Content & Media Creation · mattpocock/skills
writing-fragments
Writing, explore: mine raw fragments, no structure yet.
L2 Local · Restricted workspaceV1 Commit-PinnedWrite FileDual-pass Audit
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.
1How It Works & Mechanism
Positioned strictly as the explore stage: no commitment to structure, no outline building. Run a fragment-generating 'grilling' session to maximize the exploratory space.
skills/in-progress/writing-fragments/SKILL.md
This is pure **explore**: widen the space of what could be written without committing to structure. Committing is _exploit_, a separate skill's job. Run a grilling session that produces fragments, interviewing the user relentlessly about whatever they want to write about. Imposing phases, outlines, or article structure is out of scope here.
The explore/exploit distinction is the organizing principle of the repository's writing series: all structural commitments are forbidden during the fragments stage.
Capture discipline: even the user's initial prompt is recorded as the first batch of fragments; fragments are appended to a single markdown file, with the first write containing only an H1 working title, without metadata, TOC, or dates.
skills/in-progress/writing-fragments/SKILL.md
Capture fragments from the very first thing the user says, including the initial prompt. On first write, put a single H1 at the top with a working title (it can change later) and nothing else: no metadata, no TOC, no date.
Format is discipline: 'Fragments are separated by a horizontal rule (`\n---\n`). No headings inside the body. No tags. No order beyond the order they were added.'
Fragment quality benchmark = readable by the author, no need for cold readers to understand: 'is this a piece of good writing?' rather than 'is this a self-contained argument?'.
skills/in-progress/writing-fragments/SKILL.md
It must be _readable by the author_ (the author can tell what it means), but it does not need to define its terms or be comprehensible to a cold reader. The bar is "is this a piece of good writing?", not "is this a self-contained argument?"
Allows half-thoughts, vignettes, quotes, sharp one-liners, and other heterogeneous fragments; fragments are deliberately heterogeneous.
Leading words are marked as the most valuable fragment type: when dialogue circles around recurring ideas, actively coin a word or metaphor naming that idea to lay the foundation for subsequent exploit structures, transitions, and titles.
skills/in-progress/writing-fragments/SKILL.md
Of these, the leading word is the most valuable fragment to land. It is load-bearing: name the right one in explore and it shapes the structure, the transitions, and the title later, paying dividends through the entire exploit phase. When the conversation circles a recurring idea, push to coin a word for it.
Exemplified by terms like _tracer bullets_ / _fog of war_ that name an entire pattern—a methodological preference of the author.
Pacing discipline: silently append without asking permission per entry; reread from disk before each write and never overwrite entirely, only append or modify a specific entry in place upon user request.
skills/in-progress/writing-fragments/SKILL.md
Append silently. Don't ask permission for each fragment. Mention what you added in passing ("adding that"), but don't interrupt the conversation with save dialogs. Before every write: re-read the file from disk. The user may have edited, reordered, or deleted fragments between turns, so preserve their changes. Never overwrite the file; only append (or, if the user asks, edit a specific fragment in place).
Dialogue fluidity comes first: write actions are minimized to lowest friction; file state always reflects disk.
2Core Capabilities
01Mine writing fragments from the user (including the initial prompt) through grilling sessions
02Coin leading words: name recurring ideas with an overarching metaphor or neologism
03Single-file fragment repository maintenance: H1 title + --- dividers, no body headings, tags, or metadata
04Fragment-level editing commands: cut the last one / rewrite sharper / merge those two
05Cross-session target path memory: ask once if no path is provided and remember it
4Recommended Governance & Evidence
Suggested Controls for Safe Operation
For Agent Runtime
Restrict working directory
For Administrators
Retain operation logs
Evidence & Triggers (File Paths & Excerpts)
- 本地工作目录: File I/O and artifact output: A single markdown file specified by the user (or remembered after asking once): append-only, or edit/delete specific fragments in place upon user request
Audit Notes: Local file operations only: appends fragments to a single user-specified file (explicitly 'Never overwrite the file; only append'); additions, edits, and deletions only occur on fragments explicitly requested by the user ('cut the last one', etc., as first-class commands). No scripts, no external network requests, no credential reading. Note: write path must be specified by the user; content consists of text generated by the model from dialogue, and the user should be aware that the file will be continuously appended to.
5Second-Pass Verification
- [ok] grilling 会话引擎声明 — 'Run a grilling session that produces fragments' 原文在；调用 model-invoked 原语方向合规。
- [ok] 首次 prompt 也捕获 — 'Capture fragments from the very first thing the user says, including the initial prompt.' 原文在。
- [ok] H1 + --- 文件格式纪律 — 'put a single H1 at the top with a working title…nothing else'、'Fragments are separated by a horizontal rule' 原文在。
- [ok] 只追加不覆盖 + 写前重读磁盘 — 'Never overwrite the file; only append'、're-read the file from disk' 原文在。
- [ok] leading word 铸造指令 — 'the leading word is the most valuable fragment to land…push to coin a word for it' 原文在。
- [ok] 文件写入范围 — 仅用户指定单文件；'If the user did not pass a path, ask once where to save the document' 无其它路径。
- [ok] 外部依赖为空 — token 扫描零命中；grilling 是仓库内 skill 引用。
- [ok] 元数据 + beta 状态 — GitHub API MIT/257659/2026-09-04；本地 HEAD == 3cca18b；in-progress README 列为 beta。
6Verdict & Scenarios
Best for: Ideal for pre-writing material prospecting: authors with vague topics in mind who want to accumulate raw materials while conversing; also fits crystallizing daily noticings into a mineable fragment repository. Works best when chained with writing-shape/writing-beats (fragments → material → exploit).
Not for: Not suitable for users looking for an immediate full draft or outline; not suitable when materials are already complete and only organization is needed (should proceed directly to exploit); not suitable for cases requiring a cleanly structured, readable file (fragment files are deliberately unorganized).
Install Agent-Ready
① Verified Mirror 2026-09-09
Option A · Direct DownloadDownload writing-fragments.tar.gz
sha256:
Option B · Copy Guide for Agentca6d39f53627a5dd…Agents automatically download and verify mirrors from this JSON guide.
② Upstream GitHub
GitHub Repo ↗
Pinned to commit 3cca18b368.