Cite

SkillDocs & knowledge

Cite turns legal research into clear, actionable answers. Once added, your AI can pull together case law, statutes, and regulatory guidance and explain what they mean for your company, with risk assessments sized to your stage of business.

Use Cite in Claude, ChatGPT or Ahel Desktop

Free. Sign in, add Cite and connect your AI. About a minute.

Also: Claude Code · Cursor · Codex

Then ask your AI: use the Cite skill

Details

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

After adding Cite, ask your AI something like "research this legal question" when you need a jurisdiction comparison, a compliance check, or a plain-language legal summary.

What your AI can do with it

  • Research a legal question using case law, statutes, and regulatory guidance
  • Compare how rules differ across jurisdictions
  • Run regulatory compliance checks
  • Deliver plain-language legal summaries
  • Assess legal risk sized to your company's stage

What this skill tells your AI

The instructions your AI receives, as published by a554b554/reactant in skills/cite/SKILL.md and read by Ahel’s review.

Activation Boundary

This is a tag-only worker skill. Use it only after execute or execute-one selects an exact <@cite...> annotation in a target document and hands off its context. Never activate it from ordinary chat, semantic similarity, or any skill other than those two routers. Without that handoff, do not search or change files; direct the user to add the tag and run a router.

Find a reference that supports the claim next to the tag, insert a citation, then consume the tag.

Tag

<@cite>
<@cite: source hint or search instruction>

Examples: <@cite: search arxiv>, <@cite: ``refs.bib``>, <@cite: the original transformer paper>.

Input

  • File path — the document being processed.
  • Surrounding block — the maximal run of non-blank lines containing the tag; blank lines bound it. The claim to support is the statement the tag is attached to.
  • Prompt (optional) — where or how to find the reference: a bibliography file, an arXiv or web search, or a description of the intended source.

Workflow

  1. Read the surrounding block and identify the claim that needs support.
  2. Determine the citation source: a referenced .bib/.md file, an arXiv or web search using whatever search capability the host provides, or a specific work named in the prompt. Prefer peer-reviewed or well-cited sources when searching.
  3. Determine the citation format from the document itself: LaTeX uses \cite{key}; Markdown typically uses [number] or [@key]. Match the style already used in the document.
  4. Insert the citation at the tag's position and remove the tag.
  5. In a LaTeX project, when citing a work not already in the bibliography, add a BibTeX entry to the referenced or main .bib file. Do not duplicate an existing entry.

Context References

Double-backtick references in the prompt usually name the citation source (for example <@cite: ``refs.bib``>). Resolve paths relative to the document's directory unless absolute. Remove the delimiters from the output.

Rules

  • Insert citation material only — do not rewrite the surrounding prose.
  • Consume this annotation and its reference delimiters. Preserve every unrelated annotation, <@comment: ...> tag, <@chat-model> block, and contextual literal/example byte-for-byte.
  • If no credible source can be found, replace the tag with [citation needed] and warn the user; that is a completed no-result, not a failure.
  • If the host has no search capability and the prompt requires one, fail cleanly: leave the tag, change nothing, report the missing capability.
  • Reactant adds no verification layer on top of this protocol; the citation is whatever this workflow produces.

Signals

GitHub stars
54
Forks
11
Last commit
Sep 2026
Advanced
Item type
skill
Key
cite
Source
github.com/a554b554/reactant