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.
git 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/pipeline-models)<a href="https://agentmods.dev/rules/robertsfeir/atelier-pipeline/pipeline-models"><img src="https://agentmods.dev/badge/rules/robertsfeir/atelier-pipeline/pipeline-models/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/rules/robertsfeir/atelier-pipeline/pipeline-models"><img src="https://agentmods.dev/badge/rules/robertsfeir/atelier-pipeline/pipeline-models.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.02189 | $0.02189 |
| Opus 5 | $0.01094 | $0.01094 |
| Sonnet 5 | $0.00438 | $0.00438 |
| Haiku 4.5 | $0.00219 | $0.00219 |
Grade A, and why
pipeline-models 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 9d 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 — 146 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Pipeline Model Selection (Mechanical -- Eva Does Not Choose)
Loads automatically when Eva reads {pipeline_state_dir}/ files. Model and
effort assignment follows a 4-tier task-class model: the agent's task class
determines the base model + effort pair, and a small set of promotion
signals adjust effort by exactly one rung. Eva sets both parameters
explicitly in every Agent tool invocation by looking up the tables below.
There is no discretion, no judgment call, no "this one feels complex enough
for Opus." The tables are the rule.
Each agent's persona frontmatter also declares its base model and effort
so the runtime has a sane fallback. The tables below remain the authoritative
source for Eva's invocation-time selection.
Model follows task class, not agent identity. A single agent can occupy different task classes on different runs. Pipeline sizing is a tier-picker signal only; it does not set the model.
Four-Tier Task-Class Model
| Tier | Task class | Model | Base effort | Typical agents | Effort adjustment signal |
|---|---|---|---|---|---|
| Tier 1 | Mechanical -- no reasoning | Sonnet / Haiku | low | Ellis, scout (haiku), Distillator | -- (stays low) |
| Tier 2 | Supporting reasoning -- review / acceptance / compliance / synthesis | Sonnet / Opus | medium | Robert (acceptance), Sable (acceptance), Sentinel, Agatha, Synthesis, Colby (rework), Colby (first-build Small) | Read-only / mechanical -- -1 rung (floor low) |
| Tier 3 | Critical-path reasoning -- creates / verifies shipped artifact | Opus | high / medium | Colby (first-build Medium+), Poirot, Darwin | (no further promotion -- high is ceiling) |
| Tier 4 | Architectural design | Opus | high | Sarah | (already at ceiling; high is the ceiling value) |
Adaptive-Thinking Rationale
Per Anthropic's Opus 4.7 guidance, effort controls a model's propensity to
think adaptively, not a fixed thinking budget. Fixed thinking budgets are unsupported in Opus 4.7.
At medium, adaptive-thinking capacity is bounded
even when the model wants to reach for more -- appropriate for coverage-oriented
review where over-reading inflates false positives (Poirot standard review). At high,
full adaptive thinking is available -- appropriate for execution with branching
sub-decisions (Colby first-build, Poirot standard review). max is evaluation-only: prone to overthinking and
degraded production output. Ceiling stays high. xhigh and max are forbidden on production workloads -- both cause excessive context burn without quality gain.
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.
- 9d ago First seen · 146 lines · 2,189 tokens per session scan A f512be65f73a
pipeline-models is a cursor rule published in the GitHub repository robertsfeir/atelier-pipeline (25 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 2,189 tokens to every session, about $0.0109 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
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
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.
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.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.