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.
npx agentmods add skills/chenpg2/kw-engine/kw-explorenpx skills add chenpg2/kw-engine --skill kw-exploregit clone --depth 1 https://github.com/chenpg2/kw-engineWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00121 | $0.01465 |
| Opus 5 | $0.00060 | $0.00732 |
| Sonnet 5 | $0.00024 | $0.00293 |
| Haiku 4.5 | $0.00012 | $0.00146 |
Grade A, and why
kw-explore 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 2d 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.
How it starts
The opening of the file, as written. The whole thing — 103 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/kw-explore — drive a research project with the principle library
This is the problem-driven loop. (/kw is the corpus-driven loop: absorb papers, synthesize
globally. /kw-explore starts from the user's actual problem and pulls the library toward solving
it.) It is the workflow that turns the engine from a knowledge library into a research accelerator.
You are the orchestrator. Dispatch the kw-* subagents for reading/distilling (model boundary
below); do the framing/mapping/assembly reasoning yourself.
Separation of concerns (hard rule). Use-case outputs — the problem decomposition, the design,
the handoff — go in the user's project (cwd). Reusable principles go into the linked
knowledge base (kw status shows which one). Never mix: the knowledge base stays domain-general
and reusable across projects.
Model boundary (same as /kw). The orchestrator never reads PDFs or calls read_*/search_*/
download_* MCP tools. Acquisition → kw-fetcher (sonnet); reading → kw-reader (sonnet);
distilling → kw-distiller (opus). Pass model explicitly on every Agent dispatch.
Prerequisite: the project must be linked to a knowledge base. If kw status errors, run
kw link <name-or-path> (or /kw-init) first.
Phase 0 — Frame the problem (the most important step)
A good decomposition is most of the value. Do not skip to fetching papers.
If a Question Card already exists (<project>/problems/<slug>/QUESTION.md, produced by
/kw-question): the question's worth is already validated — do NOT re-litigate stakes/
falsifiability. Read the card and go straight to extracting the structural signature (below).
If the user only has a vague idea and no card, and it's unclear whether the question is even
worth pursuing, suggest running /kw-question first.
- Interview the user just enough to extract a problem signature: the structural axes of their problem, pushed toward the form the library indexes by — structural properties, not domain narrative. (E.g. not "vaginal microbiome typing" but "compositional/simplex-constrained data", "continuous spectrum + possible discrete sub-structure", "cross-population transferability", "must encode temporal dynamics".)
- Challenge the framing. Find the load-bearing structural difficulty, not the surface complaint. (The CST example: the real defect was "resolution too low — classifying in a space where the signal has already been integrated out", not "the clusters are wrong".)
- Write
<project>/problems/<slug>/PROBLEM.md: the signature axes, plus what any solution MUST satisfy (the constraints — e.g. "must work on both targeted-amplicon AND shotgun data"). - Show the decomposition to the user and confirm it before proceeding. A wrong signature wastes the whole loop downstream.
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.
- 2d ago First seen · 103 lines · 0 tokens per session scan A f61372862f88
kw-explore is a skill published in the GitHub repository chenpg2/kw-engine (11 stars, last pushed 2mo ago), licensed MIT. It adds 121 tokens to every session and 1,465 once invoked, about $0.0006 per session on Opus 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.
Other skills, from other repositories
paper-deep-reader
Skill "paper-deep-reader" from Linwei-Chen/paper-deep-reader-skill, covering paper deep reader, 1. 路由交付模式, 2. 按需加载参考材料, 3. 先读懂，再写报告 and 4. 强制执行双层讲解.
Datacore Methodology
Core knowledge management principles — capture, journal, learn, compound.
PBE Extractor
Extract invariant principles from any text — find the ideas that survive rephrasing.
flux-analyzer
Analyse FBA flux distributions to extract biological insights. Covers gene essentiality, phenotypic phase planes, flux sampling, pathway-level aggregation, secretion product prediction, and production of publication- quality figures.
a-evolve
Apply A-Evolve's agentic evolution methodology to improve AI agent performance across runs. Use when the user wants to diagnose agent failures, generate targeted skills from error patterns, evolve system prompts, or accumulate episodic knowledge. Works standalone or inside AutoResearchClaw pipelines. Triggers on…
statistical-method-design
Design statistical methods, baselines, diagnostics, variants, and ablations that directly address a formal problem formulation.