cold-read
Load for /vellum:cold-read, after every major editorial program, and before export.
Skill by FiredMosquito831
Chalk Labs
Need help deploying cold-read?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
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/firedmosquito831-cold-read
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
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 and code. Checked Nov 2, 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/cold-read
What's inside
- Skill name
cold-read- Description
- Load for /vellum:cold-read, after every major editorial program, and before export. The cold-read skill owns the cover-to-cover cold-read protocol: the reader charter, the rolling ledger, the append-only issue log, and the batching/orchestration model. It finds the accumulating-memory defect classes (clock drift, cross-chapter state contradictions, prop custody, promise bookkeeping, knowledge-state slips) that no scanner, chapter reviewer, or continuity query can see.
- License
- Apache-2.0
- Instructions
- 38 lines in SKILL.md · view on GitHub
Files (5)
- SKILL.md
- batching.md · resources/
- charter.md · resources/
- issue-format.md · resources/
- reader-ledger.md · resources/
SKILL.md (first lines)
# Cold Read
The cold-read skill owns the cover-to-cover cold-read protocol — a single reader, reading the whole manuscript in order, carrying a rolling ledger, and logging every defect that survives the per-chapter QA. It exists because a class of defect is invisible to every memoryless pass: a linter sees a line, a chapter reviewer sees a chapter, a continuity query sees a query — none of them accumulate reader-state. Clock drift, cross-chapter state contradictions, prop custody, promise bookkeeping at book scale, knowledge-state slips, and audit coverage bias are all *accumulating-memory* defects. The cold read is the only pass that catches them.
## What this skill is
- A **protocol**, not a single prompt. It runs on four files (charter, ledger, issues, batch reports) that together make a multi-sitting read behave like one continuous reader.
- **Assessment-only by default** (rule zero): the read measures; it does not edit the thing it measures. AUTO-FIX is opt-in, narrowly scoped, logged.
- **Batched** for sustainability and context-window safety: 8–11 chapters per sitting, the ledger the only inter-sitting state.
- A **self-contained** reference: it restates the severity taxonomy, the issue format, the batching model, and the orchestration pattern so the agent and the muse can run it without loading other skills. Cross-references to `gates/`, `kb-integrity/`, and `story-ledgers/` are pointers only.
## The document set
| File | Role |
|---|---|
| `work/cold-reads/<YYYY-MM>/charter.md` | Who the reader is, the rules, the batch plan, the severity taxonomy, autonomy tiers. Written once. |
| `work/cold-reads/<YYYY-MM>/reader_ledger.md` | The reader's accumulated memory: NEXT marker, position/clock, promise register, knowledge map, prop custody, do-not-re-explain, PROTECT, tripwires, grades. Updated every batch. |
| `work/cold-reads/<YYYY-MM>/issues.md` | Append-only issue log, one line per issue, fixed format. |
| `work/cold-reads/<YYYY-MM>/batches/batch_X.md` | Per-batch report: what works, grades, seam assessment, "would a paying reader keep going?" |
## Load the resource needed
- `resources/charter.md` — the reader persona ("devoted genre reader who just finished the previous book, with a line-editor's ear; experience first, diagnosis second"), rule zero, the batch plan, the severity taxonomy, AUTO-FIX vs PROPOSE tiers, voice cards, tripwires, PROTECT, deliverables.
- `resources/reader-ledger.md` — the rolling ledger: NEXT marker with WATCH list, position/clock table, promise register (HOT/WARM/COLD), character knowledge map, prop custody, do-not-re-explain register, PROTECT, tripwires, open threads, grades table. "A stale ledger is an amnesiac reader."
- `resources/issue-format.md` — the fixed line format `CR-### | ch:line | SEV | CAT | "quote" | reader-moment | fix direction | SCOPE`, the severity/category vocab, the severity→shared-schema mapping, and the rules (reader-moment mandatory, append-only, continuous numbering).
- `resources/batching.md` — 8–11 chapter batches cut at natural milestones, the per-sitting resume recipe (charter + ledger + last ~40 issues → continue from NEXT), the per-batch procedure, the batch report contents, the orchestration model, and when to run.
## Who runs it
- **Muse** sets up `work/cold-reads/<YYYY-MM>/` from `templates/cold-read/` (via `/vellum:cold-read`), seeds the ledger, and spawns `@cold-reader` per batch.
- **`@cold-reader`** executes one batch: reads charter + ledger + last ~40 issues, continues from NEXT, reads only the batch's chapters, returns (a) append-ready issue lines, (b) ledger section updates, (c) the batch report. Assessment-only; no peeking at outline or editorial notes during a batch.
- For **15+ chapters** muse MAY use the digest pattern from `fiction/agents/review-coordinator.md` (MIT) — spawn per-batch readers, aggregate from summaries, cache per-batch reports on disk.
## How it fits the gates
Cold-read findings are advisory by default; **BLOCKER-class findings block export** until resolved or dismissed-with-author-signoff (the same discipline as `story-ledgers/` exemptions). MAJOR/MODERATE need a triage note. The export readiness checklist (`gates/resources/readiness.md`) cross-checks the cold-read `issues.md` for open BLOCKERs.From FiredMosquito831/my-writing-companion, Apache-2.0.
More from FiredMosquito831/my-writing-companion
- 0.557score
qi-layer
FiredMosquito831
Use when writing or maintaining AGENTS.md, .context/CONTEXT.md, or CLAUDE.md mirrors: keep intent docs minimal and load-bearing.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
demolition
FiredMosquito831
Load when the author invokes demolition on a chapter draft, or when muse offers it before marking a chapter final.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
gates
FiredMosquito831
Load when muse accepts a pivotal chapter, before export, or for any scoring task.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
kb-integrity
FiredMosquito831
Load when a vellum check failed, or when muse routes integrity work.
SkillBFreshness: B★ 2Claude Code +9 - 0.553score
export
FiredMosquito831
Load for /vellum:export.
SkillBFreshness: B★ 2Claude Code +9 - 0.550score
project-setup
FiredMosquito831
One-time project setup for creative writing.
SkillBFreshness: B★ 2Claude Code +9
Signals
- Stars
- 2
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 9, 2026 (1 month ago)
- Last release
- Sep 11, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- Apache-2.0
- Website
- vellum.danubelabs.net
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from FiredMosquito831
- 0.560score
config
FiredMosquito831
Configure prompt-cache warming (keep-alive pings for idle Claude Code sessions) per scope - global, CLI, desktop app, project folder or single session -…
SkillAFreshness: AClaude Code +9 - 0.557score
demolition
FiredMosquito831
Load when the author invokes demolition on a chapter draft, or when muse offers it before marking a chapter final.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
gates
FiredMosquito831
Load when muse accepts a pivotal chapter, before export, or for any scoring task.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
kb-integrity
FiredMosquito831
Load when a vellum check failed, or when muse routes integrity work.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
qi-layer
FiredMosquito831
Use when writing or maintaining AGENTS.md, .context/CONTEXT.md, or CLAUDE.md mirrors: keep intent docs minimal and load-bearing.
SkillBFreshness: B★ 2Claude Code +9 - 0.553score
export
FiredMosquito831
Load for /vellum:export.
SkillBFreshness: B★ 2Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get cold-read running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
cold-read skill. Agent Mart. https://agentmart.dev/skills/firedmosquito831-cold-read (last updated 2026-10-24)