zoheth/theory
skills/theory/SKILL.md
Explain what's behind code — the understanding that lived in the author's head but doesn't show in the code itself. Inspired by Naur's "Programming as Theory Building". Use when the user says "explain this in non-code terms", "what's the theory here", "这段代码背后的东西是什么", or invokes /theory explicitly. Works only from code already read in this conversation; does not explore the codebase.
$ install --global
skillsauthnpx skillsauth add zoheth/vidya theoryInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 12, 2026, 4:41 AM186.5s1 file scanned
SKILL.md
- name:
- theory
- description:
- Explain what's behind code — the understanding that lived in the author's head but doesn't show in the code itself. Inspired by Naur's "Programming as Theory Building". Use when the user says "explain this in non-code terms", "what's the theory here", "这段代码背后的东西是什么", or invokes /theory explicitly. Works only from code already read in this conversation; does not explore the codebase.
Always output in Chinese (中文).
The goal: give the user what the code alone cannot — the understanding in the author's head. Naur called this the program's theory. Use his idea as a compass, not a template: the list below is what counts as "behind the code", not a form to fill in. Organize the answer however this particular code demands; lead with whatever matters most here.
Do not restate what the code does. The code is already readable; restatement is noise.
Scope: only use code already read or discussed in this conversation. Do
not re-open source files. Exception: git history (git log, git blame) is
allowed — commit messages are the author's recorded intent.
What counts as "behind the code" — pick what's real for this code, skip the rest:
- The real-world affair or constraint the code answers to, and what it deliberately ignores.
- Why it is this way and not another — especially the strange or roundabout places, and the off-page constraint behind them.
- The load-bearing assumptions nobody wrote down.
- What the structure is preparing for: which future changes it welcomes, which it would resist. Concrete examples, specific to this code.
- Where it came from: the earlier shape it grew out of, the scars left by past decisions.
Honesty:
- Mark each claim about the author's reasoning: backed by evidence (comments, commits, structure), inferred from constraints, or your own guess. Don't state a guess in the tone of a fact.
- Mark the gaps, in two kinds. Closable: knowable by reading more of the repo — name the file. Theory boundary: only the original author could answer; say so plainly.
Expand where needed — on trade-offs, constraints, and judgment calls, never on restatement. Better to say more than to flatten a key decision into one empty sentence.
Related Skills
zoheth/socratic-paper-reading
development
VerifiedTrustedCommunity
Co-read research papers with the user using a Socratic, multi-pass methodology. The agent handles all mechanical work — extracting structure, looking up terms, tracing references, generating probing questions, maintaining layered notes — while the user retains all interpretive and critical work (understanding, judgment, "if I were writing this..."). Trigger this skill whenever the user shares a research paper (PDF, arXiv link/ID, or paper title) and signals they want to engage with it deeply — phrases like "help me read this paper", "let's go through this paper", "walk me through [paper]", "I want to understand [paper]", or simply uploads a paper without specifying what they want. Especially well-suited to AI infrastructure, reinforcement learning, and embodied intelligence papers, but the methodology generalizes. Do NOT trigger when the user clearly only wants a one-shot summary or has a single specific factual question about a paper — this skill is for sustained co-reading sessions, not quick lookups.
4SKILL.mdUpdated May 11, 2026
zoheth/code-theory-reconstruction
development
VerifiedTrustedCommunity
Use this skill when the user wants to genuinely understand unfamiliar code in any of three modes — **orienting** (building a working theory of a codebase, library, project, commit, or PR), **debugging** (tracing a bug or unexpected behavior through unfamiliar code), or **extending** (planning a modification, feature addition, or refactor in code they don't fully own yet). Trigger phrases include "help me understand this code", "walk me through this codebase", "why does this commit do X", "something's broken in this module", "I need to add X to this library", "help me figure out where this bug lives", "explain the design of this library", and similar. **The user's goal is NOT a code summary — it's to grow a working theory in their own head, structured both as an adjudicated set of claims AND as a felt sense of the system's overall shape.** Trigger any time the user wants to "understand", "figure out", "debug", "fix", "extend", "modify", "trace", or "make sense of" some code, project, commit, PR, or bug — even when they don't say "theory". Do NOT use for queries answerable by a single docstring or README line.
4SKILL.mdUpdated Apr 23, 2026
zoheth/example-skill
tools
VerifiedTrustedCommunity
Describe what this skill does, when it should be used, and the kinds of user requests that should trigger it.
3SKILL.mdUpdated Apr 17, 2026
zoheth/github-qa-extractor
development
VerifiedTrustedCommunity
Extract important questions from GitHub repositories, including issues, pull requests, discussions, and code reviews, and generate Markdown question cards for deep study. Use this skill when the user wants to extract key questions from a repo, mine important technical problems from GitHub threads, or build a study set of high-value questions from open-source projects.
1SKILL.mdUpdated Apr 16, 2026