agentmods.dev

idea-spark

idea-spark is a skill for Claude Code, Codex from microsoft/ResearchStudio. It costs 65 tokens per session (2,386 once invoked), scanned A, original, MIT.

A workflow for turning a research direction into one specific research proposal with a method and a way to test whether it is wrong. It grounds the proposal in recent machine-learning research papers.

In plain words
What is it for?
Developing one research idea, checking its novelty, diagnosing a research bottleneck, and shaping a paper proposal.
Why use it?
It narrows vague brainstorming into a research idea that can be implemented, evaluated, and defended against reviewer criticism.

Skill for Claude CodeCodex ✓ vendor

Written for no agent in particular: nothing here depends on one. Also seen: mentions subagents.

not rated 3.4krepo +78 changed 1mo ago A scan SkillSpector: warn 65 tokens original MIT

Good fit Developing one research idea, checking its novelty, diagnosing a research bottleneck, and shaping a paper proposal.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/microsoft/researchstudio/idea_spark
View source ↗ microsoft/ResearchStudio aka.ms
About the project

ResearchStudio is a collection of skills that helps researchers develop an initial research direction into a paper and its related publication materials. It is used with Claude Code and Codex to create outputs such as papers, posters, videos, blogs, and reels. The catalogue entries are its own workflow skills for different stages of the research lifecycle.

microsoft/ResearchStudio · 3,378 stars · on GitHub · aka.ms

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add microsoft/ResearchStudio --skill idea_spark
Clone the repo
git clone --depth 1 https://github.com/microsoft/ResearchStudio

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for idea-spark

README.md
[![agentmods](https://agentmods.dev/badge/skills/microsoft/researchstudio/idea_spark/github.svg)](https://agentmods.dev/skills/microsoft/researchstudio/idea_spark)
Your own site
<a href="https://agentmods.dev/skills/microsoft/researchstudio/idea_spark"><img src="https://agentmods.dev/badge/skills/microsoft/researchstudio/idea_spark/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for idea-spark

Your own site · 80×15
<a href="https://agentmods.dev/skills/microsoft/researchstudio/idea_spark"><img src="https://agentmods.dev/badge/skills/microsoft/researchstudio/idea_spark.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 65 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,386 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 3 findings, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium Excessive Agency · line 74
    Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
    Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
  • medium Rogue Agent · line 94
    Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
    Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
  • medium Rogue Agent · line 150
    Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
    Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00065 $0.02386
Opus 5.5 $0.00026 $0.00954
Sonnet 5.5 $0.00013 $0.00477
Haiku 4.5 $0.00006 $0.00239

Measured 1mo ago against content hash faac45547391, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

idea-spark scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

ResearchStudio-Idea/skills/idea_spark/SKILL.md · 169 lines

How it starts

The opening of the file, as written. The whole thing — 169 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Idea Spark

Generate one literature-grounded research proposal with a concrete mechanism and honest falsification plan. A completed proposal is NOT evidence that the method works, is novel, or that a theorem is proved. Ordinary Chinese and English serve readers outside the subfield; the detailed English card retains technical obligations and implementation choices.

One run is one fresh directory: next writes a run_contract.json into an empty directory and drives only directories that carry one. A directory with files but no contract is refused, not resumed.

Driving a run

Use an absolute skill path and a fresh absolute run directory. Naming convention remains $PWD/ideaspark_run/<topic-slug>; if taken, append a numeric suffix. Never reuse phase0. Run python3 "$SKILL_DIR/scripts/run.py" next --dir "$RUN_DIR" --query "<direction>". Consume the WHOLE emitted step, including every input path. Execute it, then run next again. The navigator is read-only. It first emits quality_init to establish the run contract.

Read host-runbook.md BEFORE the first model call: context discipline (isolated sub-agent per call, artifacts to disk never to chat, compact between phases), consume every emitted step WHOLE, exit codes 10/11 are handshakes, entry points, validators and configuration. Retain the existing Phase 0/1 retrieval/diagnosis strategy: read retrieval-runbook.md when routed there. Setup remains in setup.md. Do not fabricate retrieval records. No mid-flow questions: infer missing intake fields honestly within the user's scope. Use existing connector caches/retries and report degraded grounding.

Version-bound model work

Every Phase 2+ model call uses a prepared request in .quality/<stage>/request.json. Preparation fingerprints actual inputs and prompts BEFORE execution. Read that request and the named prompt(s), then use one fresh isolated context. Selection + generation stay in the same context; generation vs audit, revision vs review, and author vs implementability review must be separate contexts. Return only output path and concise status to the host.

Read the full file on GitHub · 169 lines

Files

What ships with it

60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 1mo ago Changed · -104 lines faac45547391
  2. 1mo ago Changed · +1 lines ad9237cbc547
  3. 1mo ago First seen · 272 lines · 65 tokens per session scan A 5dfb79c52a00

Subscribe to this mod's changes

idea-spark is a skill published in the GitHub repository microsoft/ResearchStudio (3,378 stars, last pushed 1mo ago), licensed MIT. It adds 65 tokens to every session and 2,386 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Researchblog-generationidea-generationposter-generationpresentationresearchstudio-idearesearchstudio-reel
Related

Other skills, from other repositories

ppt-master

AI-driven presentation workflow for generating editable PPTX decks and slides, reconstructing page visuals, creating reusable Brand/Style/Layout/Deck workspaces, filling native PPTX templates, and enhancing finished PPTX files. Use when the user asks to create, generate, reconstruct, regenerate, beautify, redesign…

hugohe3/ppt-master · 108 tokens

slidev

Create and present web-based slidedecks for developers using Slidev with Markdown, Vue components, code highlighting, animations, and interactive features. Use when building technical presentations, conference talks, code walkthroughs, teaching materials, or developer decks.

slidevjs/slidev · 51 tokens

pptagent

Create and revise editable PowerPoint presentations from a brief using HTML slide sources, rendering, and visual review. Use when the requested deliverable is a PPTX deck. The host authors the slides; text-only hosts can use an external visual reviewer.

icip-cas/PPTAgent · 52 tokens

ppt-keynote

A browser-based Keynote-style presentation template where each slide is a full-screen section with keyboard navigation and fade transitions.

nexu-io/html-anything · 24 tokens

sciatlas-idea-evaluate

Use only the current SciAtlas automated review workflow (reviewpipeline) to take a novice user from zero setup to a final automated review of a research idea or paper, including setup, registration guidance, workflow configuration, retrieval, artifact reading, and synthesis. Trigger when the user asks whether an idea…

zjunlp/SciAtlas · 96 tokens

sciatlas-idea-generate

Use only the current SciAtlas multi-step idea-generation workflow (sciatlasideagen) to take a novice user from zero setup to final literature-grounded research idea seeds, including setup, registration guidance, workflow configuration, KG retrieval, artifact reading, novelty checking, and synthesis. Trigger when the…

zjunlp/SciAtlas · 92 tokens