Standard operating procedure for research survey and project execution. Covers deep reading methodology, information synthesis, paper evaluation, citation integrity, cross-module coordination (Library↔Tasks↔Workspace↔Monitor), and research project lifecycle (Phase 2-4).
Covers Phase 2-4 of the research workflow (after papers are found). Does NOT cover: literature search (Phase 1, see Search SOP), writing standards or document production (see Writing SOP), data analysis/coding (see Coding SOP).
Systematic reading extracts structured knowledge from each paper.
library_search.
Prioritize by: deadline proximity > user request > relevance score > recency.library_update_paper(id, read_status: "reading").workspace_save("outputs/notes/{paper-short-title}-notes.md", content)library_update_paper(id, read_status: "read", notes: "...")When reading multiple papers on the same topic, build a comparison matrix
(method, dataset, results, limitations). Note contradictions and consensus.
Save via workspace_save("outputs/notes/comparison-{topic}.md", ...)
Transform extracted knowledge into coherent research outputs.
workspace_save("outputs/drafts/{topic}-{type}.md", content)
Types: literature-review, summary, analysis, reportlibrary_export_bibtex(paper_ids: [...]) or
library_export_bibtex(paper_ids: [...], format: 'ris'). Save to:
workspace_save("outputs/exports/bibliography-{project}.bib", bibtex)Before presenting a synthesis to the user:
Structure research work into trackable tasks with deadlines.
task_create → task_link (papers/files) → task_note (progress)
→ task_complete (with summary)
task_create params: title (action statement), task_type ("human" |
"agent" | "mixed"), priority ("urgent" | "high" | "medium" | "low"),
deadline (ISO 8601).
When a task produces output: (1) workspace_save the file, (2)
task_link_file(task_id, file_path), (3) task_complete(id, notes: "Output: ...").
For papers: task_link(task_id, paper_id).
Use task_note(task_id, note) for progress updates. Set status: "blocked" with
blocker note when stuck. On completion, emit a progress_card.
Evaluate five dimensions: Venue quality (reputable, peer reviewed?), Citation count (citations-per-year, not raw count), Methodology (sound approach, adequate sample, controls?), Reproducibility (sufficient detail?), Relevance (addresses user's question?).
Rate: high / medium / low. Only add high + medium to library.
Full citation rules (4 non-negotiable rules) are in the Writing SOP skill.
Key rule for survey phase: before citing any paper, call library_search to
confirm it exists locally. If not found, add it first (Cross-Module Rule 6).
Six handoff rules keep data consistent across modules:
paper_card per finding. User selects
which to add -> library_add_paper. Emit monitor_digest. Save digest to
outputs/monitor/monitor-scan-{date}.md.monitor_create, suggest cron scheduling.
Each tick: monitor_get_context -> scan -> monitor_report -> monitor_note.task_link(task_id, paper_id).
Reversible, no confirmation needed.progress_card with papers read,
added, tasks completed, highlights.progress_card + suggest Phase 2.
After Phase 2 -> summarize findings + suggest Phase 3.library_search to confirm paper exists.
If missing, library_add_paper first. Never cite unverified papers.All research outputs MUST be persisted to workspace -- never only in chat.
outputs/drafts/literature-review-{topic}.md -- literature reviewsoutputs/notes/{paper-short-title}-notes.md -- reading notesoutputs/notes/comparison-{topic}.md -- comparison matricesoutputs/reports/analysis-{dataset}.md -- analysis reportsoutputs/monitor/monitor-scan-{date}.md -- monitor digestsoutputs/reports/weekly-report-{date}.md -- progress reportsoutputs/exports/bibliography-{project}.bib -- BibTeX exportsWhen resuming work, check: workspace_history (recent changes),
workspace_list (current files), task_list (active tasks).
Combine with MEMORY.md to reconstruct where the user left off.
Key tool call chains for survey and project execution:
library_search -> library_update_paper(read_status:"reading")
-> [read & extract] -> workspace_save(notes)
-> library_update_paper(read_status:"read", notes)
library_search(tags/collection) -> [organize by theme]
-> workspace_save(draft) -> library_export_bibtex(paper_ids)
-> workspace_save(bibliography)
task_create(title, deadline, priority) -> task_link(paper_id)
-> task_note(progress) -> workspace_save(output)
-> task_link_file(file_path) -> task_complete(notes)
monitor_create -> cron(schedule) -> monitor_get_context
-> [scan] -> monitor_report -> monitor_note
-> library_add_paper (user-selected) -> task_link
-> workspace_save(digest)
For methodology and domain guidance beyond this SOP, browse these RP subcategories:
research/methodology/ -- research design, experimental design, grounded theory, mixed methodsresearch/deep-research/ -- systematic review (PRISMA), scoping review, meta-synthesisresearch/paper-review/ -- paper summarization, peer review, quality assessmentresearch/funding/ -- grant writing (NSF/NIH), data sharing plansresearch/automation/ -- end-to-end research pipelines, automated workflowsdomains/ -- 16 discipline-specific subcategories (biomedical, cs, economics,
law, humanities, etc.). Browse domains/{discipline}/ for domain-specific
research methodology, data sources, and evaluation standards