Claude Skill

trim

Shorten docs, prompts, skills, or configs without losing meaning or behavior.

LLM Mart · 0 points · 0 views 0 listing impressions 0 install-command copies
Download fmind-dot-skills_trim-3bbe830.zip · 1 KB

Install

skills CLI npx skills add https://github.com/fmind/dot/tree/main/skills/trim
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install fmind-dot@llmmart
Git git clone https://github.com/fmind/dot.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole fmind/dot collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Trim

Handle /trim <path> or verbosity complaints by cutting words, lines, and settings that carry nothing while keeping every fact, behavior, and the author's voice. For deliverables (articles, talks) without a requested length, tighten sentences without changing length class; for code, full-review owns dead code and abstractions (trim code only with behavior tests).

Workflow

  1. Measure baseline: wc -lwm <path>, plus dot agent context --project . --details (add --source . in fmind/dot) for instructions and skills.
  2. Inventory survivors: list every claim, rule, number, command, flag, link, warning, and authority boundary. For configs, snapshot effective state first (mise tasks ls --json --hidden, mise ls --current --json, mise env --json, mise settings, dprint resolved-config, ruff check --show-settings <file>, docker compose config, chezmoi cat <target>, or yq -o json).
  3. Cut in order:
    • Duplication: keep once at the canonical owner; link elsewhere.
    • Restated defaults: drop settings matching the installed version's verified default (unless a comment/commit pins them intentionally).
    • Filler: intros, hedges, recaps, "note that", narration, history, and generic tool tutorials.
    • Dead entries: paths/options proven absent with rg or fd.
    • Wordy phrasing: collapse lists to clauses and drop redundant examples.
  4. Preserve: evidence-backed facts, numbers/dates, exact commands/flags, non-obvious gotchas with reasons, security/authority boundaries, links, required frontmatter/anchors/index markers, top-of-config docs URL comments, non-default rationale comments, and author voice.
  5. Prove equivalence: diff recaptured effective config (must match), walk the survival list, and run formatters, link checks, and affected tests.
  6. Report: before/after lines, words, and ceil(chars/4) tokens (%), cut categories, and ambiguous candidates held back for approval.

Gotchas

  • Delete, don't compress: never trade readable lines for clever one-liners or dense regexes.
  • Search shared strings first: rg before rewording triggers, error strings, or test-matched phrases.

Documentation

Files (dot)
  • SKILL.md 2.6 KB
    ---
    name: trim
    description: "Shorten docs, prompts, skills, or configs without losing meaning or behavior."
    license: MIT
    metadata:
      kind: task
      author: Médéric HURIER (Fmind)
      source: github.com/fmind/dot/tree/main/skills/trim
      created: "2026-10-04"
      updated: "2026-11-11"
    ---
    
    # Trim
    
    Handle `/trim <path>` or verbosity complaints by cutting words, lines, and settings that carry nothing while keeping every fact, behavior, and the author's voice. For deliverables (articles, talks) without a requested length, tighten sentences without changing length class; for code, [full-review](../full-review/SKILL.md) owns dead code and abstractions (trim code only with behavior tests).
    
    ## Workflow
    
    1. **Measure baseline**: `wc -lwm <path>`, plus `dot agent context --project . --details` (add `--source .` in `fmind/dot`) for instructions and skills.
    1. **Inventory survivors**: list every claim, rule, number, command, flag, link, warning, and authority boundary. For configs, snapshot effective state first (`mise tasks ls --json --hidden`, `mise ls --current --json`, `mise env --json`, `mise settings`, `dprint resolved-config`, `ruff check --show-settings <file>`, `docker compose config`, `chezmoi cat <target>`, or `yq -o json`).
    1. **Cut in order**:
       - **Duplication**: keep once at the canonical owner; link elsewhere.
       - **Restated defaults**: drop settings matching the installed version's verified default (unless a comment/commit pins them intentionally).
       - **Filler**: intros, hedges, recaps, "note that", narration, history, and generic tool tutorials.
       - **Dead entries**: paths/options proven absent with `rg` or `fd`.
       - **Wordy phrasing**: collapse lists to clauses and drop redundant examples.
    1. **Preserve**: evidence-backed facts, numbers/dates, exact commands/flags, non-obvious gotchas with reasons, security/authority boundaries, links, required frontmatter/anchors/index markers, top-of-config docs URL comments, non-default rationale comments, and author voice.
    1. **Prove equivalence**: diff recaptured effective config (must match), walk the survival list, and run formatters, link checks, and affected tests.
    1. **Report**: before/after lines, words, and `ceil(chars/4)` tokens (`%`), cut categories, and ambiguous candidates held back for approval.
    
    ## Gotchas
    
    - **Delete, don't compress**: never trade readable lines for clever one-liners or dense regexes.
    - **Search shared strings first**: `rg` before rewording triggers, error strings, or test-matched phrases.
    
    ## Documentation
    
    - Companions: [repository-docs](../repository-docs/SKILL.md), [skillify](../skillify/SKILL.md), [system-prompts](../agent-frameworks/SKILL.md#system-prompts).
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.