菜单
How to use the search_web tool well when grounding a forecast in recent news — phrase cutoff-aware queries, decide what is worth searching for, and weigh sources. Load this before your first search_web call. No scripts.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
How to use the search_web tool well when grounding a forecast in recent news — phrase cutoff-aware queries, decide what is worth searching for, and weigh sources. Load this before your first search_web call. No scripts.
How to use the search_web tool well when grounding a forecast in recent news — phrase cutoff-aware queries, decide what is worth searching for, and weigh sources. Load this before your first search_web call. No scripts.
How to use the search_web tool well when grounding a forecast in recent news — phrase cutoff-aware queries, decide what is worth searching for, and weigh sources. Load this before your first search_web call. No scripts.
Reference guide for the repo concierge catalog — domain filters, the search-then-fetch workflow, and bootcamp routing. Load references/catalog-guide.md before your first answer. No scripts.
How to use the code execution sandbox well — parse the JSON payload (not disk files), compute a couple of useful diagnostics before forecasting, and keep the session stateful within a turn. Load this before writing code. No scripts.
The output contract for producing a structured probability distribution over the rate decision (cut / hold / hike) — the JSON shape, the calibration rules, and how to submit it. Load this ONLY when your task payload asks for a forecast; ignore it for open-ended questions. No scripts.
| name | research-playbook |
| description | How to use the search_web tool well when grounding a forecast in recent news — phrase cutoff-aware queries, decide what is worth searching for, and weigh sources. Load this before your first search_web call. No scripts. |
A short guide to getting real signal out of search_web. This is a starter
skill — extend it with the queries and sources that work for your problem.
Always pass cutoff_date equal to the as_of date in your payload. It is the
temporal fence that keeps post-origin information out of a historical forecast.
A forecast that "knew" what happened after as_of is not a forecast.
search_web runs an independent verifier on every result and returns
[SEARCH_VERIFICATION_FAILED] instead of content it couldn't confirm as
pre-cutoff. Treat that as no verified news for the query — proceed on your
other signals and say so, never filling the gap from your own background
knowledge.
For a Bank of Canada rate decision, the signals that move the odds: recent CPI and core-inflation prints vs the 2% target, the labour market (employment, unemployment, wages), market pricing (overnight index swaps, economist surveys), recent BoC communications, and macro shocks (oil, the loonie, US policy, trade). Search for the current state of these, then let the base rates set the prior.
references/ file with example high-signal searches.