rumo-prose
Skill7 starsChanged 14 days agoMITView on GitHub
Use when writing, reviewing, or trimming prose in a software repository, including Markdown, code and test comments, JavaDoc/JSDoc, diagnostics, logs, API errors, configuration descriptions, UI strings, prompts, and agent-facing files such as SKILL.md, AGENTS.md, and CLAUDE.md. Keep every behavior and failure fact while removing repetition, narration, and session residue. Do not use for formal Word delivery documents.
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Write as much as a maintainer, caller, operator, or user needs to act correctly, and no more.
- Distinctive: line 13
Check wording against the code, configuration, schema, or runtime behavior it describes before judging it.
- Very specific: line 39
Read references/agent-docs.md before writing or reviewing a SKILL.md, AGENTS.md, CLAUDE.md, or a reusable prompt.
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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
Rumo Prose
Write as much as a maintainer, caller, operator, or user needs to act correctly, and no more. Cut words only after every fact they carried survives somewhere.
Scope And Authority
- A review or audit reports findings; a write, fix, or trim request authorizes edits inside the named scope.
- Check wording against the code, configuration, schema, or runtime behavior it describes before judging it.
- Edit generated text at its source. Leave fixtures, snapshots, vendored files, and historical records alone unless they are the explicit target.
- Prompts, API errors, and product strings are behavior: update their tests or snapshots when wording changes what users or models see.
Keep The Facts
Before shortening a passage, list what it asserts: who acts, under which condition and order, what is required or forbidden, side effects and cleanup, failure and recovery, authorization and data scope, compatibility, and any guarantee that something does not happen. The shorter version must still assert all of them. Keep a non-obvious reason when dropping it would invite misuse; otherwise point to where the reason lives.
Standards By Surface
| Surface | Write |
| Public API docs | caller-visible behavior that types do not show: failures, side effects, threading, cancellation, ownership |
| Code comments | the constraint or reason behind non-obvious code, never a restatement of the next line |
| Tests | the behavior and condition being pinned |
| Errors | the failed operation, the target, a safe next step, and a stable category, without secrets |
| Logs | correlation fields and state transitions needed to diagnose, without credentials or full payloads |
| UI strings | product language naming the action and its consequence, with no implementation terms |
| Repository Markdown | current behavior in present tense; rationale belongs in a decision record |
Remove Session Residue
Durable text must make sense to someone who never saw the chat, review, or plan that produced it. Remove review ordinals without a committed owner, phrases like "this change" or "later PR", reviewer-directed asides, references to uncommitted plans, and "used to / now" stories on current-state surfaces. Rewrite any real fact they contain in the present tense.
Agent-Facing Files
Read references/agent-docs.md before writing or reviewing a SKILL.md, AGENTS.md, CLAUDE.md, or a reusable prompt.
Verify And Report
Re-read each edited passage against its source of truth, search the changed scope for stale terms or duplicated statements, and run the checks the surface requires. Report the scope, the kinds of changes, the evidence used, and any wording that still needs a product decision.
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- rumo-agent-collaborationSkill
- rumo-clarifySkill
- rumo-coding-guidelinesSkill
- rumo-database-change-safetySkill
- rumo-debugSkill
- rumo-explainSkill
- rumo-filesystem-safetySkill
- rumo-frontendSkill
- rumo-git-commitSkill
- rumo-handoffSkill
- rumo-interfaceSkill
- rumo-lifecycle-safetySkill
- rumo-planSkill
- rumo-project-profileSkill
- rumo-reflectSkill
- rumo-remote-opsSkill
- rumo-reviewSkill
- rumo-setupSkill
- rumo-testSkill
- rumo-verifySkill
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.