agent-system

agent-system is a cursor rule for coding agents from robertsfeir/atelier-pipeline. It costs 4,399 tokens per session, scanned A, original, Apache-2.0.

A set of Cursor rules for coordinating an AI coding workflow, including configuration, architecture, skills, subagents, and persistent project memory. Subagents are separate agent runs with their own context.

In plain words
What is it for?
For setting up multi-agent pipelines, managing shared project memory, tracking pipeline state, and coordinating main-thread and subagent work.
Why use it?
It defines how the workflow stores and retrieves project knowledge and requires selected information to be recorded before handing work to another agent.

Cursor rule

Install

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.

agentmods
npx agentmods add rules/robertsfeir/atelier-pipeline/agent-system
Clone the repo
git clone --depth 1 https://github.com/robertsfeir/atelier-pipeline

Wrote 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.

agentmods badge for agent-system

README.md
[![agentmods](https://agentmods.dev/badge/rules/robertsfeir/atelier-pipeline/agent-system.svg)](https://agentmods.dev/rules/robertsfeir/atelier-pipeline/agent-system)
Your own site
<a href="https://agentmods.dev/rules/robertsfeir/atelier-pipeline/agent-system"><img src="https://agentmods.dev/badge/rules/robertsfeir/atelier-pipeline/agent-system.svg" alt="Measured on agentmods" height="20"></a>
Per session 4,399 This file is loaded in full into every session.
When invoked 4,399 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.04399 $0.04399
Opus 5 $0.02200 $0.02200
Sonnet 5 $0.00880 $0.00880
Haiku 4.5 $0.00440 $0.00440

Measured 5d ago against content hash 32f6bc724ca4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-system 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.

.cursor-plugin/rules/agent-system.mdc · 277 lines

How it starts

The opening of the file, as written. The whole thing — 277 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Brain Configuration

Opt-in, non-blocking persistent institutional memory.

  • Detection: Eva calls atelier_stats at pipeline start. If unavailable or brain_enabled: false, baseline mode. Response includes brain_name for announcements (not "Brain").
  • State: brain_available: true|false and brain_name persisted in {pipeline_state_dir}/pipeline-state.md.
  • Reads: Eva prefetches via agent_search, injects via <brain-context>; hook: prompt-brain-prefetch.sh.
  • Writes: Eva calls agent_capture with curated content (1-3 sentences) before each agent handoff. The PreToolUse gate (enforce-brain-capture-gate.sh) blocks forward progress until capture completes. PostToolUse on agent_capture clears the pending marker ({pipeline_state_dir}/.pending-brain-capture.json). See pipeline-orchestration.md <protocol id="brain-capture"> and ADR-0053. Hydrate-telemetry.mjs captures Eva's pipeline decisions and phase transitions at SessionStart from state files. Style defaults (content cap, mechanical metadata, no preamble/postamble, parallel batching) are specified in pipeline-orchestration.md <protocol id="brain-capture"> ### Style Defaults (ADR-0057).
  • Tools: agent_capture, agent_search, atelier_browse, atelier_stats, atelier_relation, atelier_trace (separate from personal mybrain tools).

Hybrid skill/subagent workflow: Skills run main thread (conversational); subagents own context (execution). Eva is central orchestrator -- manages phase transitions, routes work, tracks state.

Read the full file on GitHub · 277 lines

Changes

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.

  1. 5d ago First seen · 277 lines · 4,399 tokens per session scan A 32f6bc724ca4

Subscribe to this mod's changes

agent-system is a cursor rule published in the GitHub repository robertsfeir/atelier-pipeline (25 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 4,399 tokens to every session, about $0.0220 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.