Paper Search Pro

---
name: paper-search-pro
description: "Find academic papers across up to 7 sources (OpenAlex / Semantic Scholar / CrossRef / PubMed / arXiv for English, plus native-Chinese retrieval via NSSD 国家哲社文献中心 + yiigle 中华医学期刊) with adjustable depth — Quick scan (5 min) to Audit prep (3 hr). Use when the user wants to find papers, run a literature search, gather references, scope a research topic, search Chinese-language / 中文原生 literature (中文文献/中文核心/CSSCI/C刊/国内研究/国内文献/中华××期刊/心理学报/经济研究), or filter results by journal tier (中科院分区/一区/几区, Q1, JCR/SJR quartile, 影响因子/impact factor, 期刊分区, 顶刊/top journal, '按分区筛'). Triggers on search verbs ('find papers', 'literature search', 'papers about X'), review types ('scoping review', 'systematic review', 'SR prep', 'literature review', 'lit review', 'help me write a lit review'), Chinese ('找文献', '找论文', '论文搜索', '学术检索', '文献检索', '文献综述', '综述前期', '求文献', '中文文献', '中文核心', 'CSSCI', 'C刊', '国内研究', '找中文的'). Outputs Shadcn HTML report + BibTeX/RIS/CSV + PRISMA-S log. Do NOT use for: concept explanations ('what is X' / 'X 是什么', e.g. '影响因子怎么算'), writing ('帮我写' / 'help me write a paragraph'), single-paper interpretation or PDF download with metadata (use paper-downloader-portable), or when the user already has a literature set (use literature-set-review)."
license: Apache-2.0
allowed-tools: Bash, Read, Write, Edit, Glob, Grep, Task
metadata:
  author: Bo
  version: 2.3.0
  vendored-from: futurehouse/paper-qa (Apache 2.0)
---

# paper-search-pro

Multi-source literature search with adjustable depth. Four tiers, five data sources orchestrated by you (the main agent). Python helpers handle deterministic work; LLM classification is delegated to parallel Inline SubAgents — no external API key required.

## When to use this skill

- User wants to find academic papers / 找文献 / 论文搜索
- User is preparing a literature review, systematic review (SR), scoping review, or meta-analysis
- User wants to scope research on a topic for a thesis / proposal / coursework / news story
- User asks "what research exists on X" / "find me papers about Y"
- User uploads a query that suggests literature gathering (PICO, SPIDER, MeSH, RCT, etc.)

## When NOT to use

- User wants to **read** a specific paper (use PDF reader / download tool)
- User wants to **summarize** a single known paper (use a summarizer)
- User wants to **download** PDFs given DOIs (use `paper-downloader-portable`)
- User already has a literature set and wants to write a review (use `literature-set-review` / `factor-outcome-review`)
- User wants concept explanation, not papers ("what is prospect theory" → just answer)

---

## 🤖 Called by another agent / headless mode

**If you are an agent driving this Skill for your own reasoning (not for a human
who wants an HTML report)**, do NOT hand-run the 14-STEP recipe below. There is a
single structured-data channel built for you:

```bash
PYTHONPATH=$PSP_HOME python3 -m scripts.agent_search "<query>" > result.json
```

One command runs the whole deterministic core — multi-strategy retrieve → dedup →
heuristic relevance score (computed for every paper) → saturation signal → quota
snapshot → per-paper journal metric — and prints **one JSON envelope** (no HTML,
no PRISMA, no LLM classification SubAgent). The human path below is unaffected.
That command gives you a deterministic **floor**, not the finished job — agent mode
is *not* meant to stop at the machine output; `references/agent_mode.md` is where you
layer your own semantic judgement on top to reach human-recipe quality (the command
guarantees the floor; you supply the quality).

📖 **Read `references/agent_mode.md`** for the full envelope schema, every flag
(`--verify`, `--min-relevance`, `--quartile`, `--min-impact`, …), the relevance
formula, error codes / exit codes, and source selection. This is the SSOT for
agent callers — everything else in this section is just the pointer to it.

Everything from here down is the **human-facing 14-STEP recipe** (HTML report +
exports). Use it when the consumer is a person.

---

## 🔥 Execution discipline (read before running anything)

Four invariants govern every step — ignoring them is the dominant failure mode in real sessions:

- **A — NEVER `cd` into the Skill directory.** `cd $PSP_HOME` rebinds `./` to the Skill asset folder, so `./paper-search-results/...` lands inside the Skill instead of the user's workspace (and a re-install wipes it). Run every helper from the user's PWD: `PYTHONPATH=$PSP_HOME python3 -m scripts.<name> … > "$SEARCH_DIR/..."`. `$PSP_HOME` (STEP 0) is the install dir; `$SEARCH_DIR` (STEP 0) is an absolute path under the user's PWD.
- **B — Dispatch classifier SubAgents in parallel.** STEP 6 puts up to 5 `Task` blocks in one assistant message; serial dispatch inflates Standard tier from ~10 to ~17 min. The worked example lives in STEP 6 — it is not repeated elsewhere.
- **C — Announce every skip.** If you skip a STEP (budget / empty data / user choice), say *what* you skipped, *why*, *what's lost*, and *how to recover* (e.g. "re-run at `--tier deep`"). Skipping is fine; surprising the user is not.
- **D — Read a step's cited reference when that step is non-trivial for this case.** Each STEP names a `references/<file>.md` carrying edge cases not duplicated here. You won't read all of them every run, nor should you — but skipping the reference for a step you are *actually about to run* is where boundary knowledge (dict-vs-list shapes, enrich-not-search, DOI casing) gets lost. Read the one in front of you.

---

## Architecture at a glance

```
You (main agent) drive the workflow per this SKILL.md.
Python helpers do deterministic work — NO LLM inside, NO external API key.

  L1 OpenAlex (primary)  → deep top-100 multi-strategy
  L2 PubMed (medical)    → MeSH enricher (mostly; Audit-tier can search independently)
  L2 arXiv (CS/preprint) → T-0~T-4 freshness sentinel
  L3 Semantic Scholar    → influentialCitationCount + abstract fallback
  L3 CrossRef            → funder / license / clinical-trial-number

  Classification         → Inline SubAgents (parallel, file-IPC, 5 per message)
  Output                 → HTML (Shadcn) + MD + BibTeX/RIS/CSV + PRISMA-S log
```

---

## The 4 tiers — pick first

| Tier | Wall-clock | Papers | When to pick |
|------|------------|--------|--------------|
| Quick | ~5-8 min | 20-60 | "查一下" / "几篇" / "before tomorrow" / fast scope |
| **Standard** (default) | ~10-17 min | 60-180 | Scope a topic / write background / general lit search |
| Deep | ~30-45 min | 180-400 | "thorough" / writing a review article / 综述写作 |
| Audit | ~2-3 hr | 400-1000+ | "systematic review" / "PRISMA" / "Cochrane" / "meta-analysis" |

📖 **BEFORE picking, read `references/tier_decision.md`.** Tell the user your choice and why. For Audit, show limitations warning + get explicit confirmation before starting.

---

## The recipe

For every literature search, follow these steps in order. Each step references a `references/` file for details. Skip files only when the step is obviously trivial for the case at hand — and announce the skip per Rule C.

### STEP 0 — Setup ($PSP_HOME + working directory)

📖 BEFORE THIS STEP, read: `references/setup.md`.

**Resolve the Skill install path into `$PSP_HOME`** (every later step uses `PYTHONPATH=$PSP_HOME`). Prefer explicit injection / agent env var; otherwise scan the known cross-agent install locations. If your harness already exposes this SKILL.md's absolute path, just `export PSP_HOME="<that dir>"` and skip the scan. 📖 Full rationale, why this can't be a script, and the complete path list: `references/runtime_bootstrap.md`.

```bash
PSP_HOME="${PSP_HOME:-${CLAUDE_SKILL_DIR:-${CODEBUDDY_SKILL_DIR:-}}}"   # explicit / agent-injected
if [ -z "$PSP_HOME" ]; then                                            # else scan known installs
  for base in "$HOME/.claude" "$HOME/.codex" "$HOME/.agents" "$HOME/.config/opencode" \
              "$HOME/.codeium/windsurf" "$HOME/.config/goose" "$HOME/.cline" "$HOME/.roo" \
              "$HOME/.copilot" ./.claude ./.codex ./.agents ./.cursor ./.opencode ./.windsurf; do
    [ -f "$base/skills/paper-search-pro/SKILL.md" ] && PSP_HOME="$base/skills/paper-search-pro" && break
  done
fi
[ -z "$PSP_HOME" ] && { echo "ERROR: paper-search-pro install not found. Set PSP_HOME to the dir containing SKILL.md."; exit 1; }
export PSP_HOME; echo "Using Skill install: $PSP_HOME"
```

**Verify config keys** (executed from any cwd, never `cd` into the Skill dir):

```bash
PYTHONPATH=$PSP_HOME python3 -c \
  "from scripts.config import load_config; c = load_config(); print('OK' if c.openalex_api_key and c.ncbi_email else 'MISSING — see references/setup.md')"
```

If "MISSING", point the user to `references/setup.md` (5 keys, all free, ~15 min total) and halt.

**Set up the working directory variable** — every subsequent step uses `$SEARCH_DIR`:

```bash
SEARCH_ID="<topic_slug>_<tier>_$(date +%Y%m%d_%H%M%S)"   # e.g. clt_education_quick_20260522_103045
SEARCH_DIR="$(pwd)/paper-search-results/$SEARCH_ID"
mkdir -p "$SEARCH_DIR/raw" "$SEARCH_DIR/batches" "$SEARCH_DIR/classifications"
echo "Outputs will land in: $SEARCH_DIR"
```

`$SEARCH_DIR` is now an **absolute path under the user's PWD**. Use `"$SEARCH_DIR/..."` (quoted, with the variable) in every helper command below — not `./paper-search-results/...`.

### STEP 1 — Plan the query (MANDATORY for all tiers)

📖 BEFORE THIS STEP, read: `references/query_planner.md`.

**Detect the report UI language** — `UI_LANG` (`zh` for Chinese queries, `en` for everything else) selects which UI language the final HTML report renders in. Paper titles / abstracts / authors / venues are NEVER translated — only the report's UI chrome. Pass `--language $UI_LANG` to STEP 12b.

```bash
UI_LANG=$(PYTHONPATH=$PSP_HOME python3 -m scripts.detect_language "$USER_QUERY")
```

The detector routes Japanese / Korean / European queries to **English** (the bundle ships only EN + ZH dictionaries; English is the international academic default). 📖 The exact Unicode rule and why kana is checked before Han live in `references/runtime_bootstrap.md`.

**Determine the search language space** (`search_language`, axis 2 — *which literature ocean*, distinct from `UI_LANG` above which is only *report chrome*). 📖 The parsing SSOT is `references/source_routing.md` §"Language scope"; resolve the space here, before phrasing the query, because it changes how STEP 3's query is built. This is **additive and opt-in — a pure English query resolves to the `en` space with zero new prompts or behavior (R-19)**; everything below fires only for Chinese queries or explicit signals.

- Read the persisted default `config.search_language` (auto | en | zh | both) and apply the priority ladder **flags > in-query markers > config > auto**. CJK presence is the mechanical fact from `detect_language` above; markers (`CSSCI`, `中文文献`, `SSCI`, `知网`, …) and non-signals (`中科院一区`, topic-about-China) are your semantic judgment per the §"Language scope" tables.
- **`auto` + a Chinese (CJK) query + no language marker + no persisted value → ask ONE question before retrieving** (this is the human path's job; the CLI/agent path passes through instead). Two sentences, offer to persist, and don't re-ask later this session:

  > 你用中文提问——文献要英文、中文,还是都要?顺便可以说"以后都这样",我就记成默认、下次不再问。

  - "中文" / "都要" → enter that space (STEP 2 discipline routing takes over; report one line there).
  - "英文" → v2.2 behavior (Chinese topic planned as an English query), report one line.
  - "无所谓 / 都行" → **this run uses `both`** (Recall > Precision), not persisted; if the same user answers "无所谓" a second time, add one light offer to set `both` as default, then never ask again.
  - Only an explicit "以后都…" persists to `config.search_language` (single answers never auto-persist).
- **If a rank ambiguity (bare "Q1") also fired this run, merge both questions into ONE message** — ask language + platform together, never in two rounds (over-asking is a red line).
- Once the space is known, phrase the query per `references/query_planner.md` §"Cross-language query handling": **`zh` keeps Chinese search terms (no translation)**, `en` uses the English terms (v2.2 behavior), `both` builds two sets.

Apply PICO / SPIDER / PEO depending on domain:
- Medical/clinical → PICO (Population/Intervention/Comparator/Outcome)
- Qualitative → SPIDER
- Scoping → PEO (Population/Exposure/Outcome)
- Open-ended → just extract 2-4 concept blocks + 2-5 synonyms each

**Journal-rank intent recognition (additive — only acts when the query mentions a partition).** Before you extract concept blocks, check whether the user's query carries a journal-rank/partition phrase — "中科院一区", "Q1", "JCR Q1", "SJR Q2", "顶刊 / top journal". If so, that phrase is a **filter condition, not a search term**, and it MUST be **stripped from the topic** before retrieval. This roots out the failure that motivated the whole feature: "中科院一区 情绪调节" used to send "中科院一区" to the search engine as a topic word, so it searched for papers *about* 中科院一区 instead of papers *on* 情绪调节 *filtered to* CAS tier 1. The deterministic parser does both jobs (extract + strip) for you:

```bash
PYTHONPATH=$PSP_HOME python3 -c "
from scripts.rank_intent import parse_rank_intent
i = parse_rank_intent('''<original user query>''')
import json; print(json.dumps({
  'platform': i.platform, 'tiers': i.tiers, 'quartiles': i.quartiles,
  'top': i.top, 'ambiguous': i.ambiguous, 'cleaned_query': i.cleaned_query,
  'stripped': i.matched}, ensure_ascii=False))
"
```

Then act on the parse:
- **`cleaned_query`** is the real topic — use it (NOT the raw query) for STEP 3 retrieval and the query plan. When the query had no rank phrasing, `cleaned_query == query` and nothing changes (R-19 default path is untouched).
- **`platform` + `tiers`/`quartiles`/`top`** are the filter you will apply in STEP 10/11 — remember them; do not filter here.
- **`ambiguous == True`** (a bare "Q1"/"Q2" with no platform word — the recogniser never guesses a platform): **ask the user one short question inline** before going further — *"按 JCR 还是 SJR 的 Q1 筛?顺便要不要设为以后的默认?"* The CLI/headless path cannot ask, so this inline question is specifically the human path's job.
- If the query mentions no partition at all, skip this entirely — STEP 1 proceeds exactly as before.

Even Quick tier needs a lightweight version of this step — never skip silently. Output: 1-3 search strategies (concept blocks + year range + work type filter). Write to `"$SEARCH_DIR/query_plan.json"` so PRISMA-S logger can pick it up later (STEP 13).

### STEP 2 — Route supplemental sources within the STEP-1 language space

📖 BEFORE THIS STEP, read: `references/source_routing.md`.

**You make these routing calls by judging the query's domain — the reference's keyword tables are calibration examples, not a match list** (mechanical facts — CJK detection, explicit `--flags` — stay deterministic). Within the language space fixed in STEP 1, route the per-discipline boosters:

- **English space** (`en`, or the English half of `both`) — unchanged from v2.2:
  - Medical signals (RCT, PRISMA, MeSH, clinical, disease names) → enable PubMed
  - CS/preprint signals (preprint, arXiv, NeurIPS, transformer, "最新", 2024+) → enable arXiv
  - Cross-domain (e.g. "AI in radiology") → enable both
  - Pure social science / humanities → OpenAlex only
  - *(Judgment call: a core AI/CS query may also raise the primary engine to Semantic Scholar — see `source_routing.md` §"AI / CS queries → consider Semantic Scholar as primary".)*
- **Chinese space** (`zh`, or the Chinese half of `both`) — route the Chinese boosters the same way, by discipline:
  - Social-science / humanities signal → add **NSSD** (国家哲社文献中心; carries the CSSCI 收录标识 OpenAlex has ≈0 coverage of)
  - Medical signal → add **yiigle** (中华医学期刊全文数据库); PubMed still covers MEDLINE-indexed 中华 journals, so the two are complementary
  - Pure sci-tech with neither → Chinese side runs on OpenAlex only (sci-tech Chinese core journals mostly register DOIs, so OpenAlex covers them well)

**Report one line** (axis-3 style — a statement, not a question; 22 §6.3). For an English-only run this is the existing PubMed/arXiv notice, unchanged (*"I detected medical + CS signals — also searching PubMed and arXiv. Override with `--no-pubmed`."*). For a Chinese space, e.g.:

> 本次按「中英都要」检索;中文侧检测到社科主题,已加 NSSD(国家哲社文献中心)。想去掉说 `--no-nssd`,只查一边说"只要英文/中文"。

**Coverage honesty rides with the notice:** if the zh space has a social-science topic but the user declined NSSD, add that OpenAlex hits ≈0 on CSSCI flagship journals (经济研究 / 管理世界 …), so that layer is missing. User can override any of this with an explicit instruction (a per-query override wins over everything).

**On the `--flag` shorthands above (`--no-nssd`, `--no-pubmed`, `--source …`):** on this human path they are **natural-language override *notation*** — a compact way to write what the user can *say* ("去掉 NSSD" / "只查 OpenAlex"), which you (the LLM) interpret. They are **not executable CLI flags** — no script parses them here. The only real, script-parsed flags live on the agent/headless path (`agent_search`), and there the Chinese-source control is opt-**in**: `--with-nssd` / `--with-yiigle` (there is no `--no-nssd` / `--source` there). See `references/agent_mode.md`.

### STEP 3 — Retrieve from OpenAlex (deep)

📖 BEFORE THIS STEP, read: `references/openalex_helper_cheatsheet.md`.

Always run OpenAlex first. (OpenAlex is the default primary source; only if `primary_source` is set in `config.yaml` or the OpenAlex quota is exhausted, see *Primary source selection & quota fallback* in `references/source_routing.md` for the additive SS-fallback flow — default behavior is unchanged.) For Standard+ tiers, use multi-strategy deep crawl:

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.openalex_helper double-sort "<query>" \
    --n 50 --year-min 2018 \
    > "$SEARCH_DIR/raw/openalex.json"
```

For Quick tier, single-strategy is fine:

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.openalex_helper search "<query>" \
    --limit 30 --year-min 2018 \
    > "$SEARCH_DIR/raw/openalex.json"
```

The full subcommand + flag reference (`search` / `double-sort` / `seminal` / `reviews` / `journal-list` / `citation-network`, all verified against argparse) is in `references/openalex_helper_cheatsheet.md` — read it before reaching for anything beyond the two commands above. For Deep+Audit, also call topic-specific subcommands (e.g. `seminal`, `reviews`, `journal-list`), append outputs to `$SEARCH_DIR/raw/openalex_*.json`, and federate them all together in STEP 5.

### STEP 4 — Run L2 boosters (if enabled by STEP 2)

📖 BEFORE THIS STEP, read: `references/pubmed_helper_cheatsheet.md` and `references/arxiv_helper_cheatsheet.md`.

**PubMed — default mode is `enrich`, NOT `search`**:

- **Standard / Deep tier**: enrich OA-found papers with MeSH terms (mutates the openalex.json file in place):
  ```bash
  PYTHONPATH=$PSP_HOME \
    python3 -m scripts.pubmed_helper enrich \
      --input-file "$SEARCH_DIR/raw/openalex.json" \
      --output-file "$SEARCH_DIR/raw/openalex.json"
  ```
- **Audit tier with explicit MeSH query**: independent MeSH search (produces a new file to federate later):
  ```bash
  PYTHONPATH=$PSP_HOME \
    python3 -m scripts.pubmed_helper search-mesh "Diabetes Mellitus, Type 2" \
      --year-min 2020 --limit 30 --pub-type "Randomized Controlled Trial" \
      > "$SEARCH_DIR/raw/pubmed.json"
  ```
- Generic `pubmed_helper search` is a fallback when no MeSH term is known — prefer `enrich` or `search-mesh` whenever possible.

**arXiv — only if query contains freshness signals (preprint, 最新, 2024+):**

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.arxiv_helper freshness "<query>" \
    --days 4 --limit 30 \
    > "$SEARCH_DIR/raw/arxiv.json"
```

Subcommand reference:
- `arxiv_helper freshness <query> --days N --limit M [--all-cats]`
- `arxiv_helper search <query> --limit M --sort submitted|relevance|lastUpdated [--all-cats]`
- `arxiv_helper get <arxiv_id>`

**NSSD / yiigle — Chinese boosters (ONLY when STEP 1-2 put this run in the `zh` space, and only the one(s) the discipline routing selected):**

Each is an independent primary source for the Chinese space (same role as `ss_helper --search`) and emits the **same `UnifiedPaperEntity` shape** as openalex.json, so STEP 5 federates them identically. Keep the query in **Chinese** (do NOT translate — query_planner §Cross-language) and write to `raw/nssd.json` / `raw/yiigle.json`:

```bash
# NSSD — Chinese social-sciences & humanities (adds the CSSCI-tier layer OpenAlex lacks)
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.nssd_helper --search "<中文检索式>" \
    --n 50 --year-min 2018 \
    > "$SEARCH_DIR/raw/nssd.json"

# yiigle — Chinese medical (中华医学期刊全文数据库; native-Chinese titles + abstracts)
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.yiigle_helper --search "<中文检索式>" \
    --n 50 --year-min 2018 \
    > "$SEARCH_DIR/raw/yiigle.json"
```

Both degrade gracefully to `[]` on any network / HTTP failure (they never raise) — an empty file just federates to nothing. Both take **only** 题录 + 摘要 (compliance: no full-text download, no caching) and print their source attribution to stderr. `--year-min` filters client-side; drop it to keep all years.

**config `search_language: en` but a Chinese query arrived (hard boundary 2):** do NOT enable Chinese boosters, but say one line — never a silent translation (22 §6.4):

> 按你的默认(只查英文),我把中文主题规划成英文检索式了。想要中文文献这次说一声即可,想改默认说"以后…"。

**User names 知网 / CNKI / 万方 / 维普 (marker hit + compliance):** these are closed subscription databases PSP does not scrape. Say one line, offer the substitute, don't re-argue (22 §6.5):

> PSP 不接知网/万方(合规原因,不做封闭库抓取)。中文侧用 OpenAlex 中文底座 + NSSD(社科,含 CSSCI 标识)/yiigle(医学)覆盖;如需知网全文,结果里的题录可去知网人工检索。继续吗?

### STEP 5 — Federate (dedup + merge)

📖 BEFORE THIS STEP, read: `references/source_routing.md` §"Field priority table".

Combine all retrieval results into a single deduped KG. **Default output is a dict keyed by canonical_key** — that's what `rcs_parser` expects later, so do NOT pass `--as-list`:

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.federated_kg_resolver \
    --input-files "$SEARCH_DIR/raw/openalex.json" \
                  "$SEARCH_DIR/raw/pubmed.json" \
                  "$SEARCH_DIR/raw/arxiv.json" \
    --output "$SEARCH_DIR/kg.json"
```

Pass only the input files you actually produced — skip ones that were not enabled by STEP 2. **If STEP 4 ran the Chinese boosters, add `"$SEARCH_DIR/raw/nssd.json"` / `"$SEARCH_DIR/raw/yiigle.json"` to the same `--input-files` list** — they carry the identical entity shape and federate exactly like the others (CJK-safe dedup is handled by the Phase 0 canonical-key fix, so distinct Chinese titles don't collapse). For an English-only run those files don't exist, so the call is byte-identical to v2.2 (R-19). This handles DOI normalization (arXiv X→x case), version stripping, E5b guard (same title+year but different DOIs are kept separate), and field-priority merge.

`--as-list` exists but is only for consumers that want a sorted list (by citation_count); do not use it in this pipeline.

### STEP 6 — Classify in parallel batches (LLM happens here — main agent + SubAgents)

📖 BEFORE THIS STEP, read: `references/classifier_subagent_prompt.md` and `references/rcs_rubric.md`.

Split the KG into batches of 10 papers each. Write to `"$SEARCH_DIR/batches/batch_NNN.jsonl"`.

**Before dispatch**, expand `$PSP_HOME/references/rcs_rubric.md` into the actual absolute path (e.g. `/Users/alice/.claude/skills/paper-search-pro/references/rcs_rubric.md`) and substitute it for `{rubric_path}` in the classifier prompt template. Each SubAgent runs in its own shell where `$PSP_HOME` is **not** exported — passing the literal `$PSP_HOME` token would leave the SubAgent unable to find the rubric, which silently degrades scoring quality. See `references/classifier_subagent_prompt.md` for the full placeholder table.

🔥 **PARALLELISM IS MANDATORY** (Rule B):

You MUST dispatch up to **5 classifier SubAgents in a single assistant message** using multiple `Task` tool_use blocks. Serial dispatch (one Task per message, waiting for each result) is the single biggest performance failure observed — it inflates Standard tier from ~10 min to ~17 min.

✅ **CORRECT — in ONE assistant message:**

```
Task tool_use #1  → subagent_type="general-purpose", prompt="<classifier prompt for batch_001.jsonl>"
Task tool_use #2  → subagent_type="general-purpose", prompt="<classifier prompt for batch_002.jsonl>"
Task tool_use #3  → subagent_type="general-purpose", prompt="<classifier prompt for batch_003.jsonl>"
Task tool_use #4  → subagent_type="general-purpose", prompt="<classifier prompt for batch_004.jsonl>"
Task tool_use #5  → subagent_type="general-purpose", prompt="<classifier prompt for batch_005.jsonl>"
```

All five tool_use blocks live in the same `<assistant>` message. The harness fires them in parallel; you receive five tool_result blocks back together.

❌ **WRONG — five separate messages (this is what serial dispatch looks like):**

```
Message N:    Task tool_use #1 ─→ wait for result
Message N+1:  Task tool_use #2 ─→ wait for result   ← SERIAL, makes Standard run 70% slower
Message N+2:  Task tool_use #3 ─→ wait for result
...
```

If you have more than 5 batches, send 5-at-a-time across multiple messages — each message still contains 5 parallel Task blocks.

Each SubAgent reads its batch file, applies the RCS rubric, and writes `"$SEARCH_DIR/classifications/batch_NNN_result.json"`. Then merge classifications into the KG:

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.rcs_parser \
    --input-dir "$SEARCH_DIR/classifications/" \
    --kg "$SEARCH_DIR/kg.json" \
    --output "$SEARCH_DIR/kg_classified.json"
```

### STEP 7 — Compute saturation curve (MANDATORY for all tiers)

📖 BEFORE THIS STEP, read: `references/stop_decision.md`.

This step is NOT optional, even for Quick. The curve.json drives both STEP 8 stop decision and STEP 12 HTML chart rendering. If you skip it, the report shows an empty curve and PRISMA-S transparency suffers.

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.discovery_curve \
    --kg "$SEARCH_DIR/kg_classified.json" \
    --output "$SEARCH_DIR/curve.json"
```

The curve has `saturation_estimate` (0-1) + `ci_low` + `ci_high`. Optional `--prior-snapshots` lets you chain curves across iterations; `--papers-evaluated` overrides the auto-count.

### STEP 8 — Decide next action (MANDATORY)

📖 BEFORE THIS STEP, read: `references/stop_decision.md`.

This step is NOT optional. Make the decision **explicitly** — based on curve.json + tier budget + intent — and state the reasoning to the user. Do not skip based on intuition.

Decision tree:
- saturation < 0.6 AND budget remaining AND tier in {standard, deep, audit} → expand citations (STEP 9)
- saturation > 0.85 OR budget exhausted → stop, write report (STEP 10+)
- ambiguous → tell user the numbers and ask

### STEP 9 — Expand citations (if applicable)

📖 BEFORE THIS STEP, read: `references/citation_chasing.md`.

For top-rcs papers (rcs >= 7), get the citation network:

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.openalex_helper citation-network <openalex_id> \
    --refs-limit 25 --cited-by-limit 25 \
    >> "$SEARCH_DIR/raw/citations.json"
```

Then loop back to STEP 5 (federate the new papers into the KG, then re-classify only the new entries in STEP 6).

### STEP 10 — Enrich top-N papers (L3, optional but recommended)

📖 BEFORE THIS STEP, read: `references/ss_helper_cheatsheet.md` and `references/crossref_helper_cheatsheet.md`.

For papers with rcs >= 6, enrich with SS (influentialCitationCount + abstract fallback + tldr) and CrossRef (funder/license/clinical-trial-number). Both helpers consume a JSON **list** — the KG is currently dict-shaped. Convert first, enrich, then federate back; or supply a paper_list.json produced by data_materialization in STEP 12.

For Quick tier, skipping STEP 10 is acceptable — but **announce the skip** per Rule C ("Skipped L3 enrichment → no influentialCitationCount or funder fields; re-run at `--tier standard` to include this").

```bash
# Semantic Scholar — adds influentialCitationCount + abstract fallback + tldr
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.ss_helper \
    --input-file "$SEARCH_DIR/paper_list.json" \
    --mode enrich \
    --output-file "$SEARCH_DIR/paper_list.json"

# CrossRef — adds funder + license + refs + clinical_trial_number in one fetch
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.crossref_helper \
    --input-file "$SEARCH_DIR/paper_list.json" \
    --mode all \
    --output-file "$SEARCH_DIR/paper_list.json"
```

This adds ~135-170s for 100 papers — only do it on top-N, not the full set.

**Optional (additive) — journal partitions (中科院 / JCR / SJR).**
The multi-platform partition layer labels every paper with
**all three** platforms and, when a tier was requested, filters on **one**. Like
everything else in this step it is **opt-in and off by default — skip it and the
report is byte-for-byte unchanged** (R-19). 📖 Read `references/journal_metrics.md`
first (it is the SSOT for sources, the ISSN join, attribution, and R-04 naming).

- **First use needs a one-time fetch** (init-once; data is pulled at runtime into
  `~/.paper-search-pro/ranks/` and **never bundled in the repo**). If you have not
  fetched before, run it once (and tell the user it is a one-time step):
  ```bash
  PYTHONPATH=$PSP_HOME python3 -m scripts.journal_rank fetch          # all three
  # or a single platform: ... journal_rank fetch --platform cas
  PYTHONPATH=$PSP_HOME python3 -m scripts.journal_rank info           # what's cached
  ```
- **Annotate (label all three platforms — do this once per result set):**
  ```bash
  PYTHONPATH=$PSP_HOME python3 -c "
  from scripts import journal_rank, rank_filter
  # ... load your papers as UnifiedPaperEntity list, then:
  lk = journal_rank.load()                         # RankLookup | None (None → graceful degrade)
  n  = rank_filter.annotate_papers(papers, lk)     # fills paper.journal_rank (三家全标)
  "
  ```
  `journal_rank.load()` returns **None** when nothing is cached — then this layer
  silently degrades (no partitions; the OpenAlex open-impact figure from the block
  above is still the influence placeholder) and you tell the user they can
  `journal_rank fetch` to enable partitions.
- **Filter (only when a tier was requested — see STEP 11 for the full flow):** call
  `rank_filter.filter_by_rank(papers, platform, tiers=…, quartiles=…, top=…)`. It
  returns `(kept, filtered_out, no_platform_data)` — the third bucket (journals not
  on the chosen platform) is **reported, never silently dropped**.
- **R-04 naming** is enforced for you in the serialised dict: only JCR exposes an
  `impact_factor` (the real IF); 中科院"区" and SJR quartile are **分区/quartile**.

### STEP 11 — Write the executive summary

📖 BEFORE THIS STEP, read: `references/summary_writer.md`.

Write a ~300-word executive summary in your own words based on the classified papers:
- The field's main consensus
- Key methods / theoretical frameworks
- Notable disagreements or open questions
- Top 3-5 most influential papers (by `influential_citation_count` when available)
- *(Optional)* journal tier of the leading papers, **if** you attached SJR metrics in STEP 10 — phrase as "SJR分区 / 期刊影响力", never "影响因子 / JCR" (R-04). Skip this bullet entirely when no metrics were attached.

Save to `"$SEARCH_DIR/summary.md"`.

**Partition default / ask / filter / report / switch flow (additive — only when partitions are in play).** When you annotated the multi-platform journal_rank in STEP 10, follow this flow; it is entirely opt-in and changes nothing on the default no-partition path (R-19):

- **Factory default standard = JCR.** The persistent default lives in config `rank.default_platform` (out of the box: `jcr`; the user can set it to `cas`/`sjr`). The default platform only **labels** every paper — it does **not** filter unless the user actually asked for a tier.
- **No partition mentioned → do not filter.** Just show all three platforms' labels per paper (STEP 10 annotate already did this) and let the user read / refine. Never invent a tier filter the user didn't ask for.
- **A tier was requested (from STEP 1 intent or the user this round) → filter this once.** Use the STEP 1 parse: `platform` + `tiers`/`quartiles`/`top`. A per-request tier filter is **transient — never auto-persist it** to config. The persistent default is only ever changed when the user explicitly says "以后都用 X".
- **Ambiguous bare "Q1" with no platform and no persistent default → ask one short question** (you should already have asked in STEP 1; if not, ask now): *"按 JCR 还是 SJR?顺带设默认吗?"* Small confirmations are welcome, but do not over-ask.
- **Always report what this run did.** After filtering, tell the user in one line: *"本次按 {platform} 筛(留 N / 滤 M",* plus a light offer: *"可换中科院/JCR/SJR 或换档位、可设为以后的默认。"* Include the per-platform attribution (`journal_rank.ATTRIBUTION[platform]`).
- **Switching standard or tier = RE-FILTER the already-annotated pool, NOT a re-search.** When the user then says "换成中科院二区" or "看看 SJR Q1", do **not** re-run the search. `annotate_papers` already stamped all three platforms onto the same candidate pool, so a switch is a pure in-memory re-filter — call `rank_filter.filter_by_rank(papers, new_platform, tiers=new_tiers, …)` again and it returns instantly. **Only when the re-filter leaves too few survivors** do you go back to STEP 3 and deepen the search (retrieve more, re-annotate, re-filter). This "切换=重筛不重搜" rule is what makes partition exploration cheap.
- **Persisting the default** (only on an explicit "以后都用 X"): set `rank.default_platform` in `~/.paper-search-pro/config.yaml`. Tier档位 is never persisted — only the platform default is.
- **R-04 naming in the summary bullet too:** 中科院"区" and SJR quartile are **分区 / quartile**; only JCR IF(2024) is an **影响因子 / Impact Factor**. The OpenAlex 2yr-mean-citedness figure is "期刊影响力" (open), never a JIF.

### STEP 12 — Render the report

📖 BEFORE THIS STEP, read: `references/output_files.md`.

```bash
# 12a. Materialize data for the renderer (also writes sibling chart_data / paper_list / metadata / prisma_log)
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.data_materialization \
    --kg "$SEARCH_DIR/kg_classified.json" \
    --summary "$SEARCH_DIR/summary.md" \
    --query "<original query>" \
    --tier "<quick|standard|deep|audit>" \
    --search-id "$SEARCH_ID" \
    --snapshots "$SEARCH_DIR/curve.json" \
    --output "$SEARCH_DIR/report_data.json"

# 12b. Render HTML (Shadcn webartifacts — only renderer; no size cap)
#      --language $UI_LANG selects EN vs ZH UI; the bundle ships with both
#      dictionaries inlined, $UI_LANG just picks which one mounts. Resolution
#      order inside the renderer is: explicit --language > metadata.language > en.
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.html_renderer_webartifacts \
    --data "$SEARCH_DIR/report_data.json" \
    --output "$SEARCH_DIR/report.html" \
    --query "<original query>" \
    --language "$UI_LANG"

# 12c. MD report (uses materialized-dir for speed)
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.md_report \
    --materialized-dir "$SEARCH_DIR" \
    --query "<original query>" \
    --tier "<quick|standard|deep|audit>" \
    --output "$SEARCH_DIR/report.md"

# 12d. Exports (BibTeX / RIS / CSV / papers.json — only rcs >= 5 by default)
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.generate_exports \
    --kg "$SEARCH_DIR/kg_classified.json" \
    --output-dir "$SEARCH_DIR/" \
    --min-rcs 5
```

`data_materialization` accepts `--wall-clock-seconds` if you tracked elapsed time yourself; otherwise the helper computes it from session timestamps when available.

### STEP 13 — Write PRISMA-S log

📖 BEFORE THIS STEP, read: `references/prisma_s_checklist.md`.

```bash
PYTHONPATH=$PSP_HOME \
  python3 -m scripts.prisma_s_logger \
    --search-id "$SEARCH_ID" \
    --kg "$SEARCH_DIR/kg_classified.json" \
    --user-query "<original query>" \
    --tier "<quick|standard|deep|audit>" \
    --query-plan "$SEARCH_DIR/query_plan.json" \
    --snapshots "$SEARCH_DIR/curve.json" \
    --output "$SEARCH_DIR/execution_log.json"
```

This captures the 16 PRISMA-S items for transparency / audit.

### STEP 14 — Open the report + report to user

**First**, auto-open the HTML report in the user's default browser (do NOT wait for the user to ask). Platform-aware Bash:

```bash
# macOS — most common dev setup
open "$SEARCH_DIR/report.html"
# Linux fallback — xdg-open "$SEARCH_DIR/report.html"
# Windows fallback — start "" "$SEARCH_DIR/report.html"
```

Use `open` on macOS by default. If it fails (rare — only bare Linux containers), fall through to `xdg-open` then `start`. **Do NOT skip this step** — the user just waited 5-30 minutes for the report; they should see it the moment it's ready.

**Then** tell the user:
- "Opened report in your default browser." (1 line confirmation)
- Where the report is on disk (absolute path: `$SEARCH_DIR/report.html`) — so the user can find it later
- Top findings (3-5 sentences from your executive summary)
- Any caveats — including any steps you skipped per Rule C (e.g. "PubMed wasn't queried because no medical signals were detected", "Skipped STEP 10 L3 enrichment because Quick tier; re-run at standard to include funder/license fields")

---

## Output convention

📖 See `references/output_files.md` for the full directory layout. All paths below are **relative to the user's working directory (PWD)** — never the Skill asset directory.

```
$(pwd)/paper-search-results/<search_id>/
├── report.html              # Main deliverable (Shadcn style)
├── report.md                # Markdown copy
├── papers.csv               # Spreadsheet export
├── papers.bib               # Citation manager import (BibTeX)
├── papers.ris               # Alternative citation format
├── papers.json              # Full structured data
├── kg_classified.json       # Internal KG with RCS scores
├── summary.md               # Your executive summary
├── execution_log.json       # PRISMA-S 16-item log
├── report_data.json         # Renderer bundle
├── chart_data.json          # Sibling: chart series
├── paper_list.json          # Sibling: per-paper list
├── metadata.json            # Sibling: run metadata
├── prisma_log.json          # Sibling: PRISMA log JSON view
├── curve.json               # Saturation snapshot
├── query_plan.json          # STEP 1 output
├── raw/                     # Raw per-source dumps (openalex.json, pubmed.json, arxiv.json, citations.json)
├── batches/                 # batch_NNN.jsonl files
└── classifications/         # batch_NNN_result.json files
```

---

## Error handling

📖 See `references/error_handling.md`. Common cases:

| Error | What to do |
|-------|-----------|
| Config missing keys | Direct user to `references/setup.md`, halt |
| Rate limit (SS 429 / NCBI 429) | Helper auto-retries; if persistent, drop that enricher |
| OpenAlex 404 on DOI | Use title search fallback (helper handles) |
| L2 booster returns 0 papers | Skip silently, note in PRISMA-S log via STEP 13 |
| SubAgent classifier returns invalid JSON | `rcs_parser.py` has 5-layer fallback (regex parse) |
| HTML output size | No size cap or fallback — `html_renderer_webartifacts` always produces the full Shadcn bundle. Typical 250-paper report is ~1.7 MB; pathological 1000+ paper Audit may reach 5-10 MB. All modern browsers handle 10+ MB HTML cleanly. |

---

## References (progressive disclosure — read the one for the step you're on)

You won't read all of these every run, and shouldn't. Read a step's reference when you reach that step and it's non-trivial for the case (Rule D). **core** = read for its step; **cond** = only when its trigger fires.

| File | Load | Read when |
|------|------|---------|
| `tier_decision.md` | core | choosing the tier (before STEP 0) |
| `setup.md` | core | STEP 0 — config + 5-key acquisition |
| `runtime_bootstrap.md` | cond | STEP 0/1 — only if `$PSP_HOME` env injection failed, or you need the full install-path list / language-routing rationale |
| `query_planner.md` | core | STEP 1 — PICO / SPIDER / PEO frameworks |
| `source_routing.md` | core | STEP 1 language scope (§"Language scope" SSOT) + STEP 2 routing + STEP 5 field-priority merge |
| `openalex_helper_cheatsheet.md` | core | STEP 3 + STEP 9 — subcommands, params, gotchas |
| `pubmed_helper_cheatsheet.md` | cond | STEP 4 — only if PubMed enabled |
| `arxiv_helper_cheatsheet.md` | cond | STEP 4 — only if arXiv enabled |
| `classifier_subagent_prompt.md`, `rcs_rubric.md` | core | STEP 6 — SubAgent prompt + RCS 0-10 rubric |
| `stop_decision.md` | core | STEP 7 + STEP 8 |
| `citation_chasing.md` | cond | STEP 9 — only if expanding citations |
| `ss_helper_cheatsheet.md`, `crossref_helper_cheatsheet.md` | cond | STEP 10 — only if enriching top-N |
| `summary_writer.md` | core | STEP 11 |
| `journal_metrics.md` (SSOT) | cond | STEP 1 / STEP 10-11 — only if the user wants journal partitions (中科院 / JCR / SJR) or SJR metrics; ISSN join, attribution, R-04 naming |
| `output_files.md` | core | STEP 12 — output dir layout (PWD-relative) |
| `prisma_s_checklist.md` | core | STEP 13 |
| `agent_mode.md` (SSOT) | cond | only when another agent / headless calls this Skill — `agent_search` envelope + flags |
| `error_handling.md` | cond | any unexpected error |

---

## Examples

### Example 1: Quick scan (5-8 min)

User: "find 5-6 high-impact papers on prospect theory in decision making, classics + a couple recent ones"

You: Pick Quick tier (signals: "5-6", "high-impact", short query). Run STEP 0-2 lightweight. In STEP 3 use `openalex_helper seminal` for classics + `openalex_helper search` for recent (year >= 2020). No L2 boosters in STEP 4 (pure social science). In STEP 6 classify 20-30 papers via 2 parallel SubAgents in one message. STEP 7 + 8 still run (curve renders in the report). Announce skip of STEP 9 + STEP 10 per Rule C. Render report.

### Example 2: Standard ZH (10-17 min)

User: "用 paper-search-pro 帮我找一些关于工作记忆训练干预的文献 老板让我看 我对这块完全不懂 要给老年人群体的最好 谢谢🙏"

You: Pick Standard tier (default; signals: "找一些", "老板让我看"). Detect medical signal ("干预" + "老年") in STEP 2 → enable PubMed enricher. Query plan: PICO (P=elderly, I=working memory training, O=cognitive outcomes). OpenAlex `double-sort` top-100 in STEP 3. PubMed `enrich` of openalex.json in STEP 4. Federate in STEP 5 (dict output). Classify 60-180 papers via 4 batches × 5 SubAgents — **all 5 Tasks in one message** (Rule B). STEP 7 curve, STEP 8 expand if saturation < 0.6. Render report.

### Example 3: Deep × Lit review writing (30-45 min)

User: "I'm writing a proper literature review article on attachment and human-robot interaction in elderly care contexts. Need real depth..."

You: Pick Deep tier ("proper literature review article" + "real depth"). Cross-domain (psychology + CS) in STEP 2 → enable arXiv freshness sentinel. SPIDER plan in STEP 1. OpenAlex `double-sort` top-200 + `reviews` subcommand in STEP 3. Classify 200+ papers via 8 batches in STEP 6 — dispatch 5 parallel Tasks per message, two waves. STEP 9 expand citations 2 hops. STEP 10 enrich top-50 with SS + CrossRef. Render report with PRISMA-S log.

### Example 4: Audit × SR-prep (2-3 hr)

User: "Need help — preparing a systematic review on dietary interventions for IBS in adults. Inclusion criteria: RCTs, adult populations (≥18), low-FODMAP or fiber-based interventions, English-language, published 2010-present."

You: Pick Audit tier ("systematic review" + PICO + IC). **Show limitations warning first** ("This is not a PRISMA replacement — it's SR-prep assist. Cochrane Library + Embase still needed for full SR rigor."). Get user confirmation. STEP 4 use `pubmed_helper search-mesh "Irritable Bowel Syndrome" --pub-type "Randomized Controlled Trial"` for independent MeSH search. STEP 3 also call `openalex_helper journal-list --preset Cochrane`. STEP 10 add CrossRef enrichment for funder + clinical-trial-number. Render with PRISMA flow chart in STEP 12.