Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add rules/robertsfeir/atelier-pipeline/step-sizinggit clone --depth 1 https://github.com/robertsfeir/atelier-pipelineWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/rules/robertsfeir/atelier-pipeline/step-sizing)<a href="https://agentmods.dev/rules/robertsfeir/atelier-pipeline/step-sizing"><img src="https://agentmods.dev/badge/rules/robertsfeir/atelier-pipeline/step-sizing.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00607 | $0.00607 |
| Opus 5 | $0.00303 | $0.00303 |
| Sonnet 5 | $0.00121 | $0.00121 |
| Haiku 4.5 | $0.00061 | $0.00061 |
Grade A, and why
step-sizing scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Step Sizing Gate
After drafting the implementation plan, apply these five tests to each step. Steps touching 10+ files (files_to_create + files_to_modify) MUST pass all five. Steps under 10 files are reviewed at Sarah's judgment but should still pass.
| # | Test | Question | Fail Action |
|---|---|---|---|
| S1 | Demoable | Can you state what this step enables in one user-facing sentence? ("After this step, I can ___") | Split along demo boundaries -- each demoable behavior becomes its own step |
| S2 | Context-bounded | Does Colby need <= 8 files (create + modify) to implement it? | Extract excess files into a prerequisite or follow-up step |
| S3 | Independently verifiable | Can Poirot test this step without the next one existing? | Split so each verifiable behavior is its own step |
| S4 | Revert-cheap | If Poirot fails it, can Colby redo it in one fresh invocation? | Split until each piece is one-invocation sized |
| S5 | Already small | Is this step <= 6 files with one clear behavior? | Do NOT split -- over-splitting wastes orchestration overhead |
Evidence
Sub-slicing from ~15 files/step to ~8 files/step improved first-pass QA from 57% to 93% and reduced rework findings by 90% (same model, same pipeline, same project -- only variable was step granularity). Context window research shows LLM accuracy degrades above ~32k tokens of effective context.
Split Heuristics
When a step fails the gate, look for these seams:
| Seam | Example |
|---|---|
| CRUD separate from lifecycle/state machine | Firm create/list vs suspend/archive |
| Read-only separate from mutations | User search/list vs suspend/grant-PM |
| Foundation separate from first consumer | Auth guard + layout vs audit log page |
| Security separate from config CRUD | AI settings vs key encryption/rotation |
| Data pipeline separate from dashboard | Cost event logging vs cost trend UI |
| Schema/job separate from UI | Analytics aggregation vs metrics page |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 52 lines · 607 tokens per session scan A 5c78a1b9b39b
step-sizing is a cursor rule published in the GitHub repository robertsfeir/atelier-pipeline (25 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 607 tokens to every session, about $0.0030 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other cursor rules, from other repositories
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
cli-error-handling
CLI command error handling patterns.
family-instance-domain-actions
Family instance domain action implementation patterns.
prefer-assertions-over-defensive-checks
Prefer assertions over defensive checks when data is guaranteed to be valid.
prefer-direct-imports-over-module-mocks
Prefer extracting a testable core over vi.mock / vi.resetModules when unit tests need to reach production logic entangled with config, env, or singletons.