Initial Feynman research agent scaffold
This commit is contained in:
49
skills/research/reading-list/SKILL.md
Normal file
49
skills/research/reading-list/SKILL.md
Normal file
@@ -0,0 +1,49 @@
|
||||
---
|
||||
name: reading-list
|
||||
description: Use this when the user wants a curated reading sequence, paper shortlist, or tiered set of papers for learning or project onboarding.
|
||||
---
|
||||
|
||||
# Reading List
|
||||
|
||||
## When To Use
|
||||
|
||||
Use this skill for:
|
||||
- getting up to speed on a topic
|
||||
- onboarding into a research area
|
||||
- choosing which papers to read first
|
||||
- constructing a project-specific reading order
|
||||
|
||||
## Procedure
|
||||
|
||||
1. Start with `alpha_search` in `all` mode.
|
||||
2. Inspect the strongest candidates with `alpha_get_paper`.
|
||||
3. Use `alpha_ask_paper` for fit questions like:
|
||||
- what problem does this really solve
|
||||
- what assumptions does it rely on
|
||||
- what prior work does it build on
|
||||
4. Classify papers into roles:
|
||||
- foundational
|
||||
- key recent advances
|
||||
- evaluation or benchmark references
|
||||
- critiques or limitations
|
||||
- likely replication targets
|
||||
5. Order the list intentionally:
|
||||
- start with orientation
|
||||
- move to strongest methods
|
||||
- finish with edges, critiques, or adjacent work
|
||||
6. Write the final list as a durable markdown artifact in `outputs/`.
|
||||
|
||||
## Pitfalls
|
||||
|
||||
- Do not sort purely by citations.
|
||||
- Do not over-index on recency when fundamentals matter.
|
||||
- Do not include papers you have not inspected at all.
|
||||
|
||||
## Deliverable
|
||||
|
||||
For each paper include:
|
||||
- title
|
||||
- year
|
||||
- why it matters
|
||||
- when to read it in the sequence
|
||||
- one caveat or limitation
|
||||
Reference in New Issue
Block a user