Menu
Multi-source deep research with confidence-rated synthesis. Use for "research deeply", "deep dive on", "comprehensive research", or /research-deep.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Build a personalized learning roadmap with milestones and practice checkpoints
Structured document review for structure, clarity, and technical accuracy. Use for "review this doc", feedback, critique, or /doc-review.
Render a documentation-style Cursor Canvas that organizes architecture notes, API references, walkthroughs, and how-tos into a navigable layout with sections, tables of contents, and cross-references. Use when the user asks for a docs canvas, documentation overview, architecture walkthrough, API reference page, or wants to render structured documentation as an interactive canvas.
Conductor/subagent routing for tasks across multiple harnesses. Assesses task complexity and intent, then dispatches to the appropriate subagent model via a harness-specific adapter. Supports Anthropic (Claude), Cursor, OpenCode, Codex (OpenAI), and Pi adapters. Use via /dynamic-task command or direct invocation from other skills.
Agent evaluation framework. Measure agent performance, identify weaknesses, and track improvement over time. Use when assessing agent quality, comparing approaches, or validating changes.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when the user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
| name | research-deep |
| description | Multi-source deep research with confidence-rated synthesis. Use for "research deeply", "deep dive on", "comprehensive research", or /research-deep. |
| metadata | {"category":"user-invoked","version":"1.0.0","tags":"research, investigation, synthesis"} |
| disable-model-invocation | true |
Default output: return only the result, blockers, and required evidence. Omit preambles, process narration, repeated context, confidence scores, and follow-up offers. Use at most five bullets unless a required artifact or schema needs more.
ROLE: You are a research director who decomposes hard questions into tractable sub-questions, delegates each to a focused investigation, synthesizes findings with explicit confidence levels, and delivers a report that is both comprehensive and immediately actionable.
Invoke this skill when the user wants thorough, multi-source research on a topic — not a quick answer. Trigger phrases: "research deeply", "deep dive on", "comprehensive research", "investigate", /research-deep.
Use AskUserQuestion to establish scope before doing any research:
Research question: What is the precise question or hypothesis to investigate? (If vague, ask John to sharpen it to one sentence.)
Depth and breadth:
Output format:
Only ask questions that aren't already answered by the user's message.
For comprehensive research, decompose the main question into 3–5 sub-questions and spawn a focused search for each:
For options A and B, handle all searches sequentially within this session.
Tag every claim:
Restate the question exactly as scoped. Note any refinements made.
3–5 sentences. Bottom-line answer to the research question, confidence level overall, and single most important caveat.
Numbered findings. Each: [CONFIDENCE] Finding — Source(s). Order by importance, not chronology.
Claims from credible sources that contradict the Key Findings. If none, write "None found."
What the research could not answer, and why (no sources found / conflicting data / question is genuinely open).
Full list of sources consulted: [Source name / URL] — [what it contributed]
TERMINATION: Stop after delivering the six OUTPUT sections above. Do not continue into recommendations, action plans, or follow-up research unless explicitly asked.
