For AI agents#

acquaint ships artifacts for coding agents alongside its code. This page lists them, says where each lives in the repository, and points at the machine-readable copies of this documentation.

Skills#

Skills are folders holding a SKILL.md (the Agent Skills format): a description that tells an agent when to use it and a body with the procedure. Install one into your agent with gh skill (any host: --agent claude-code, copilot, cursor, codex, gemini), or use the copy bundled in the wheel.

acquaint#

Who someone is, how to reach them, and how to write to them, from the operator’s local people records (acquaint). Use whenever a person, project, org or group is named in a task — to get a handle, email or alias; to check whether two names are the same person before writing them as two; before drafting anything a specific person will read; and to record something learned about someone, with its source. Triggers on “who is X”, “what’s X’s email/handle/GitHub”, “is X the same person as Y”, “how do I reach X”, “remember that X…”, “add X to my profiles”, “brief me on X”, and on any name in prose about to be published. Routes heavier work to acquaint-write, acquaint-read, acquaint-profile, acquaint-sync, acquaint-meeting-todos and deslop. Query one record; never read the whole store.

gh skill install thorwhalen/acquaint acquaint --agent claude-code

Source: acquaint/data/skills/acquaint (bundled with the pip package).

acquaint-meeting-todos#

Turn the todos from a meeting into GitHub issues for a company or project the operator works on — one manual-task issue per participant in the project’s private meta repository, items copied verbatim, each linked to the notes and to the issue or page it concerns. Use when given meeting notes, minutes, a transcript, a Slack thread or an email with “next steps”, “action items” or “todos” and asked to “put these in issues”, “file the meeting todos”, “make issues from the next steps”, “assign the action items”, or “track what everyone said they’d do”. Checks first, through acquaint, that the project is unambiguous, that every participant is registered and belongs to it, and that its declared meta repository is private; when any of these fails, it tells the user what is missing and how to fix it, and files nothing.

gh skill install thorwhalen/acquaint acquaint-meeting-todos --agent claude-code

Source: acquaint/data/skills/acquaint-meeting-todos (bundled with the pip package).

acquaint-profile#

Build, research, update or consolidate a profile in acquaint — gather a person’s own writing, extract observable communication habits with evidence, and write sourced lines into their entry file, writing card and views. Use when asked to “make a profile of X”, “research how X writes”, “update X’s profile”, “what do we know about how X communicates”, “consolidate X’s observations”, “turn these notes into a profile”, or before a high-stakes message to someone whose record is thin. Runs parallel profile-reader agents on one extraction brief. Records behaviour with sources, never personality labels or sensitive categories.

gh skill install thorwhalen/acquaint acquaint-profile --agent claude-code

Source: acquaint/data/skills/acquaint-profile (bundled with the pip package).

acquaint-read#

Interpret a message from a specific person using their acquaint record — what they are saying and asking, what they likely mean and how strongly, how it differs from their usual, and what to do next — and judge, with evidence and never as an accusation, whether it was AI-processed. Use when asked “what does X mean by this”, “read this message from X”, “how should I take X’s reply”, “is X unhappy with this”, “did X write this with AI”, or when a reply from a known person arrives during a task. Records anything newly learned with acquaint remember, sourced.

gh skill install thorwhalen/acquaint acquaint-read --agent claude-code

Source: acquaint/data/skills/acquaint-read (bundled with the pip package).

acquaint-sync#

Set up, use and check private sync of the acquaint profile store through a private GitHub repository. Use when asked to “sync my profiles”, “back up my acquaint store”, “set up profiles on the server”, “use my profiles on another machine”, “is my profile store private”, “push/pull profiles”, or when acquaint sync status reports a problem. Covers what the visibility guard does and, plainly, what it does not protect, plus the git-remote-gcrypt upgrade.

gh skill install thorwhalen/acquaint acquaint-sync --agent claude-code

Source: acquaint/data/skills/acquaint-sync (bundled with the pip package).

acquaint-write#

Write, edit or review anything a specific person will read — an email, a reply, a chat message, an issue or PR comment, a pitch, a document — using that person’s acquaint record; and spar with a simulated version of them to stress-test a draft or an idea. Use when asked to “write to X”, “reply to X”, “draft this for X”, “how should I put this to X”, “would X like this”, “red-team this against X”, “argue with me as X”, or whenever the reader of a draft is a known person. Runs acquaint brief, check and style-lint, applies deslop, and keeps the operator’s voice and positions.

gh skill install thorwhalen/acquaint acquaint-write --agent claude-code

Source: acquaint/data/skills/acquaint-write (bundled with the pip package).

deslop#

Write and edit prose that carries no machine-writing tells and sounds like its author, calibrated to the reader. Use whenever drafting or editing text a person will read — emails, messages, READMEs, docs, posts, pitches, reports — and whenever the user mentions AI slop, AI tells, “sounds like AI”, “make it sound human”, “my voice”, or asks to de-AI a draft; also whenever another skill asks for it (acquaint-write does). Runs the deterministic acquaint style-lint check at the reader’s tolerance of AI-sounding text, then judgment passes for rhythm, specificity and ownership. Never fakes humanity and never invents specifics.

gh skill install thorwhalen/acquaint deslop --agent claude-code

Source: acquaint/data/skills/deslop (bundled with the pip package).

The bundled skills are also on disk after pip install acquaint, under the package’s data/skills/ directory; link them into an agent without network access with skill link-skills <that directory>.

Subagents#

Subagents are Markdown files with a frontmatter (name, description, tools) and a system prompt as the body. Copy one into your project’s .claude/agents/ (or your agent host’s equivalent) to delegate the task it describes.

profile-reader#

Parallel extraction worker for the acquaint-profile skill. Reads one batch of a person’s own writing and returns evidence rows (dimension, observation, short quote, anchor, date) exactly as the skill’s extraction brief specifies — never a summary, never personality labels or sensitive categories. Spawn several at once, one batch each.

Source: acquaint/data/agents/profile-reader.md; tools: Read, Grep, Glob, WebFetch.

recipient-reader#

Simulates how a specific, known person is likely to read a draft or answer an idea, grounded only in the acquaint brief it is handed. Used by the acquaint-write skill for its red-team check (a report on a draft, without rewriting it) and for sparring (arguing from the person’s documented positions). Always presented as a simulation; every point cites the brief line it rests on.

Source: acquaint/data/agents/recipient-reader.md; tools: Read.

Instruction files#

Files agents read before working in this repository.

Machine-readable documentation#

This site publishes the same documentation in forms that fit an agent’s context window:

  • llms.txt: an index of every page with a one-line description (llms.txt format)

  • acquaint.md: the whole documentation as one Markdown file

  • <page>.html.md: a rendered Markdown twin of every page, advertised from each page’s <head> with <link rel="alternate" type="text/markdown">

  • objects.inv: the Sphinx inventory: a symbol-to-URL index (sphobjinv convert plain objects.inv -)