ossianravn

discover, filter, sort, and shortlist polymarket markets from the full active market universe. use when the user wants to browse all available bets, find candidates by category, horizon, liquidity, structure, catalyst profile, resolution risk, opportunity type, create discovery-driven watchlists, or run scheduled universe triage before deeper research. do not use for placing, previewing, or submitting trades.

perform deep research for a polymarket market, event, or thesis. use when the user wants fair-value work, evidence gathering, catalyst mapping, linked-market comparisons, or a detailed explanation of whether the market looks mispriced. for wide candidate sets or triage questions, use opportunity-classifier first.

evaluate whether passive polymarket quoting looks reward-eligible or operationally sensible. use when the user wants market-making guidance, passive quote placement, spread analysis, rewards checks, or two-sided quoting ideas.

write a structured memo for a polymarket market or event. use when the user asks to analyze a single market, summarize current odds, liquidity, spread, recent flow, resolution mechanics, or prepare a trade brief before acting. for broad candidate sets or ranking questions, use opportunity-classifier first.

turn a polymarket strategy into a guarded order plan. use when the user wants a limit order, marketable order, preview, cancel plan, or explicit trade parameters. use this skill before any live execution.

review polymarket positions, open orders, and concentration risk. use when the user asks what they own, where they are overexposed, which orders are stale, how correlated their book is, or what should be reduced or canceled. when helpful, reuse the opportunity-classifier taxonomy to bucket exposure by structural type, horizon, and resolution risk.

monitor polymarket markets that are close to resolution or likely to be resolved soon. use when the user wants official-source checking, resolution readiness, ambiguity flags, or a scheduled workflow for near-resolution markets.

turn a polymarket thesis into a practical strategy draft. use when the user wants entry ideas, exit logic, invalidation, scenario planning, passive versus aggressive execution guidance, or a trade plan that still stops short of sending orders. use opportunity-classifier or deep-market-research first if the market has not been screened or modeled yet.

Work with WireSpec Markdown UI specifications and review artifacts. Use when creating, editing, validating, or adopting `.wirespec.md` files; turning product/UI requests into WireSpec before code; resolving WireSpec annotation sidecars or agent tasks under `.wirespec/reviews`; running WireSpec parser/runtime/bridge/core commands; or using WireSpec as an agent-facing contract in another repository.

build or edit product, marketing, and application ui from scratch or within an existing codebase while avoiding generic ai frontend habits, preserving reference designs, reusing existing tokens and components, and verifying hierarchy, states, breakpoints, and task clarity. use when codex is asked to create, restyle, refactor, review, or match html, css, react, vue, svelte, design-system, figma, or screenshot-based ui.

research and produce explicit probability forecasts and fair-odds views for current events, elections, policy outcomes, company metrics, mergers, regulatory approvals, and other uncertain questions. use when the user asks for odds, fair value, probability distributions, scenario weights, stress tests, or pricing for a prediction market or other yes/no or multi-outcome event. especially use for deadline-driven questions, election markets, quarterly company forecasts, and requests to compare market price versus fair probability without using market prices as evidence.

Set up, resume, pause, and run an autonomous experiment loop inside a git repository. Use when asked to run autoresearch, optimize something through repeated experiments, continue an existing loop from autoresearch.md or autoresearch.jsonl, or apply new user steering to the next experiment while keeping the loop going.

Read .agent-annotations (open inbox + assets) and turn UI annotations into an actionable plan. Optional browser validation only when requested.

subagent-native incumbent-challenger workflow for ideation, concept descriptions, proposals, wireframes, specs, implementation plans, and code or feature improvements. use when you want codex to explicitly spawn real subagents for planning, research, critique, challenger generation, synthesis, or blind judging instead of simulating fresh perspectives in one shared context. also use when you want repo artifacts, resumable round logs, mermaidjs flowcharts, a browser-viewable html report, web browsing by default, cross-platform bootstrap helpers, and recurring critiques promoted into rubrics, tests, or checks. do not use for trivial one-step edits or pure factual lookup requests.