haowjy-creative-writing-skills

Wide-open option generation agent — spawn with `meridian spawn -a brainstormer`, passing a scoped exploration prompt and relevant story context with -f. Explores a question or angle in depth, generates options, tags speculative content, and produces a structured brainstorm report under `work/brainstorm/`. Fan out multiple brainstormers for creative breadth; each explores its assigned angle in depth. Does not commit to structural decisions — that is the outliner's job.

Character simulation for voice discovery and relationship testing — spawn with `meridian spawn -a character-sim`, specifying the character in the prompt and passing voice/state files with -f if they exist. Stays in character for freeform, unscripted conversation. Works with full character profiles or just a sketch in the prompt. Doesn't write files — the conversation itself is the output, mined later by session-miner.

Chapter fact extractor — spawn with `meridian spawn -a chronicler`, passing the chapter file and relevant existing knowledge files with -f. Reads written chapters and updates the kb with what changed in the project's factual state. Not a summary — a synthesis of what's now true that wasn't before.

Cross-project continuity checker — spawn with `meridian spawn -a continuity-checker`, passing the content to check and relevant canon files with -f. Cross-references against timeline, character state, geography, and established facts across the full project. Reports contradictions with evidence.

Adversarial draft critic — spawn with `meridian spawn -a critic`, passing the draft and a focus area in the prompt along with relevant reference files via -f. Reports findings, doesn't edit.

Autonomous drafting orchestrator that executes write/critique loops until converged. Spawn with `meridian spawn -a draft-orchestrator`, passing the scene brief, style files, and relevant context with -f. Selects style files, fans out critics with different focus areas, iterates revisions. Produces final draft + critique synthesis under work/.

Fast project explorer — spawn with `meridian spawn -a explorer`, passing the question in the prompt and optionally target files with -f. Reads files, searches content, navigates the knowledge graph, mines past conversations. Cheap and high-throughput for bulk context gathering. Reports findings, doesn't edit.

Knowledge graph maintainer — spawn with `meridian spawn -a graph-maintainer`, passing relevant kb/ files with -f or pointing to the directory. Keeps relationship maps and cross-links current across all kb/ content. Runs meridian kg for link topology, flags broken links, updates mermaid diagrams.

Autonomous knowledge maintenance orchestrator — keeps kb/ current after brainstorms, chapter drafts, and critique rounds. Spawn with `meridian spawn -a knowledge-orchestrator`, passing conversation context with --from and a description of what changed. Dispatches session-miner, chronicler, and graph-maintainer as needed. Reports what was updated.

Story structure agent — spawn with `meridian spawn -a outliner`, passing relevant story context with -f. Produces outlines at arc, chapter, and beat levels, plus mermaid diagrams for structure visualization. Output goes to `work/outline/`. For open-ended exploration and option generation, spawn the brainstormer instead.

Experiential reader-response agent — spawn with `meridian spawn -a reader-sim`, passing the draft with -f and optionally a focus spec or style references. Performs the act of reading a draft and reports the experience as it happens, structured by the four reader reward channels (transportation, aesthetic, social simulation, flow). Produces an experiential report, not a craft critique — the orchestrator or critic translates experience into craft findings. Report lands at `work/reader-reports/`.

External web researcher — spawn with `meridian spawn -a researcher` with the research question in the prompt. Produces thorough reports with trade-off analysis and source URLs.

Session knowledge extractor — spawn with `meridian spawn -a session-miner`, passing conversation context with --from or session IDs in the prompt. Pulls durable knowledge out of brainstorm and planning conversations and writes it inline to the kb with decision annotations.

User-facing story creation entry point — owns the author relationship. Understands intent, fans out brainstormers and researchers, synthesizes results, and presents options. Kicks off draft-orchestrator when direction is confirmed and knowledge-orchestrator when decisions need recording. Spawn with `meridian spawn -a story-orchestrator`, passing conversation context with --from and relevant files with -f. Never writes files directly.

Creates style reference files from sample chapters or requirements — spawn with `meridian spawn -a style-creator`, passing sample chapters, existing style files, or written requirements with -f. Produces standalone style reference files to kb/styles/ that writer agents load when drafting. Does not evaluate prose against styles — that is the critic agent with a voice focus.

Wiki page editor — spawn with `meridian spawn -a wiki-editor`, passing the topic and relevant source material with -f. Creates and updates polished, reader-facing reference pages in wiki/. Maintains link discipline and mermaid relationship diagrams.

Prose writer — spawn with `meridian spawn -a writer`, passing scene brief, style files, and context with -f. Drafts fiction from briefs following project voice and conventions. Output is draft text subject to critique and revision.

Story brainstorming capture — minimal notes that preserve creative freedom. Use when exploring narrative ideas, discussing characters, planning chapters, or thinking through story possibilities. Supports interactive conversation and autonomous report mode for fan-out exploration.

Shared coordination model for writing orchestrators — delegation discipline, convergence loops, critique synthesis, and artifact persistence. Load when you coordinate work across multiple spawns rather than producing content directly.

Adversarial reading methodology for narrative fiction — find what doesn't work, not confirm what does. Focus-area driven with dedicated resources per area. Use when reviewing drafts, evaluating prose quality, or assessing changes at any stage.

Prose drafting technique for narrative fiction. Use when writing new scenes, chapters, or dialogue — whether in conversation with an author or producing a draft autonomously from a brief. Covers craft fundamentals; project-specific voice comes from style files passed alongside this skill.

Arc structure, narrative design, and pacing at multiple scales — saga, arc, chapter, scene. Use when structuring story at any level, planning arcs, designing chapter outlines, or evaluating whether narrative structure serves the story's goals. Not prescriptive about methodology.

Context scoping for writing agent spawns — use when deciding what context a spawned agent should receive, whether ephemeral story decisions should be materialized before handoff, and how much to pass. Poor context handoffs cause writers to invent contradictions and critics to miss relevant history.

Story decision capture and mining — recording what was decided, what was rejected, and why, inline with the artifacts they relate to. Use whenever story direction is being chosen, brainstorm options are being narrowed, character or world choices are being made, or past decisions need to be recovered from session history.

Encyclopedic reference documentation for fictional worlds — wiki pages, character profiles, lore entries. Use when creating or updating reader-facing reference pages. Covers link discipline, citation conventions, and mermaid relationship diagrams. Distinct from the knowledge base, which is author/agent-facing knowledge.

Shared artifact convention between orchestrators — what goes where in the kb and work directories, how artifacts flow between phases, and what each directory means. Use whenever work artifacts, style files, knowledge entries, drafts, or critique reports are being created, referenced, or discussed.

Logging and referencing writing issues — craft problems, tics, inconsistencies, and structural concerns found during analysis, critique, or review. Use when an agent identifies something worth tracking beyond a single critique report: repeated tics across chapters, inconsistencies that affect multiple scenes, structural problems that need the author's attention, or patterns that should be fixed in revision.

What fiction readers actually want, framed as four composable reward channels (transportation, aesthetic, social simulation, flow), and the specific documented ways alignment training damages each one. Grounded in reader-psychology research and empirical NLP findings. Load when drafting prose, critiquing a draft, deciding whether to show or tell, diagnosing why a passage feels flat, or reasoning about why a scene is or isn't working.

Team composition for writing workflows — which agents to spawn, how many, what focus areas to assign, and how to scale effort. Use when composing critic panels, dispatching researchers, staffing draft/revise loops, or setting up brainstorm fan-outs.