adopt
Take a folder that already holds research — papers, notes, code, half-written drafts — into MAGI, without breaking what is…
Answer a question from this project's own knowledge — retrieved, read and cited, never from memory.
Claude Code
$ npx -y skills add misaka16384/magi --skill ask --agent claude-code
Codex
$ npx skills add misaka16384/magi --skill ask -a codex -y
Cursor
$ npx skills add misaka16384/magi --skill ask -a cursor -y
Gemini CLI
$ npx skills add misaka16384/magi --skill ask -a gemini-cli -y
GitHub Copilot
$ npx skills add misaka16384/magi --skill ask -a github-copilot -y
OpenCode
$ npx skills add misaka16384/magi --skill ask -a opencode -y
Windsurf
$ npx skills add misaka16384/magi --skill ask -a windsurf -y
name: ask description: "Answer a question from this project's own knowledge — retrieved, read and cited, never from memory." commands: ask: "Answer a question from this project's own sources, with citations." origin: magi
Any question about what this project knows. Also when you are about to answer one from memory.
1. `magi search "<question>"` — hybrid retrieval, the default route. Add `--line <line>` when the question comes from inside a research line: it ranks what that line is looking at higher without hiding the rest. 2. **Read the files it returns.** A search result is a pointer, not a source; citing a snippet you did not open is how a half-sentence becomes a claim. 3. About one concept: `magi wiki context --name "<concept>"` gives every paragraph that mentions it, which beats reassembling the card from search hits. 4. Structural questions — what links to what, which sources back a claim, what is unlinked: `magi graph query "<SQL>"`, or `magi graph browse` for the same without SQL. Exact strings: `magi grep "<text>" wiki/`. 5. Answer with inline `[[wikilinks]]` to the cards you actually read. Where the project does not know, say that instead of filling the gap. 6. A question the project cannot answer but should be able to is a gap worth keeping: one line in `inbox/notes.md`, or open it as a question with `magi thread new <slug> --kind question --title '<the question>' --purpose '<why it matters now>'`.
answer is "this project does not have that", which is useful and true.
file it was compiled from. A card can be wrong in the way the claim is trying to rule out.
says so, and `magi index` fixes it.
MAGI 是一个 agent-native 的科研工作环境:人是驾驶员,LLM agent 是机体,确定性的 magi CLI 是拘束具——三者同步率越高,科研越快。它把学术论文(PDF/LaTeX)摄入、编译为 Obsidian 兼容的概念卡片项目,并用三核架构管理完整的科研状态: 进入任意项目先跑 magi sync——它输出同步率、三核状态和逐条可执行的修复提示。magi radar 是文献雷达:定时发现相关新论文,并侦察"知识上应引用我方论文却未引用"的候选。同一份 skills 通吃 Claude
Take a folder that already holds research — papers, notes, code, half-written drafts — into MAGI, without breaking what is…
Write the one thing a person reads after an unattended run — its report — and the lecture notes that prepare them for it, in…
Compile raw sources into dense, interlinked reference and concept cards, and mine the concepts a compiled card left implicit.
Talk a run over with the person until it can be written down — motivation, directions, what is not worth it, when to stop, their…
Turn whatever the human has — a PDF, a link, a DOI, a screenshot of a citation — into raw/ sources this project can compile.
Entry point for a MAGI research project: run magi next, do what it says, and call the skill it names.
Use this skill to extract structured Markdown/JSON from PDFs and document images—tables with cell-level precision, formulas as…
Estimates intracellular metabolic fluxes from steady-state carbon-13 isotope-tracing measurements using validated atom maps,…