spike
Skill1 starsChanged yesterdayMITView on GitHub
Run throwaway experiments, sized in story points, to answer a feasibility question before committing to a real build. Use when the user says 'spike this', 'is this even possible?', 'let me try X', 'quick prototype', 'proof of concept', 'compare A vs B', 'before I commit to Y', or 'what should I spike next?'. Not for questions answerable by reading docs or code, and not for production implementation.
56 other projects use a version of this.
Adopted by
Most recently taken up by
- rodos4312/Meu-Agenttheir own version, first seen Nov 15, 2026
- fabioc-aloha/Alex_ACT_ONEtheir own version, first seen Nov 15, 2026
- metaspartan/cybaratheir own version, first seen Nov 15, 2026
- NousResearch/hermes-agenttheir own version, first seen Nov 13, 2026
- hnqlv/skillstheir own version, first seen Nov 12, 2026
- ice-lfernandes/claude-spring-architecttheir own version, first seen Nov 8, 2026
Used with
Projects with it also use teams-meeting-pipeline, touchdesigner-mcp and 4 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
john-data-chenSkillteams-meeting-pipeline
Operate the Teams meeting summary pipeline via Hermes CLI — summarize meetings, inspect pipeline status, replay jobs, manage Microsoft Graph subscriptions.
Also in 8 projects that use this
john-data-chenSkilltouchdesigner-mcp
Control a running TouchDesigner instance via twozero MCP — create operators, set parameters, wire connections, execute Python, build real-time visuals. 36 native tools.
Also in 5 projects that use this
Tommy-ywSkillsongwriting-and-ai-music
Songwriting craft, AI music generation prompts (Suno focus), parody/adaptation techniques, phonetic tricks, and lessons learned. These are tools and ideas, not rules. Break any of them when the art calls for it.
Also in 8 projects that use this
mae616Skillp5js
p5.js(クリエイティブコーディング)のProcessing由来「コードでスケッチ」思想を軸に、setup/drawループ・座標系・インタラクション・パフォーマンス最適化を整理する。p5.js、Processing、クリエイティブコーディング、ジェネラティブアート、インタラクティブ表現、ビジュアルプログラミング、Canvas描画、パーティクル、フラクタル、ノイズ表現に関する相談で必ず使うこと。ユーザーが「コードでアート」「動くビジュアル」「インタラクティブな描画」と言った場合にも適用する。
Also in 8 projects that use this
Tommy-ywSkillpopular-web-designs
54 production-quality design systems extracted from real websites. Load a template to generate HTML/CSS that matches the visual identity of sites like Stripe, Linear, Vercel, Notion, Airbnb, and more. Each template includes colors, typography, components, layout rules, and ready-to-use CSS values.
Also in 8 projects that use this
math-incSkillfindmy
Track Apple devices and AirTags via FindMy.app on macOS using AppleScript and screen capture.
Also in 8 projects that use this
Other versions56 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 56
- openclaw/openclaw★ 395ksame name
- moltbot/clawdbot★ 394ksame name
- steipete/warelay★ 394ksame name
- steipete/clawdis★ 394ksame name
- moltbot/moltbot★ 394ksame name
- clawdbot/clawdbot★ 394ksame name
- molt-bot/clawdbot★ 391ksame name
- NousResearch/hermes-agent★ 264ksame name
- OpenBMB/PilotDeck★ 5ksame name
- friuns2/openclaw-android-assistant★ 573same name
46 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository28 other files it gives its agents.
- AGENTS.mdAGENTS.md
- vercel-react-best-practices/AGENTS.mdAGENTS.md
- architecture-reviewSkill
- awsSkill
- code-review-frontendSkill
- comment-sickoSkill
- demo-scriptSkill
- documentation-writerSkill
- dotnetSkill
- factory-reviewSkill
- factorySkill
- lets-talk-about-itSkill
- nextjs-proxy-csp-hardeningSkill
- performance-review-humanizerSkill
- pr-body-styleSkill
- prd-modeSkill
- prove-it-worksSkill
- pythonSkill
- qa-reviewSkill
- reconcile-reviewsSkill
- security-reviewSkill
- simplifySkill
- sweSkill
- test-reviewSkill
- typescript-strict-styleSkill
- unslopSkill
- vercel-react-best-practicesSkill
- whats-going-onSkill
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 9 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 52 edited versions we read keep these.
9
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 24
Read the reference for each step when you reach it.
- Very specific: line 44
Read references/evidence.md before you build and before you report.
- Distinctive: line 47
Never invent or "tidy" output, numbers or results.
- Distinctive: line 48
Every verdict rests on pasted output from a run of the actual question.
- Distinctive: line 50
Before you tell the user anything, trace each claim back to output in a README.
- Distinctive: line 51
Every spike has one testable question and a story-point estimate.
- Distinctive: line 52
Never write "it works" after one happy-path run.
- Distinctive: line 53
If the experiment can be run, run it.
- Distinctive: line 54
Default stack is AWS and Terraform.
- Distinctive: line 56
No real credentials, and no touching other shared systems without asking.
“Common” is measured: kept by at least 6 in 10 of the 52 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Spike
A spike is a disposable experiment that answers one specific question so the team can decide or estimate with less risk. The deliverable is the verdict, not the code.
When NOT to use this
- The answer is in the docs or the existing code: research it and answer, don't build.
- The idea is already validated: implement it.
- The work is production code: the throwaway rules here do not apply.
Loop
decompose → align → research → build → verdict
↑______________________________________↓
iterate on findingsRead the reference for each step when you reach it. Do not load them all up front.
| Step | What you do | Read |
| 1. Decompose | Split the idea into 2-5 feasibility questions with Given / When / Then, risk and story points. Order by risk. | references/decomposition.md |
| 2. Align | Show the table. Ask "Build all in this order, or adjust?". No code until the user answers. | references/decomposition.md |
| 3. Research | Pick an approach per spike. Reuse what the repo has. Keep it short. | references/research.md |
| 4. Build | One standalone directory per spike. Interactive output, real evidence, edge and failure cases. | references/building.md, references/template.md |
| 5. Verdict | VALIDATED, PARTIAL, INVALIDATED or INCONCLUSIVE, backed by pasted results. Run the final check, then give a short summary. | references/verdicts.md, references/evidence.md |
Other situations:
| Situation | Read |
| The spike touches any AWS service | references/aws-floci.md |
| Two or more approaches answer the same question | references/comparison.md |
| Spikes already exist and the user asks what to spike next | references/frontier.md |
Non-negotiables
Read references/evidence.md before you build and before you report. It overrides everything else.
- Nothing is reported as true unless you ran it in this session and saw the output. Your memory, the docs and your reading of the code are not tests. Label them as untested.
- Never invent or "tidy" output, numbers or results. Paste what the run printed. If you could not run something, say so.
- Every verdict rests on pasted output from a run of the actual question. With no run, the verdict is INCONCLUSIVE.
- Say what you did not test, every time.
- Before you tell the user anything, trace each claim back to output in a README. Remove or relabel what you cannot trace.
- Every spike has one testable question and a story-point estimate. Run the riskiest spike first.
- Never write "it works" after one happy-path run. Test an edge case and a failure case.
- If the experiment can be run, run it. For AWS that includes Terraform, which works against Floci. Never answer from assumption when a test was possible.
- Default stack is AWS and Terraform. Use Terraform for any infrastructure-as-code in a spike, unless the repo already uses another tool. Use CDK or OpenTofu only if the user asks.
- AWS operations run on Floci in Docker, and never on real AWS. Results are reported as "on Floci [version]", never as "works on AWS".
- No real credentials, and no touching other shared systems without asking.
- INVALIDATED and INCONCLUSIVE are honest outcomes. Report them plainly.
Output
spikes/at the repo root (or the folder the repo already uses), oneNNN-name/directory per spike.- Each directory has a
README.mdfilled in fromreferences/template.md. - End with a chat summary: one line per spike with its verdict, the overall recommendation, what was not tested, and the biggest remaining unknown.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

