litcritic
Skill0 starsChanged 6 days agoMITView on GitHub
Works on a novel manuscript with the literary-critic toolkit — setting it up for a writer ("I want to use literary-critic for my novel"), ingesting a draft, recording scenes and characters in the Codex, answering questions about the novel from it, running the analysis checklist, recording findings, revision rounds, comparing drafts. Use whenever the user talks about their novel, a manuscript, a draft, chapters, characters, an editorial letter, revision rounds or a reader report, even without naming the tool, and whenever an `lc` or `lc codex` command or a workspace under `$LITCRITIC_HOME` comes up. Not for changes to the tool's own code (those are [[...]] requests handled as normal development).
- Installs packages
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- claude-arsenal/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- coverage-gapsSkill
- dep-upgradeSkill
- designSkill
- executionSkill
- explore-ideaSkill
- gate-checkSkill
- githubSkill
- initSkill
- mutmut-reportSkill
- pin-checkSkill
- pypi-releaseSkill
- python-bootstrapSkill
- queue-addSkill
- queue-nextSkill
- queue-statusSkill
- reviewSkill
- session-endSkill
- shipSkill
- specifySkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 25
Otherwise: use the installed lc command.
- Very specific: line 28
If that fails, ask the writer to add the literary-critic Skill (a zip they upload once in Claude's settings, under Skills) and start again.
- Very specific: line 30
Check it works with lc doctor.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
literary-critic
CANARY: litcritic-loaded-2026-10-11-2ebb846e930ff66d
A writer who says "I want to use literary-critic for my novel" (in any
words or language) starts with prompts/05-setup.md: it checks what this
Claude can do, finds the novel and agrees how to work, step by step.
Running the tool
- This folder has
lc.py(the Skill uploaded to Claude, or unzipped into~/.claude/skills/): wherever the procedures run lc with a command, runpython3 <this folder>/lc.pywith that command instead (pythonwhen there is nopython3, as on Windows); there is nolccommand to look for. The procedures are in this folder'sprompts/. - In a source checkout of the tool (this skill under
.claude/skills/):uv run lc, never an installed copy, so the working tree is what runs. - Otherwise use the installed lc command. If it is missing and the
network allows it, install it:
pip install git+https://github.com/nuncaeslupus/literary-critic. If that fails, ask the writer to add the literary-critic Skill (a zip they upload once in Claude's settings, under Skills) and start again. - Check it works with
lc doctor. Work from a writable folder (the novel's folder, or a scratch folder in a chat): never create anything inside this Skill's folder, which may be read-only.
The procedures live in prompts/ so any chatbot can follow them; this skill
says which one to read. Read prompts/00-guide.md at the start of every
session that touches a novel. It holds the rules that matter most: the writer
writes (no drafted prose unless the writer chose small fixes or suggested
wording, after the AI-text notice), never edit the manuscript without
permission for that change, plain language, locate everything, one change per
message.
What is known about a novel lives in its Codex (<ws>/codex/, read and
written only through lc codex; reference in docs/codex/README.md). Ask
it before reading text, and record what reading teaches.
The index is built in bounded packets, never by reading the whole novel:
lc index plan, then entity packets (strong tier), plan again, scene packets
(standard tier; in Claude Code or Cowork one subagent per packet in parallel,
answers imported one at a time by the main session), plan again for the n1
new-names packet (strong tier). Print a packet with lc index packet, import
an answer with lc index import; lc index status shows what is left.
Details in prompts/20-codex-items.md.
| The user wants to… | Read |
| start using the tool on a novel ("I want to use literary-critic for my novel"), set it up | prompts/05-setup.md |
| start, or where things stand is unclear | prompts/00-guide.md, then the workspace log/ |
| bring in a draft, fix chapters/scenes | prompts/10-ingest-and-structure.md |
index the novel for analysis (lc index packets), fix one Codex item | prompts/20-codex-items.md |
| a first reading, brief, profile | prompts/30-orientation.md |
| a question about the novel ("how often…", "where is…", "what changed…") | prompts/35-ask.md |
| analysis, feedback, "what's wrong with it" | prompts/40-checks.md |
| look at a new draft | prompts/50-new-draft.md |
| a revision round, the editorial letter, recording decisions | prompts/60-rounds.md |
| an agency/editor assessment, a reader report | prompts/70-reader-report.md |
| go to a passage in Word, apply an approved change to the Word file | prompts/80-live-word.md (only lc docx and lc word; lc word acts in Word through doc-control when it can reach it, else gives doc-control's tool call and computer-use steps). "Take me to chapter 7": lc docx locate 7 --go --workspace <ws> at once, then "Done." if it answers "status": "done"; read the file only otherwise |
Workspaces live outside the tool's own files: in the novel's folder
(<folder>/literary-critic/, made by lc setup) or under
$LITCRITIC_HOME/<slug>/. When working in the tool's source repo: before
any commit, run make guard (nothing from a working novel may enter the
repo; see AGENTS.md), and turn what a novel teaches into generic changes.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

