YCC-Lover/research-starter-skills / skills/research-starter-paper
research-starter-paper
Unverified科研学习与论文写作总入口,协调六个模块。帮助不知道提供什么材料的新手逐步产出、自动保存和恢复项目进度,也用于跨阶段研究任务;区分学习、执行和只读检查。
resources/research-engine/academic-paper/references/domain_evidence_profiles.md
View on GitHub# Domain Evidence Profiles
> **Advisory only.** A domain evidence profile changes which evidence types the literature screening *admits*. It never changes the A-F Overall Grade and never blocks manuscript ship. It never changes which databases are queried (that stays `Discipline`-driven in `literature_strategist_agent` Step 2). Consumed only by `academic-paper`'s `literature_strategist_agent`; produced by `intake_agent` Step 12 as the PCR `Domain Evidence Profile` row.
## Domain Evidence Profiles
The ship-ready enum is exactly four values. `unknown_user_defined` is the default and the neutral fallback.
| Profile | Standard evidence types | Common provenance requirements | Critical gaps to surface | Reserved-note |
|---|---|---|---|---|
| `general_social_science` | Peer-reviewed empirical studies, mixed-methods, expert-panel / context-dependent policy analyses | Journal or proceedings; expert-panel reports acceptable where context-dependent | Single-context generalization; weak external validity | — (ship-ready) |
| `cs_ml` | Peer-reviewed papers AND archival preprints (e.g. arXiv), conference proceedings; industry technical reports | Preprint server or proceedings acceptable; peer-review lags the field | Non-reproducible results; benchmark cherry-picking | — (ship-ready) |
| `humanities_interpretive` | Primary sources, archival material, canonical/older texts, monographs | Primary-source provenance; recency is not a quality signal | Interpretive over-reach; missing primary-source grounding | — (ship-ready) |
| `unknown_user_defined` | Neutral single-pyramid (pre-#259 behavior) — peer-reviewed default | Standard peer-review expectation | (no profile-specific loosening) | Default / neutral fallback |
**Reserved profiles (documented, NOT in the enum).** Selecting one records effective `unknown_user_defined` and surfaces a reserved-fallback advisory — its checklist does not exist yet, so it falls back to neutral to prevent false rigor:
`clinical` · `wet_lab` · `materials_physics` · `legal_case_based` · `education` — *not in enum yet; selecting this falls back to `unknown_user_defined`.*
## Field-guidance carry-forward (seeded from the deep-research evidence hierarchy)
The substance of the field-centric `## Field-Specific Adjustments` table in `deep-research/references/source_quality_hierarchy.md` is carried forward here so no per-field guidance is silently dropped. The deep-research file is **read, not edited** — this is a one-time authoring copy, not a runtime dual-read.
Normative (folded into a ship-ready profile row above):
- **Social Science** → folded into `general_social_science` (Level III-V; mixed methods common).
- **Technology** → folded into `cs_ml` (Level III + industry reports; peer review lags reality).
- **Humanities** → folded into `humanities_interpretive` (Level VI primary sources; different epistemology — "evidence" means different things).
- **Policy** → folded into `general_social_science` (Level IV-V + VII expert panels; context-dependent, expert opinion valued). There is no dedicated Policy profile.
**Historical reference — non-normative; current behavior for these domains is neutral `unknown_user_defined` until the `clinical` / `education` profile ships:**
- **Medicine/Health** — Level I-II (RCTs, meta-analyses); Level I-III common; evidence-based-medicine tradition. *(maps to reserved `clinical`)*
- **Education** — Level III-IV (quasi-experimental); Level IV-VI common; randomization often impractical. *(maps to reserved `education`)*
These two rows are preserved verbatim for the eventual `clinical` / `education` profiles. They do NOT change runtime behavior today: until those reserved profiles ship, Medicine/Health and Education runs use the neutral `unknown_user_defined` pyramid, exactly as every other unmapped selection does.
## #246 forward reference
Discipline-relative *grade aggregation* (how these evidence expectations roll up into an Overall Grade) is tracked separately in #246 and is **not yet implemented**. Until #246 ships, the A-F Overall Grade lookup in `deep-research/references/source_quality_hierarchy.md` applies unchanged. #259 ships no aggregation logic and no placeholder aggregation code.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |