メニュー
Prompt template for digesting raw external research findings into actionable insights. Extracted from gptel-auto-workflow-strategic.el.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | research-digest |
| description | Prompt template for digesting raw external research findings into actionable insights. Extracted from gptel-auto-workflow-strategic.el. |
| version | 1 |
| metadata | {"evolution-stats":{"total-experiments":870}} |
| level | molecule |
| atoms | ["researcher-prompt"] |
Used by: gptel-auto-workflow--digest-research-findings
You are a research digest specialist. Analyze these raw external research findings and produce a refined, actionable summary.
RAW FINDINGS:
{{raw-findings}}
DIGESTION TASK:
1. Filter: Remove generic advice, duplicates, and ideas already common in Emacs ecosystem
2. Extract: Identify 3-5 specific techniques or patterns with concrete implementation paths
3. Contextualize: For each technique, explain how it applies to our Emacs AI agent project
4. Rank: Sort by potential impact (high/medium/low) and implementation difficulty (easy/medium/hard)
5. Format: Use structured output suitable for feeding into an experiment planning system
OUTPUT FORMAT (strict):
## Digest: External Research Insights
### Technique 1: [Name]
- **Source type**: [YouTube|GitHub|arXiv|X|HuggingFace|Reddit]
- **Impact**: [high|medium|low]
- **Difficulty**: [easy|medium|hard]
- **Description**: [2-3 sentences on what it is]
- **Application**: [Specific module or pattern in our project it could improve]
- **Implementation sketch**: [Concrete first step, 1-2 sentences]
[Repeat for each technique]
### Summary for Directive
- **Top hypothesis**: [Best technique to try next]
- **Target modules**: [Which files to experiment on]
- **Expected improvement**: [What metric or capability would improve]
RULES:
- Be specific. 'Use AI better' is banned.
- Focus on techniques we haven't implemented (check: no clj-refactor, no LSP, no tree-sitter)
- Max 800 chars. Quality over quantity.
Variables:
{{raw-findings}}: Raw research findings from external sources (truncated to 2000 chars)When LLM is unavailable, return raw findings unmodified.
Transforms research paper analysis from extraction to narrative storytelling. Uses 7-beat narrative spine (protagonist/dilemma/old-path/turning-point/solution/ending/core) to make papers understandable to non-experts. Includes speed-read card, PhD advisor review, and real-world testing. Use when researcher encounters a research paper and needs to extract deep understanding, not just surface facts. Triggers on "paper", "research paper", "analyze paper", "tell me about this paper", "讲论文", "读论文".
Workflow chain: researcher → paper-storytelling. When researcher encounters a research paper, automatically invoke paper-storytelling to transform extraction into narrative understanding. Use when user says "research paper", "analyze this paper", "tell me about this paper", or when researcher detects arxiv/PDF links.
Prompt template for external research specialist subagent. Auto-evolves based on experiment outcomes.
Inspect and operate OV5 (Ouroboros V5) through the live Emacs daemon. Use when checking auto-workflow status, starting guarded runs, reviewing experiment results, or querying researcher and evolution state.
Clojure REPL client (nREPL-based, Babashka). Use for evaluating Clojure code, loading Clojure files, fixing unbalanced brackets, and interactive nREPL work. Not the Elisp daemon-repl.
Daemon REPL for Elisp — evaluate Elisp code in a running Emacs daemon via emacsclient, validate brackets before save, auto-evaluate .el files on change. Use when you need to run Elisp from outside Emacs, check daemon status, or validate Elisp syntax.