New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Deep Research Loop

by madebyaris in madebyaris/advance-minimax-m3-cursor-rules

Runs multi-step research with a loop of search, compress, reflect and synthesize, scaling effort from a quick sourced answer to an exhaustive cited report.

MITAuto-check passedResearch & Science

Install Deep Research Loop

skills CLI
$ npx skills add madebyaris/advance-minimax-m3-cursor-rules --skill deep-research -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install madebyaris/advance-minimax-m3-cursor-rules deep-research --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/madebyaris/advance-minimax-m3-cursor-rules.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.cursor/skills/deep-research .claude/skills/deep-research && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
deep-research
GitHub stars
126
Token cost
~2.9k tokens
SKILL.md length
892 words
Files
2
Skills in repo
6
Repo updated
First seen
Licence
MIT

At a glance

Runs multi-step research with a loop of search, compress, reflect and synthesize, scaling effort from a quick sourced answer to an exhaustive cited report.

  • Works in 5 steps: Scope → Plan → Research Loop → …
  • Surveying the options for a technology or architecture decision
  • SKILL.md covers Effort Scaling, Phase 0: Scope, Phase 1: Plan and Phase 2: Research Loop, plus 5 more sections
  • Reaches react.dev

What it does

Before searching, the agent calibrates depth. A quick tier suits a focused factual question and uses two to three searches with no delegation, a standard tier handles multi-faceted topics and comparisons with five to eight searches and a sectioned analysis, and an exhaustive tier covers surveys and architecture decisions with 10+ searches, parallel `Task` investigations and a full report with citations.

Phase 0 scopes the request. The agent classifies it as a comparison, explanation, investigation, survey or fact-check, then picks sources: web search and fetch for general knowledge and library docs, semantic search, grep and file reads for codebase questions, or a mix when external practice and internal conventions both matter. It writes a one-paragraph research brief that it does not show you, and Phase 1 splits the brief into sub-queries, for a comparison one per item plus one for the criteria. A `reference.md` file is bundled, and the excerpt ends partway through the planning phase.

When your agent uses it

  • Surveying the options for a technology or architecture decision
  • Comparing two libraries or approaches with sourced evidence
  • Investigating why something happens using both the web and the codebase
  • Checking whether a claim is true before relying on it

Example prompts

  • “Do a deep dive on the current state of CSS-in-JS and cite your sources.”
  • “Compare Postgres and MongoDB for our event-logging workload.”
  • “Research why our build got slower using both our repo and public discussions.”
  • “Verify whether React 19 still needs keys in lists.”

Requirements

  • Web search and page fetch tools
  • Codebase search tools for questions about a repository

Workflow steps

5 steps, taken from the step headings in SKILL.md.

  1. Scope
  2. Plan
  3. Research Loop
  4. Synthesize
  5. Deliver

What it can do on your machine

Read from SKILL.md and the folder at commit 4d6c552. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Hosts in commands or code, which the agent is likely to contact:

    • react.dev

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Deep Research Loop loads about 2.9k tokens when it runs. Until then it costs about 90 tokens; SKILL.md has 892 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~90
When it runs · the whole SKILL.md, loaded when a task matches
~2.9k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from madebyaris/advance-minimax-m3-cursor-rules at commit 4d6c552, republished under its MIT licence (© madebyaris). 892 words, ~2,900 tokens.

Download SKILL.mdSave it as .claude/skills/deep-research/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
deep-research
description
Conducts multi-step deep research on any topic using iterative search, reflection, and synthesis. Use when the user asks to research, investigate, survey, compare, analyze, deep-dive, or explore a topic in depth. Covers web research, codebase analysis, documentation review, mixed-source investigation, and M3 long-context compression discipline.
license
MIT
metadata.version
1.2.0
metadata.category
research
metadata.sources
Cursor-native tool workflows, Documentation review and comparative research practice, Repo and web synthesis patterns

Deep Research

Conduct thorough, multi-step research using an iterative loop of search, compress, reflect, and synthesize. Works with any Cursor-supported model.

Effort Scaling

Before starting, calibrate depth to the question:

TierWhenSearchesDelegation (Task)Output
QuickFocused factual question, single concept2-3NoneConcise answer with sources
StandardMulti-faceted topic, comparison, how-something-works5-8NoneStructured analysis with sections
ExhaustiveComprehensive survey, architecture decision, landscape review10+Parallel Task investigationsFull report with citations
Calibration:
  Question complexity: [single-fact / multi-faceted / comprehensive]
  Source diversity needed: [one source type / mixed]
  User expectation: [quick answer / detailed analysis / full report]
  -> Tier: [Quick / Standard / Exhaustive]

Phase 0: Scope

Immediately classify the research request before any searching.

Step 1 -- Classify research type:

TypeSignalExample
Comparison"vs", "compare", "which is better", "difference between""React vs Vue for enterprise apps"
Explanation"how does", "what is", "explain", "why does""How does Raft consensus work?"
Investigation"debug", "find out why", "what caused", "root cause""Why is our build 3x slower?"
Survey"landscape", "options for", "state of", "overview""State of CSS-in-JS in 2026"
Fact-check"is it true", "verify", "confirm""Does React 19 still need keys?"

Step 2 -- Determine sources:

SourceWhen to use
WebSearch + WebFetchGeneral knowledge, current events, library docs, community solutions
SemanticSearch + Grep + ReadCodebase-specific questions, internal patterns, project architecture
Mixed"How should we implement X?" (need both external best practices and internal conventions)

Step 3 -- Generate a one-paragraph research brief:

Research brief:
  Question: [exact user question]
  Type: [comparison / explanation / investigation / survey / fact-check]
  Sources: [web / codebase / mixed]
  Tier: [quick / standard / exhaustive]
  Key dimensions to cover: [list 3-5 specific aspects]
  Out of scope: [anything explicitly excluded]

Do NOT present this brief to the user. Proceed to Phase 1 immediately.

Phase 1: Plan

Decompose the research brief into concrete sub-queries.

Decomposition strategy by type:

  • Comparison: One sub-query per item being compared, plus one for the comparison criteria
  • Explanation: Start broad (overview), then narrow (mechanism, edge cases, alternatives)
  • Investigation: Hypothesis-first -- form 2-3 hypotheses, create sub-queries to test each
  • Survey: One sub-query per category/dimension in the landscape
  • Fact-check: One sub-query for the claim, one for counter-evidence, one for authoritative source

For Standard/Exhaustive tier, create a TodoWrite tracker:

TodoWrite(todos=[
  { id: "DR-scope", content: "Research: [brief summary]", status: "completed" },
  { id: "DR-q1", content: "Sub-query: [first sub-query]", status: "in_progress" },
  { id: "DR-q2", content: "Sub-query: [second sub-query]", status: "pending" },
  ...
  { id: "DR-synth", content: "Synthesize findings into report", status: "pending" }
], merge=false)

For Exhaustive tier, evaluate which sub-queries are independent (can run in parallel via Task) vs. dependent (must run sequentially because results inform next query).

Phase 2: Research Loop

This is the core iterative cycle. Execute it per sub-query.

Web research pattern:

1. WebSearch(search_term="[specific, well-formed query] [current year if recency matters]")
2. If a result looks highly relevant, WebFetch the full page
3. Immediately compress: extract only the facts relevant to the sub-query

Codebase research pattern:

1. SemanticSearch(query="[natural language question]", target_directories=[relevant dir])
2. If results point to specific files, read them with `Read`
3. If searching for exact symbols, use `Grep`
4. Compress: extract the pattern/answer, not the full file contents

Parallel Task pattern (Exhaustive tier only):

Launch up to 3 parallel `Task` investigations for independent sub-queries:

Task(
  subagent_type="generalPurpose",
  model="fast",
  readonly=true,
  description="Research [topic]",
  prompt="Research the following question and return a compressed summary with sources:
    Question: [sub-query]
    Search using WebSearch and WebFetch. Return:
    1. Key findings (bullet points)
    2. Sources (title + URL for each)
    3. Confidence: certain / likely / uncertain
    Do NOT return raw search results. Summarize.",
)

Do NOT accumulate raw search results. After each search or WebFetch:

Compression template:
  Source: [URL or file path]
  Key finding: [1-3 sentences of relevant information]
  Confidence: [certain / likely / uncertain]
  Relevance: [directly answers sub-query / provides context / tangential]

Drop tangential results immediately. Only carry forward "directly answers" and "provides context" findings.

M3 nudge: with a 1M-token context, the failure mode shifts from "ran out of room" to "kept too much raw output." Apply the compression template aggressively. If you have run 3+ searches without compressing, the next reflection must include a compression pass. See minimax-m3-long-context for the broader retention/discard plan.

Reflect (after every 2-3 searches)

Pause and evaluate using this checklist:

Reflection checkpoint:
  1. Coverage: Which sub-queries are answered? Which have gaps?
  2. Confidence: Am I seeing convergence across sources, or contradictions?
  3. Diminishing returns: Are my last 2 searches finding new information, or repeating what I already know?
  4. Pivots needed: Has anything I found changed what I should be searching for?
  5. Sufficiency: Can I answer the original question with what I have?

  Decision: [continue searching / pivot strategy / proceed to synthesis]

Stop searching when:

  • 3+ independent sources confirm the same finding
  • Last 2 searches returned no new information
  • All sub-queries are answered at the target confidence level
  • Maximum search budget for the tier is reached

Pivot when:

  • Initial hypothesis was wrong -- reformulate sub-queries
  • A new dimension emerged that the original plan missed -- add a sub-query
  • Sources contradict each other -- search for authoritative tiebreaker
Evolving Summary

Maintain a running summary that gets updated (not appended to) after each reflection:

Working summary (updated, not appended):
  [Paragraph 1: What I know with high confidence]
  [Paragraph 2: What I know with moderate confidence]
  [Paragraph 3: Open questions / contradictions / gaps]
  Sources so far: [numbered list]

This is the "evolving report as memory" pattern. Previous raw search results can be released from active context once compressed into this summary.

Show full SKILL.md (358 more words)Show less

Phase 3: Synthesize

Generate the final output in a SINGLE pass from the evolving summary and compressed findings.

Do NOT:

  • Generate sections independently and merge them (produces disjointed output)
  • Copy-paste raw search results into the report
  • Include findings you flagged as "tangential" during compression

Do:

  • Write the full response in one coherent pass
  • Resolve contradictions explicitly ("Source A claims X, while Source B claims Y. Based on [reasoning], Y is more credible because...")
  • Organize with clear headings for Standard/Exhaustive tier
  • Include inline citations: [Source Title](URL) or file path references

Structure by research type:

  • Comparison: Table or side-by-side, then analysis of tradeoffs, then recommendation
  • Explanation: Overview, then mechanism/details, then edge cases/caveats
  • Investigation: Hypothesis, evidence for/against, conclusion
  • Survey: Categories, key players/options per category, trends, recommendations
  • Fact-check: Claim, evidence, verdict (confirmed/partially true/false/unverifiable)

Phase 4: Deliver

Citation Format

Every factual claim must have a source. Use inline links:

React Server Components reduce bundle size by up to 30% [React Blog](https://react.dev/blog/...).

For codebase findings, cite file paths:

The auth middleware uses JWT validation (`src/middleware/auth.ts:42-58`).
Confidence Flags

Honesty by construction: never assert an API, version, or fact from memory as current, and never cite a URL you did not actually retrieve. Label each finding verified / unverified / assumption, and if a path yields no usable signal after two reformulations, stop and escalate with one concrete question rather than looping. See the Anti-Hallucination & Failure Recovery section in reference.md.

End the report with an honest assessment:

Confidence assessment:
  - High confidence: [claims well-supported by multiple sources]
  - Moderate confidence: [claims from single authoritative source]
  - Low confidence / needs verification: [claims from informal sources or with contradictions]
Mark Completion

Update TodoWrite to mark all research sub-queries and synthesis as completed.

Model Compatibility

This skill uses only Cursor-native tools and plain behavioral instructions:

  • No model-specific prompting syntax
  • No assumptions about thinking/reasoning format
  • Tool names in this skill are illustrative; use the exact identifiers and schemas from the active session. In Composer-style Cursor agents you will typically see Read, Grep, StrReplace, Task (delegation), WebSearch, WebFetch, SemanticSearch, TodoWrite, and others—names differ in older docs or other products (ReadFile, ApplyPatch, Subagent, etc.)
  • Reflection happens in whatever reasoning mechanism the model supports

The iterative search-compress-reflect loop is a behavioral pattern, not a code construct. Any model that can call tools and reason about results can execute it.

Quick Reference

SCOPE  -> Classify type + sources + tier (no searching yet)
PLAN   -> Decompose into sub-queries, create tracker
SEARCH -> Execute queries, compress each result immediately
REFLECT -> Every 2-3 searches: coverage? gaps? pivot? stop?
SYNTH  -> One-shot report from compressed findings
DELIVER -> Citations, confidence flags, completion

Additional Resources

  • For detailed examples, process failure modes, and the anti-hallucination / failure-recovery layer (fabricated APIs, bad citations, dead-end loops, confidence honesty), see reference.md

© madebyaris, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file in .cursor/skills/deep-research of madebyaris/advance-minimax-m3-cursor-rules.

  • SKILL.md
  • reference.md

Open the folder on GitHubat commit 4d6c552

Compare with similar skills

Deep Research Loop next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Deep Research Loop compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Deep Research Loop this skillmadebyaris/advance-minimax-m3-cursor-rules126—~2.9kAutomated safety check: PassMIT
Net Deep Researchh4444433333/net-deep-research123—~3.3kAutomated safety check: PassMIT
Ray Trend Searchimraywang/rayskills159—~2.1kAutomated safety check: PassCustom licence
Argo Search and Verificationtaxueseek/argo186—~1.2kAutomated safety check: PassMIT
Multi Source Searchsandbaseai/sandbase-skills202—~1.6kAutomated safety check: PassApache-2.0
Research LookupK-Dense-AI/claude-scientific-writer2.4k2 repos~3.6kAutomated safety check: PassMIT

Similar skills

  • Net Deep Research

    h4444433333/net-deep-research

    Runs cross-source web research to verify whether an online claim is true, distinguishing confirmed facts from rumor, marketing claims or stale information.

    123 GitHub stars~3.3k tokensUpdated 1 mo ago
    Research & ScienceAuto-check passed

  • Ray Trend Search

    imraywang/rayskills

    Researches what people are saying about a topic over a recent window across X, Reddit, YouTube and the public web, reporting each source's status with links.

    159 GitHub stars~2.1k tokensUpdated 16 days ago
    Research & ScienceAuto-check passed

  • Argo Search and Verification

    taxueseek/argo

    Unified web search, page fetching and evidence checking across hundreds of sources, with result verification, a research-dossier mode and vertical search engines.

    186 GitHub stars~1.2k tokensUpdated 3 days ago
    Research & ScienceAuto-check passed

  • Multi Source Search

    sandbaseai/sandbase-skills

    Portable multi-source research with cross-source validation and an offline evidence ledger.

    202 GitHub stars~1.6k tokensUpdated 13 days ago
    Research & ScienceAuto-check passed

  • Research Lookup

    K-Dense-AI/claude-scientific-writer

    Compile current scholarly evidence for a scientific manuscript or research brief.

    2.4k GitHub starsUsed in 2 repos~3.6k tokens
    Research & ScienceAuto-check passed

  • Deep Research

    asgeirtj/system_prompts_leaks

    Deep research harness — fan-out web searches, fetch sources, adversarially verify claims, synthesize a cited report.

    69k GitHub stars~390 tokensUpdated yesterday
    Research & ScienceAuto-check passed

More from madebyaris/advance-minimax-m3-cursor-rules

  • 3D Web Experiences

    madebyaris/advance-minimax-m3-cursor-rules

    Builds 3D web scenes with Three.js and React Three Fiber, with attention to art direction, a performance budget, mobile behavior and fallbacks.

    126 GitHub stars~3k tokensUpdated 3 mo ago
    Auto-check passed

  • Incident Triage Harness

    madebyaris/advance-minimax-m3-cursor-rules

    Walks an agent through an evidence-first incident investigation across logs, metrics, code and screenshots, from first symptom to the smallest safe mitigation.

    126 GitHub stars~984 tokensUpdated 3 mo ago
    Auto-check passed

  • MiniMax M3 Long-Context Discipline

    madebyaris/advance-minimax-m3-cursor-rules

    Teaches how to work within MiniMax M3's 1M-token context: decide per source what to keep, summarize or drop, plan the loading, and cap raw blocks across iterations.

    126 GitHub stars~1.6k tokensUpdated 3 mo ago
    Auto-check passed

  • MiniMax M3 Multimodal Input

    madebyaris/advance-minimax-m3-cursor-rules

    Teaches an agent on MiniMax M3 to ground visual claims in attached images, screenshots, mockups and clips, and to re-read results after a visual change.

    126 GitHub stars~1.6k tokensUpdated 3 mo ago
    Auto-check passed

  • MiniMax Multimodal Toolkit

    madebyaris/advance-minimax-m3-cursor-rules

    Routes image, video, voice, music and media-processing requests to the smallest suitable MiniMax or local path, with FFmpeg-style processing around generated media.

    126 GitHub stars~1.6k tokensUpdated 3 mo ago
    Auto-check passed

Categories

Questions about Deep Research Loop

What does Deep Research Loop do?

Runs multi-step research with a loop of search, compress, reflect and synthesize, scaling effort from a quick sourced answer to an exhaustive cited report. Before searching, the agent calibrates depth. A quick tier suits a focused factual question and uses two to three searches with no delegation, a standard tier handles multi-faceted topics and comparisons with five to eight searches and a sectioned analysis, and an exhaustive tier covers surveys and architecture decisions with 10+ searches, parallel `Task` investigations and a full report with citations.

When should I use Deep Research Loop?

Deep Research Loop fits situations like: surveying the options for a technology or architecture decision; comparing two libraries or approaches with sourced evidence; investigating why something happens using both the web and the codebase; checking whether a claim is true before relying on it.

How do I install Deep Research Loop in Claude Code?

Run `npx skills add madebyaris/advance-minimax-m3-cursor-rules --skill deep-research -a claude-code`. Or copy the skill folder (.cursor/skills/deep-research in madebyaris/advance-minimax-m3-cursor-rules) into .claude/skills/deep-research in your project. Claude Code loads it when a task matches its description.

How do I install Deep Research Loop in Codex?

Run `npx skills add madebyaris/advance-minimax-m3-cursor-rules --skill deep-research -a codex`. Or copy the skill folder (.cursor/skills/deep-research in madebyaris/advance-minimax-m3-cursor-rules) into .agents/skills/deep-research in your project. Codex loads it when a task matches its description.

Can I use Deep Research Loop in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add madebyaris/advance-minimax-m3-cursor-rules --skill deep-research -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/deep-research, .gemini/skills/deep-research, .github/skills/deep-research and .opencode/skills/deep-research in your project.

What does Deep Research Loop need to run?

SKILL.md names no scripts, command-line tools or credentials: Deep Research Loop is instructions for the agent only. Our summary lists: Web search and page fetch tools; Codebase search tools for questions about a repository.

Does Deep Research Loop access the network?

SKILL.md names 1 domain. In commands or code: react.dev; the agent is likely to contact it when it follows the instructions. This is read from the text; nothing was executed.

Is Deep Research Loop safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Deep Research Loop use?

Deep Research Loop is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Deep Research Loop use?

About 2.9k tokens (SKILL.md is roughly 12k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Deep Research Loop?

Skills that share tags, products or a category with Deep Research Loop: Net Deep Research (h4444433333/net-deep-research, 123 stars), Ray Trend Search (imraywang/rayskills, 159 stars), Argo Search and Verification (taxueseek/argo, 186 stars) and Multi Source Search (sandbaseai/sandbase-skills, 202 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Deep Research Loop?

madebyaris (a GitHub user) maintains it in madebyaris/advance-minimax-m3-cursor-rules, which has 126 GitHub stars. The repository holds 6 skills in this directory. The repository was last updated on June 16, 2026.

Source: madebyaris/advance-minimax-m3-cursor-rules on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.