Command
Apply cross-national survey experiment design expertise to the task below. Cover per-country power, measurement equivalence, sensitivity bias auditing, instrument localization, and multi-country estimation as relevant.
Command
Delegate creative divergence to Codex (GPT-5.6 "Sol" at xhigh effort). Compose a brainstorm prompt from the task below and run it through codex exec --model gpt-5.6-sol -c modelreasoningeffort=xhigh (read-only sandbox, stdin closed with < /dev/null); Codex returns 3–5 conceptually distinct approaches labeled by…
Command
Before implementing anything, generate 3–5 conceptually distinct approaches to the task below — each labeled by creativity dimension ([Novel], [Surprising], [Diverse], [Conventional]), with at least one [Surprising] and one [Novel]. Differences must be in underlying mechanism, not surface vocabulary. For each, give a…
Command
Read or convert the document below. Decide whether to read it directly or convert it, route to the right converter based on the document's actual structure, guard against each converter's known failure modes, and decide whether the resulting Markdown is a tracked artifact or a scratch file to delete.
Command
Act as the multi-model orchestrator, with Fable 5 (reasoning /effort max) as tech lead — the strongest model on the team, so the hard reasoning stays with you rather than being offloaded. Plan and decompose the task, then show the plan first — state your decomposition and which executor each piece routes to — before…
Command
Fact-check the claims in the manuscript below against their cited sources. First run the pre-flight gate: if there is no per-source Markdown knowledge base, or raw PDFs/docs are still unconverted, refuse to run and return a Pre-flight: NOT READY checklist that tells the user to build the per-source Markdown knowledge…
Command
Audit the completed manuscript and any replication package against FAIR open-science principles: Findable, Accessible, Interoperable, and Reusable. Check data, code, materials, prompts, preregistrations, repository links, persistent identifiers, metadata, licenses, access restrictions, and availability statements.…
Command
Audit figures, tables, captions, cross-references, statistical notes, accessibility, SI links, and text-to-table consistency for the task below. Flag visual-only claims that require author verification rather than inferring exact values from plots.
Command
Design or revise a publication-quality figure for the task below. State the comparison first, then choose chart type, scales, color, and legend ordering. Match legend order to the visual order in the chart, prefer direct labeling over legends when possible, and write a self-contained caption. Generate from code; defer…
Command
Apply causal hypothesis architecture expertise to the task below. Cover falsifiability, counterfactuals, DAGs, FPCI, three-level hypothesis specification, equivalence testing, and SESOI as relevant.
Command
Draft a senior-peer referee report on someone else's manuscript for a social-science journal. Five parallel finder sub-agents (Breaker, Butcher, Shredder, Void, Situator) plus a Blue Team filter, Chief Reviewer synthesis, and Tone Guard sanitization produce a tight 1,200–2,000 word report with Recommendation + Summary…
Command
Apply list experiment (item count technique) design, estimation, and diagnostic expertise to the task below. Cover pre-design sensitivity assessment, control list design, estimator choice, assumption testing, and common failure modes as relevant.