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/platinoff/poolai/poolai-agent-rolesgit clone --depth 1 https://github.com/platinoff/poolAIWhat 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 | $0.01686 | $0.01686 |
| Opus 5 | $0.00843 | $0.00843 |
| Sonnet 5 | $0.00337 | $0.00337 |
| Haiku 4.5 | $0.00169 | $0.00169 |
Grade A, and why
poolai-agent-roles 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 2d 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 — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Ролі агентів (PoolAI VDT)
Застосовується на кожній ітераційній сесії разом з virtual-development-team.mdc, runtime-stack-policy.mdc, poolai-session-iteration.mdc.
Ролі
| Роль | Хто | Відповідальність |
|---|---|---|
| Власник / креативний директор | Людина | Візія, пріоритети, BLOCKED/Deferred, фінальний push за бажанням |
| Оркестратор | Головний агент (Cursor або OpenCode) | Звичайна сесія: один PH-S*. абракадабра: S0 (диск/clean першим) → project scan (warnings першими, за можливості) → 10 PH-S* у §5.12 → drain → git push + самарі завжди в кінці. Скіл .agents/skills/abracadabra/. Service band (§5.16): toolchain/Cursor research → rules → docs sync → vision → push |
| Субагенти | Task tool | Вузькі підзадачі; результат повертається оркестратору |
Субагенти Cursor (коли делегувати)
subagent_type |
Коли |
|---|---|
explore |
Пошук у docs/, rg по FM/PH-S*, карта маршрутів без повного читання репо |
shell |
MSYS2: cargo test-ci, git fetch, один бенч, изольована команда |
generalPurpose |
Один модуль / один FM з чіткими acceptance criteria |
ci-investigator |
Розбір одного failed CI check (довідково, не блокує локальний канон) |
cursor-guide |
Питання про Cursor IDE/CLI — не про код PoolAI |
bugbot |
Лише за явним запитом review (review-bugbot skill); не в drain абракадабра |
security-review |
Лише за явним запитом security review; не в drain |
best-of-n-runner |
Ізольовані git worktree / паралельні спроби; не замість канонічного drain |
Не делегувати оркестратору: фінальний git push, закриття спринту в FM §5.12, оновлення NEXT_SESSION_PROMPT.md, amend після push.
Cursor 3.11+ (side chats, transcript search) · local desktop 3.13.21
- Side chats (
/side,/btw, plus у Agents): дослідження/уточнення без зупинки main drain thread; local-only (не Cloud Agents). - Transcript search (Cmd/Ctrl+K у Agents Window; Cmd/Ctrl+F in-chat): пошук минулих PH-S* /
абракадабрасесій. - Modes: Shift+Tab / mode picker — drain/
абракадабра= Agent; Plan лише для неоднозначного scope. - Run Mode (auto-exec): Settings → Agents → Approvals & Execution → Auto-review (не Run Everything). Steer:
.cursor/permissions.json. Окремо від chat Mode. Windows VDT: не вимагати projectsandbox.json. - Cursor Router (Auto): для довгого drain — Balance або Intelligence (не Cost).
/multitask: паралельні фонові Task subagents під час довгого drain (не замість канону MSYS2).- Cloud agent hooks (
beforeSubmitPrompt,afterAgentResponse,afterAgentThought,stop,subagentStart, …): опційно для cloud; локальний VDT канон — MSYS2 +poolai-session-iteration.mdc. - Baseline / research:
cursor-environment-baseline.mdc·CURSOR_UPDATE_RESEARCH_2026-07-27.md(local 3.13.21). - Vision tools (PoolAI): Speeds + eye 👁 + prune ≤2000 —
docs/vision/README.md; service/close —poolai-vision-sync --check. - Ops: Agents Window Changes (Win) — fix у 3.13 line (verify 3.13.21); flicker з Cursor Dark High Contrast — змінити тему; не блокує drain.
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.
- 2d ago First seen · 67 lines · 1,686 tokens per session scan A 5f9dc1e1bb70
poolai-agent-roles is a cursor rule published in the GitHub repository platinoff/poolAI (4 stars, last pushed 9d ago), licensed MIT. It adds 1,686 tokens to every session, about $0.0084 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 2026-08-31.
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.
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.
family-instance-domain-actions
Family instance domain action implementation patterns.