research-briefing

name	research-briefing
description	Use when starting a nontrivial research task to frame scope, success criteria, evidence plan, and the right downstream workflow before expensive browsing or multi-agent work begins.
metadata	{"short-description":"Brief a research task before execution"}

OpenCode Runtime Notes

This skill is installed as an OpenCode-native SKILL.md. For runtime-backed helpers, prefer the shared ai-agents-skills runtime root and the AAS_RUNTIME_ROOT override instead of assuming a Codex-specific runtime path.

Research Briefing

Use this as a lightweight gate before nontrivial research.

When to use

• the task needs deeper research, not a quick factual answer
• the user asks for a report, survey, comparison, or careful investigation
• the task may branch into source-research, deep-research-workflow, prose, or agent-group-discuss
• you want to surface scope, assumptions, and evidence needs before spending time

When not to use

• trivial lookups that can be answered directly
• after the user already approved a detailed research plan and nothing material changed

Output contract

Produce a short visible section titled Research Brief.

Keep it brief and include:

• Goal — what question the work must answer
• Scope — what is in and out
• Constraints — time, tools, source class, or formatting limits
• Context/style artifacts — prior posts, templates, house style, examples, source ledgers, or supplied materials to inspect before drafting or matching a publication format
• Style profile — whether writing-style-settings.md and any overlay such as math-manuscript-style.md will be active for downstream writing, plus the expected style_profile_ref, active_overlays, and active_requirement_ids
• Evidence plan — primary source types and verification expectations
• Workflow — which downstream research skill or path to use
• Risks — likely ambiguity, missing evidence, or live-data concerns

Use the compact template in references/brief-template.md when helpful.

Guardrails

• keep the brief short enough to read in a few seconds
• if the user's request is only to scope, analyze, investigate, diagnose, or report, stop after the brief or findings; do not continue into source retrieval, drafting, artifact creation, or remediation unless the user explicitly asks for that next action
• state assumptions explicitly instead of hiding them in later research
• for writing or publication-format tasks, inspect old posts, templates, house style, and supplied examples before drafting; if they are absent, say so and state the style assumption before writing
• for downstream writing, require the selected workflow to load writing-style-settings.md and record style_profile_ref, active_overlays, active_requirement_ids, and style_applied before final output
• if the task is simple, say so and skip heavyweight planning
• if the user already provided a plan, validate and tighten it rather than replacing it