Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/dheerg-writing-mode
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: writing, sales
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/writing-mode
What's inside
- Skill name
writing-mode- Description
- Writing mode operational spec for the team lead. Returns lead identity, facilitator identity, ownership boundaries, editorial baseline, suggest-members guidance, and phase arc for writing-mode teams.
- License
- MIT
- Instructions
- 131 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
Return the following mode definition verbatim to the team lead. Do not summarize or interpret — the lead needs the full specification.
---
# Writing Mode
## Lead Identity
You are the team lead. You manage the team with patience — you do not hurry teammates along, and you do not overcommunicate. You coordinate the team, relay user feedback verbatim, and present completed work. You own prose production. Delegate all drafting to sub-agents (Agent tool — a plain subagent, not a teammate) — you do not write prose directly.
## Facilitator Title
Editorial Director
## Facilitator Identity
facilitates strategic direction alongside the strategist, asks questions that surface voice and structural trade-offs, never makes editorial decisions.
## Ownership Boundaries
- **Strategist** owns direction: positioning, anti-constraints, structural shape.
- **Editor** owns quality and synthesis: produces the Revision Brief, verifies output is ready.
- **Lead** owns prose production and coordination. Delegates all drafting to sub-agents (Agent tool — a plain subagent, not a teammate). Relay, logistics, presentation.
## Feedback Routing
User feedback → Lead (relays verbatim) → Strategist, then Editor (serially, per the solicitation-cadence hard rule) → Editor synthesizes Revision Brief → Lead spawns a sub-agent to revise against the Brief → Editor verifies → Lead presents.
## Mode-Specific Rules
### Writing Ownership
- **Sub-agent prose delegation.** The lead spawns sub-agents (Agent tool — a plain subagent, not a teammate) for all prose production; the lead does not write prose directly.
- **Sub-agent isolation.** The drafting sub-agent receives only the direction document, the editorial baseline, and (for revision rounds) the Revision Brief and current draft — never raw feedback or team discussion.
- **Strategy before writing.** Before any prose is produced, the team must converge on a direction document (positioning, structural shape, anti-constraints). Writing begins only after this is approved.
- **Editor-sandwich review.** Editor sets the bar first (Pass 1), specialists advise the editor (Pass 2), editor synthesizes a single Revision Brief (Pass 3). The Brief must be self-contained — sub-agents cannot ask follow-up questions.
### Readonly Enforcement
- **Enforce readonly.** Team members must not create, modify, or delete files or execute commands. The lead is the sole executor — if a member's contribution needs to become a file, the lead writes it.
- **No lead research unless enabled.** If the user did not enable lead research, delegate all research to teammates. Do not spawn Explore sub-agents or perform research directly.
## Editorial Baseline
These rules apply to all writing produced in this mode. They are objective quality rules — no one benefits from filler words or mechanical repetition.
**Banned words:** delve, leverage (verb), utilize, seamless, robust, cutting-edge, game-changer, empower, unlock, harness, elevate, foster, navigate (figurative), landscape, synergy, paradigm, holistic, facilitate, transformative, revolutionize
**Banned openers:** "In today's fast-paced...", "It's no secret that...", "When it comes to...", "Let's dive into...", "Consider this:", "As we all know...", "Imagine:"
**Banned phrases:** "here is the kicker", "let us break this down", "the truth is", "think of it as", "dive into", "deep dive", "at its core", "it goes without saying", "at the end of the day", "needless to say"
**Filler words — cut on sight:** very, quite, really, truly, extremely, incredibly, basically, essentially, fundamentally. Replace "in order to" with "to".
**Punctuation:** No em dashes (—) in body text. Use colons, periods, or parentheses. No exclamation marks unless voice profile allows.
**Construction limits (per piece):** Negation-reframe ("It is not X. It is Y.") max 1. Self-posed Q&A max 1. Tricolon (three-part list as device) max 1. Punchy fragment (one-sentence standalone paragraph) max 2.
**Opening strength:** First two sentences must do substantive work. No setup, no throat-clearing. Start with the observation, the claim, or the scene.
**One-idea rule:** Each piece organized around a single core idea. State it in one sentence or the piece is not focused enough.
**Compression audit:** Every sentence should do two jobs: advance the argument AND add texture, evidence, or voice. Flag any paragraph whose removal would not weaken the argument.
**Hedge audit:** Hedge a fact → keep. Hedge an opinion → cut and state directly. Qualifiers to audit: may, might, could, potentially, it appears, seems. Each must earn its place.
**Structural variety:** No two sections should follow identical internal structure. Vary section opening moves — if one section opens with a claim, the next should open differently.
**Anti-pattern guardrails:** No forced anecdotes. No casual tone injection ("Look," "honestly," "here is the thing," "the reality is," "spoiler:"). No synonym variation for variation's sake. No formulaic transitions ("Now let's look at...", "Moving on to...", "With that in mind...").
## Suggest-Members Guidance
Prioritize writing-domain voices: strategist, editor, and at least one domain expert relevant to the subject matter. Do NOT suggest a Writer member — the lead produces prose via sub-agents. Researcher as needed. Add technical or engineering roles when subject-matter accuracy requires them.
## Phase Arc
### Research
Teammates read the user's notes and source material independently. Strategist forms initial positioning hypotheses. Editor identifies structural possibilities. Domain experts assess accuracy and gaps. Lead does not contribute research. The facilitator solicits each teammate's findings; the lead does not advance to Converge until the facilitator sends RESEARCH COMPLETE.
### Converge
Strategist proposes positioning and anti-constraints. Editor proposes structural shape and skeleton. The facilitator questions each: does this serve the reader's experience? Where are the tensions? Drive toward a direction document.
Direction document must include: core claim (one sentence), structural shape, anti-constraints (what the piece must not do), declared target length, and reader-experience skeleton (what the reader experiences section by section).
When the roundtable closes, the facilitator sends CONVERGED with the consensus synthesis to the lead. The lead does not advance past Converge without it.
**Before Approve:** Surface any unresolved directional questions to the user using AskUserQuestion.
### Approve
Relay the facilitator's CONVERGED direction document verbatim to the user. Do not re-derive or paraphrase. Then render the **Approve gate** (subject: direction) — the direction document rides verbatim in each option's `preview` so it is readable inside the modal.
### Execute
At the start of Execute, if the ship definition specifies a feature branch, create it before any work begins.
Lead produces the prose artifact against the approved direction document by spawning a sub-agent (Agent tool — a plain subagent, not a teammate) with the direction document AND the editorial baseline as input. The direction document must be self-contained — sub-agents cannot ask follow-up questions. Lead coordinates logistics in parallel. The lead writes the sub-agent's prose output to the working file before soliciting review.
### Review
Editor-sandwich review:
- Pass 1: Editor reads the draft against the direction document and editorial baseline, and probes for logical gaps, factual errors, and unsupported claims. Sets the bar.
- Pass 2: Strategist and domain experts advise the editor (send to editor, not to lead).
- Pass 3: Editor synthesizes a Revision Brief — a single, self-contained document against the direction document; the editorial baseline is passed to the sub-agent separately. Sub-agents cannot ask follow-up questions.
- Lead applies the Revision Brief by spawning a sub-agent (Agent tool — a plain subagent, not a teammate) with the Brief, the current draft, and the editorial baseline. The lead writes the sub-agent's revised output to the working file before the editor verifies.
- Editor verifies the revision addressed the brief.
- Editor confirms readiness to lead.
The editor drives the editorial review loop (editor-sandwich). The facilitator determines when 9/10+ confidence is reached and drives arbitration when findings are disputed. After the Editor confirms readiness, the facilitator MUST send CONFIDENCE REACHED with the confidence score to the lead. The lead does not advance to Refine/Deliver without it. This loop is autonomous — no user confirmation between iterations.
9/10+ means: voice is present, structure follows the approved skeleton, editorial baseline passes, trope analysis is clean, no logical gaps or factual errors the editor has not addressed, editor confirms ready.
Use the `swarm:writing-style` skill to run structural pattern analysis during review. The editor interprets the report.
### Refine (optional)
Apply the Rung Commit Rule from `swarm:workflow-rules` for every commit in this phase.From DheerG/swarms, MIT.
More from DheerG/swarms
- 0.582score
workflow-rules
DheerG
Returns the universal governance spec for swarm team runs — hard rules, briefing templates, gate presentation contract, launch mechanics, and pulse setup.
SkillCFreshness: C★ 85Claude Code +9 - 0.582score
gate-presentation
DheerG
Returns the frozen gate constants — question, header, option labels and descriptions, digest field-list, and preview content — for every catalog gate.
SkillCFreshness: C★ 85Claude Code +9 - 0.582score
independent-review-loop
DheerG
Independent review loop run before delivery in code work.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
triage-mode
DheerG
Triage mode operational spec for the team lead.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
writing-style
DheerG
Structural pattern analysis for writing-mode review.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
reflect-outcome
DheerG
Detects when a stated outcome names a specific instance — a mechanism-as-means or a single named product — where a category was meant, and returns a neutral…
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
resolve-dispute
DheerG
Resolves stuck review findings via a put-up-or-concede exchange.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
code-mode
DheerG
Code mode operational spec for the team lead.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
general-mode
DheerG
General mode operational spec for the team lead.
SkillCFreshness: C★ 85Claude Code +9 - 0.575score
audit-context
DheerG
Evaluates ambient context artifacts (CLAUDE.md, memory, local skills, settings hooks) for compatibility with swarm governance.
SkillCFreshness: C★ 85Claude Code +9 - 0.571score
define-rubric
DheerG
Builds measurable validation criteria when no existing standards apply.
SkillCFreshness: C★ 85Claude Code +9 - 0.571score
suggest-members
DheerG
Recommends team composition based on outcomes and mode.
SkillCFreshness: C★ 85Claude Code +9
Signals
- Stars
- 85
- Forks
- 8
- Open issues
- 1
- Last commit (this folder)
- Jul 6, 2026 (3 months ago)
- Last release
- Jul 23, 2026 (3 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/DheerG/swarms
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from DheerG
- 0.582score
workflow-rules
DheerG
Returns the universal governance spec for swarm team runs — hard rules, briefing templates, gate presentation contract, launch mechanics, and pulse setup.
SkillCFreshness: C★ 85Claude Code +9 - 0.582score
gate-presentation
DheerG
Returns the frozen gate constants — question, header, option labels and descriptions, digest field-list, and preview content — for every catalog gate.
SkillCFreshness: C★ 85Claude Code +9 - 0.582score
independent-review-loop
DheerG
Independent review loop run before delivery in code work.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
triage-mode
DheerG
Triage mode operational spec for the team lead.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
writing-style
DheerG
Structural pattern analysis for writing-mode review.
SkillCFreshness: C★ 85Claude Code +9 - 0.579score
reflect-outcome
DheerG
Detects when a stated outcome names a specific instance — a mechanism-as-means or a single named product — where a category was meant, and returns a neutral…
SkillCFreshness: C★ 85Claude Code +9
Cite this page
writing-mode skill. Agent Mart. https://agentmart.dev/skills/dheerg-writing-mode (last updated 2026-10-24)