明を表示しています。

QZhang2111/Research-Pilot262026年6月9日 更新

Use when the user wants to add source identity from PDF, URL, arXiv, DOI, note, experiment result, manual reference, or Zotero item into a Research Pilot project.

日本語の概要は準備中です。原文の説明を表示しています。

QZhang2111/Research-Pilot262026年6月9日 更新

Use when a Research Pilot project needs an experiment proposal from a weak or bounded project claim.

日本語の概要は準備中です。原文の説明を表示しています。

QZhang2111/Research-Pilot262026年6月9日 更新

Use when a Research Pilot project needs read-only gap detection or Evidence Need extraction from the Project Understanding Graph.

日本語の概要は準備中です。原文の説明を表示しています。

QZhang2111/Research-Pilot262026年6月9日 更新

QZhang2111 のスキルをすべて見る

このスキルの問題を報告する