writing-principles
What fiction readers want (four reward channels) and the specific ways LLM training damages each one. Load when drafting prose, critiquing, or diagnosing why a passage feels flat.
What this skill does
# Writing Principles Load `/llm-writing` if it isn't already loaded. This skill adds the fiction-specific layer. ## Trust the Reader The reader is an active collaborator. They reconstruct emotions from behavior, infer motives from action, hold tension across scenes, fill gaps the text leaves open, and make assumptions about what's coming next. That work is where the reward lives: reconstruction, inference, anticipation. Your training pulls in the opposite direction. The helpfulness instinct wants to explain, resolve, clarify, and complete. In fiction, every one of those impulses can damage the reading experience by doing work the reader wanted to do themselves. The specific failure modes below are all forms of this: not trusting the reader to interpret an emotion, hold an ambiguity, follow subtext, or tolerate unresolved tension. Trust doesn't mean obscurity. Readers also need coherent narrative, stable geography, and enough access to model characters. The discipline is knowing when to leave space and when to orient. ## Economy Every element does more than one thing. A line of dialogue advances plot AND reveals character. A sensory detail grounds the scene AND shows who the POV character is. A transition compresses time AND carries an emotional beat. Single-purpose prose makes fiction go flat: description that only describes, dialogue that only informs, interiority that only labels. Economy isn't minimalism. Dense, lyrical prose can be economical when every phrase carries weight. Sparse prose can be wasteful when it takes ten short sentences to do what one image could do. The measure is whether removing the element would cost the reader something. The LLM pull is toward completeness: covering every beat, naming every emotion, resolving every ambiguity. Economy is the counter-discipline: what can you leave out and still have the scene work? What's the reader already doing for you? ## Four Reward Channels Readers enjoy fiction through four separable channels. Good prose protects all four at once; damaging any one damages the reading experience. - **Transportation**: entering the story world. Protected by coherent narrative progression, consistent POV, concrete sensory grounding. Consistent POV means writing from inside the character's knowledge state: what have they experienced, what do they actually know right now, what would they notice and miss? The full story is in your context window; the character only has what they've lived through. Separate those. - **Aesthetic**: sentence-level pleasure. Protected by variety in rhythm, word choice, sentence shape, and punctuation. Style is a reward channel, not decoration. - **Social simulation**: modeling characters as minds. Protected by access through behavior and interiority, distinct voices, emotion the reader interprets rather than being told. - **Flow**: readable challenge. Protected by pacing that matches the scene's work, sentences that support comprehension. The channels compose: optimizing one at the expense of others fails. Over-explaining breaks social simulation. Under-explaining breaks transportation. Generic style breaks aesthetic pleasure. Impenetrable style breaks flow. ## Punctuation Tells Readers increasingly associate em dashes with AI-generated prose. Default to punctuation that leaves less visible AI residue: sentence breaks, commas, colons, semicolons, parentheses, or dialogue beats. Rewrite the sentence around the actual relationship between clauses instead of substituting a hyphen. Use dashes only when a project style file or author instruction makes them part of the voice. When a line needs interruption, prefer the project's documented interruption pattern and keep it consistent. ## Applying the Principles The craft skills carry the execution: `/prose-writing` for immersion patterns (psychic distance, rhythm, sensory grounding, interiority) and `/scene-construction` for how scenes work on the page (entry, dialogue, pacing, transitions). This skill's job is the diagnostic layer. When a passage feels off and you can't name why, check the four channels: which one broke? Then see `resources/failure-modes.md` for common patterns and fix heuristics. ## Resources - [`resources/failure-modes.md`](resources/failure-modes.md): per-pattern deep dives with examples and fix heuristics. - [`resources/citations.md`](resources/citations.md): research backing for the four-channel model and documented failure modes.
Related in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.