{"slug": "jane-street-open-sources-agent-skills-with-a-socratic-tutor-skill-md", "title": "Jane Street Open-Sources agent_skills With a Socratic Tutor SKILL.md", "summary": "Jane Street published agent_skills, an MIT-licensed GitHub repository, in early November 2026, drawing roughly 2,100 stars by November 10, according to the repository README and GitHub counter. The repo's first skill, socratic-tutor, is a host-agnostic SKILL.md reference implementation that turns a model into a strict tutor: it decomposes a topic into 4–10 subskills, runs a 3–5 question diagnostic before teaching, renders an ASCII skill map with levels 0–5, and asks one question per message, with anti-sycophancy rules requiring two independent checks before any level bump. The release matters because it packages a portable, auditable Agent Skills folder pattern that can be copied into Claude Code skills folders or other harnesses without a Python or npm install.", "body_md": "**Jane Street** — the quantitative trading firm known for OCaml-heavy engineering — published **[agent_skills on GitHub](https://github.com/janestreet/agent_skills?ref=explainx)** in early November 2026. The repo crossed **roughly 2,100 stars** within days, per the [repository README](https://github.com/janestreet/agent_skills?ref=explainx) and GitHub counter on November 10.\n\nIt is not a model release. It is a **reference implementation** of the same **Agent Skills folder pattern** AWS Quick [adopted in its November skills guide](https://explainx.ai/blog/aws-quick-skills-agents-catalog-build-guide-2026) and Anthropic popularized for Claude agents. The first skill, **`socratic-tutor`**, turns a host model into a **strict personal tutor** that teaches through questions, diagnostics, and leveled mastery checks.\n\n## TL;DR\n\n| Question | Answer | \n|---|---|\n| License? | **MIT** per LICENSE.md in the repo | \n| Skills shipped? | **`socratic-tutor`** today; README invites more | \n| Format? | Directory with root **SKILL.md** plus optional assets | \n| Install? | Copy folder into agent skills path or paste SKILL body | \n| Trigger phrases? | **teach me** ,**tutor me** ,**quiz me** ,**help me master** | \n| Star velocity? | **2.1k+** stars since**November 3, 2026** creation | \n\n## Why a trading firm shipped a tutor skill\n\nJane Street's public engineering brand centers on **disciplined learning** — interns famously work through **Socratic problem sets**. Publishing a **SKILL.md** that encodes that pedagogy for external agents extends the same idea to **any model host** that reads the Agent Skills spec.\n\nFor explainx.ai readers, the interesting part is **structure**, not mystique about Jane Street's internal stack. The skill is **portable text** you can audit, diff in git, and reuse in [Claude Code skills folders](https://explainx.ai/blog/what-are-agent-skills-complete-guide), Quick, or custom harnesses like [Executor 2's MCP gateway](https://explainx.ai/blog/executor-2-beta-open-source-mcp-gateway-one-catalog-claude-code-cursor-codex-policies-mit-4k-stars-guide-2026).\n\n## Repository layout\n\nThe [GitHub tree](https://github.com/janestreet/agent_skills?ref=explainx) on November 10 listed:\n\n- `socratic-tutor/SKILL.md` — primary instructions\n- `CONTRIBUTING.md` ,`LICENSE.md` ,`.ocamlformat` ,`.gitignore`\n\nREADME copy: **\"Agent skills from Jane Street\"** — **\"A collection of skills for AI agents and chatbots. Each skill is a directory containing a SKILL.md file with instructions for the agent.\"**\n\nNo Python package, no npm install — intentionally **host-agnostic**.\n\n## Inside `socratic-tutor/SKILL.md`\n\nThe [published SKILL.md](https://github.com/janestreet/agent_skills/blob/master/socratic-tutor/SKILL.md?ref=explainx) opens with YAML metadata:\n\n```\nname: \"socratic-tutor\"\ndescription: \"Structured Socratic tutoring when the user explicitly asks to be taught, quizzed, or drilled to mastery on a topic...\"\n```\n\n### Guard: do not hijack casual questions\n\nThe skill mandates a **guard** before the teaching loop. If the user only asks **\"what is X\"** or **\"explain X\"**, answer normally. If intent is ambiguous, ask **one** confirmation: quick answer versus full tutoring. That prevents accidental **quiz mode** — a common failure when skills descriptions are too broad.\n\n### Session flow\n\n1. **Decompose** the topic into**4–10 subskills** (fewer for narrow subjects).\n2. Run a **3–5 question diagnostic** spanning easy to hard**before teaching** .\n3. Render an **ASCII skill map** with levels**0–5** per subskill and an overall percentage bar.\n4. Enter the **teaching loop** :**one question per message** , wait, no trailing hints.\n\nLevels are defined plainly: **0 unaware** through **5 can teach it / handle edge cases**.\n\n### Strict evaluation rules\n\nJane Street's copy is unusually explicit about **anti-sycophancy**:\n\n- **No level bumps** without**two independent checks** , one**unprompted recall** .\n- **Vague answers count as wrong** — probe until understanding is demonstrated.\n- **Spaced review** can**lower** levels if the user forgets earlier material.\n- **Level 5** requires an edge-case or adversarial question**and** teach-back quality.\n- **Do not flatter** — prefer**\"Not quite — why would X happen if Y were true?\"**\n\n### User commands\n\nThe skill documents escape hatches:\n\n- `show progress` — redraw the map\n- `harder` /`easier`\n- `just tell me` — one-question lecture override\n- `save state` /`resume from:` — portable session blobs\n\nThat last pair is valuable for **long-running mastery paths** across sessions — a pattern more skills should copy for education use cases.\n\n## How to try it today\n\n**With a SKILL.md-capable agent** (Claude Code, Cursor agents, other spec hosts):\n\n1. Clone `https://github.com/janestreet/agent_skills` .\n2. Copy `socratic-tutor/` into your host's skills directory per that product's docs.\n3. Prompt: **\"Use socratic-tutor — teach me how KV cache eviction works.\"**\n\n**With a plain chatbot:**\n\n1. Paste the **body** of`SKILL.md` after the frontmatter into a custom instruction field.\n2. Start with an explicit trigger: **\"Tutor me on backpropagation.\"**\n\nCompare results with AWS Quick's approach to **progressive disclosure** in [skills publishing](https://explainx.ai/blog/aws-quick-skills-agents-catalog-build-guide-2026): Jane Street loads everything when triggered; Quick loads metadata first. Host behavior still matters.\n\n## How this compares to other team skills repos\n\n| Repo / product | What it ships | Jane Street difference | \n|---|---|---|\n| [Tencent TeamAI-CLI](https://explainx.ai/blog/tencent-teamai-cli-git-based-team-skills-2026) | Git-synced team skills for CLI agents | Jane Street ships **pedagogy** , not CI sync | \n| AWS Quick catalog | Hundreds of business templates | Jane Street is **minimal** ,**MIT** ,**auditable** | \n| [Stack Diagram agent skill](https://explainx.ai/blog/stack-diagram-agent-skill-architecture-diagrams-github-trending-2026) | Visual architecture diagrams | Different task domain | \n\n## Star velocity and maintenance signals\n\nThe candidate feed logged **2,124 stars** with repo **created November 3, 2026** — fast uptake for a **single-skill** repo. Commit history showed **two commits** on November 10 fetch — early but not abandoned.\n\nWatch for:\n\n- Additional skills directories beyond `socratic-tutor`\n- CONTRIBUTING.md norms for external PRs\n- Whether Jane Street pins **model-specific** addenda\n\nTreat star count as **interest in agent skills as a format**, not proof the pedagogy beats human tutors.\n\n## Skill map example (what the agent should render)\n\nJane Street specifies an ASCII template the host should print after diagnostics. The pattern looks like this in `SKILL.md`:\n\n```\nTopic: overall: ######.... 58%\n-----------------------------------------------------------------\nsubskill A [########..] 4/5 * independent\nsubskill B [######....] 3/5 <- current focus\nsubskill C [##........] 1/5\n-----------------------------------------------------------------\n```\n\nThe file asks for solid block **U+2588** and light block **U+2591** characters when the environment supports them. That visual discipline matters for **long-running tutoring sessions** where users lose track of progress. It also makes **save state** blocks easier to parse when pasted into a new chat.\n\n### Teaching loop rules in plain language\n\nAfter the map appears, the agent should:\n\n- Ask **one** question, then**stop the message** — no trailing hints, no example answers in the same turn.\n- Prefer questions that force **prediction** ,**error finding** , or**transfer** to a new case.\n- On wrong answers, **narrow** the question twice before explaining.\n- On right answers, occasionally **devil's advocate** to test robustness.\n- Run **spaced review** on earlier subskills without announcing it.\n\nThe skill explicitly forbids attributing insights the user did not state. That line fights **hallucinated progress reports**, a common failure mode when models want to be encouraging.\n\n### Final assessment gate\n\nCompletion requires **every subskill** at the target level (default **4/5**, or **5/5** if the user asked for full mastery). Then the agent runs a **mixed assessment** with at least one question per subskill plus one **integrative** problem. Failures send the user back into the loop rather than issuing a participation trophy.\n\n## Porting `socratic-tutor` to your agent harness\n\n| Step | Action | Pitfall | \n|---|---|---|\n| 1 | Copy `socratic-tutor/` verbatim into your skills root | Renaming `name:` in frontmatter breaks triggers | \n| 2 | Ensure host loads **description** for routing | Broad descriptions cause false positives — this skill already narrows triggers | \n| 3 | Disable **parallel tool calls** during tutoring | Extra tools interrupt the one-question-per-turn contract | \n| 4 | Cap **max output tokens** per turn | Long lectures violate the format even if the model \"helps\" | \n| 5 | Log **level changes** externally if you grade employees | SKILL.md does not persist state unless user runs `save state` | \n\nIf you maintain skills in git, add a **CI check** that `SKILL.md` frontmatter `name` matches the directory name — the same lint AWS Quick recommends for production skills in [its November guide](https://explainx.ai/blog/aws-quick-skills-agents-catalog-build-guide-2026).\n\n## Jane Street and the wider skills arms race\n\nNovember 2026 is crowded with **skills-shaped** releases:\n\n- **AWS Quick** productized**publish/share** for business users.\n- **Jane Street** open-sourced**pedagogy** .\n- Trending repos such as [Stack Diagram](https://explainx.ai/blog/stack-diagram-agent-skill-architecture-diagrams-github-trending-2026) target**diagram generation** .\n\nJane Street's contribution is **normative text**: it tells the model how **not** to teach. That is harder to copy than a script because it depends on **turn-by-turn discipline** from the host runtime.\n\nFor education startups, the skill is a **free baseline** to compare against proprietary tutors. Run A/B tests: default system prompt versus `socratic-tutor` on the same **calculus or ML** syllabus and measure **human-rated mastery**, not token count alone.\n\n## OCaml culture and `.ocamlformat` in the repo\n\nThe repository root includes **`.ocamlformat`**, a wink at Jane Street's OCaml ecosystem, even though the skill itself is Markdown. Contributors editing future skills should keep **CONTRIBUTE.md** norms in mind if the firm accepts external skills — today the repo is tiny, but the file signals **reviewed contributions** rather than drive-by prompts.\n\n## Limitations and risks\n\n- **Token burn** : Socratic loops are long; pair with[output-token awareness](https://explainx.ai/blog/ai-labs-output-token-inflation-pricing-playcode-2026) when using reasoning models.\n- **No sandbox** : The skill assumes the host enforces tool policies; it does not ship execution sandboxing — irrelevant for pure Q&A tutoring but critical if you add code-exec exercises.\n- **Not financial advice** : Jane Street's brand does not make this a trading signal.\n\n## When not to use this skill\n\nSkip `socratic-tutor` for:\n\n- **Incident response** or**on-call** pages where the user needs an answer in one shot.\n- **Compliance answers** that must quote regulation verbatim — use retrieval with citations instead.\n- **Creative brainstorming** where leading questions constrain ideation.\n- **Children's homework without oversight** — the skill is strict, not a substitute for teacher supervision.\n\nFor **security training**, pair tutoring with factual runbooks and restrict tool access in the host so a student agent cannot exfiltrate data while \"learning.\"\n\n## Related reading\n\nJane Street agent skills library card pulled from shelf illustrating SKILL.md collections\n\n*Star counts and commit totals are accurate as of November 10, 2026 per GitHub.*", "url": "https://wpnews.pro/news/jane-street-open-sources-agent-skills-with-a-socratic-tutor-skill-md", "canonical_source": "https://explainx.ai/blog/janestreet-agent-skills-socratic-tutor-github-2026?utm_source=rss&utm_medium=feed&utm_campaign=blog", "published_at": "2026-11-10 00:00:00+00:00", "updated_at": "2026-11-10 01:51:19.007075+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "developer-tools", "agent-protocols", "artificial-intelligence"], "entities": ["Jane Street", "agent_skills", "socratic-tutor", "GitHub", "Anthropic", "Claude", "AWS Quick"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/jane-street-open-sources-agent-skills-with-a-socratic-tutor-skill-md", "markdown": "https://wpnews.pro/news/jane-street-open-sources-agent-skills-with-a-socratic-tutor-skill-md.md", "text": "https://wpnews.pro/news/jane-street-open-sources-agent-skills-with-a-socratic-tutor-skill-md.txt", "jsonld": "https://wpnews.pro/news/jane-street-open-sources-agent-skills-with-a-socratic-tutor-skill-md.jsonld"}}