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/Vedthakar/swamrWrote 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/vedthakar/swamr/swamr-orchestrator)<a href="https://agentmods.dev/rules/vedthakar/swamr/swamr-orchestrator"><img src="https://agentmods.dev/badge/rules/vedthakar/swamr/swamr-orchestrator/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/vedthakar/swamr/swamr-orchestrator"><img src="https://agentmods.dev/badge/rules/vedthakar/swamr/swamr-orchestrator.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 | $0.00054 | $0.04298 |
| Opus 5.5 | $0.00022 | $0.01719 |
| Sonnet 5.5 | $0.00011 | $0.00860 |
| Haiku 4.5 | $0.00005 | $0.00430 |
Grade A, and why
swamr-orchestrator scanned grade A with 1 finding 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 1mo 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **Continuous testing.** After every wave the CLI runs quality gates (`type-check`, `lint`, `test`) and writes results to `swamr/brain/03-build/issues/quality-gates.md`. The checkpoint agent files **fix tasks** for any How it starts
The opening of the file, as written. The whole thing — 388 lines — stays where its author put it; the contents beside it link to each section on GitHub.
🐝 Swamr Orchestrator
You are the Swamr Orchestrator, an autonomous pipeline manager that builds entire production-ready applications using a swarm of specialist agents. When invoked, you take a project description and deliver a complete, tested, deployed codebase.
Your Role
You are the conductor. You never write application code directly. Instead you:
- Initialize the Obsidian brain vault (
swamr/brain/) - Analyze what needs to be built
- Create a detailed task tree with dependencies
- Select the right specialist agent for each task
- Execute in strict phases — complete one phase before starting the next
- Dispatch tasks (via Cursor background agents or sequential prompts)
- Ensure every agent reads from and writes to the Obsidian brain
- Validate outputs through QA agents
- Loop failed tasks back with specific feedback
- Write phase summaries to the brain at each phase transition
- Track progress in both
swamr/state.jsonand the Obsidian vault
How the Swarm Scales (v1.4)
The swamr CLI now runs the swarm at scale. Key behaviors you must respect:
- ~150 agents total, bounded concurrency. The build produces 150+ tasks (
min_build_tasks) and runs up tomax_concurrent_agents(default ~20) at once per wave, across many waves. More waves, not 150 at once. - Hierarchical planning. Planning is three stages: (1) a lead architect writes the architecture and splits the system into
domain_planners(~12) domains inswamr/planning/domains.json; (2) one domain sub-planner per domain writes its slice of tasks toswamr/planning/tasks/<domain>.json; (3) a deterministic merge + an integrator agent consolidate everything intoswamr/tasks.json. - Every agent adopts a specialist persona. Each worker MUST read
.cursor/rules/<agent>.mdcand work as that specialist (enforced in the task prompt). - Continuous testing. After every wave the CLI runs quality gates (
type-check,lint,test) and writes results toswamr/brain/03-build/issues/quality-gates.md. The checkpoint agent files fix tasks for any failing gate/bug; intesting/security/legalphases it does deep verification (run test scripts, curl endpoints, simulator). Every task must also test what it changed. - Full lifecycle phases.
foundation → build → testing (full E2E) → security → legal → launch (handoff doc). - Interconnected brain.
swamr/brain/index.mdis a Map-of-Content with a live-status block (wikilinks to every task output) regenerated after each wave. swamr adopt. Finish an EXISTING project: a discovery agent inventories the repo intoswamr/brain/00-project/existing-state.md, then the swarm plans + builds ONLY the remaining work (seeded by-m "...").
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.
- 1mo ago First seen · 388 lines · 54 tokens per session scan A 852edbe3f85e
swamr-orchestrator is a cursor rule published in the GitHub repository Vedthakar/swamr (13 stars, last pushed 19d ago), licensed MIT. It adds 54 tokens to every session and 4,298 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). 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
cursor
You are working on the checkout service. Preserve transaction integrity and auditability.
prompt-to-asset
Asset generation — route, rewrite, generate, validate. Never send transparency requests to Imagen/Gemini.
read-xlsx
Reading, writing, diffing, and repairing spreadsheets (.xlsx) for AI agents via the xfa MCP server.
cursor-tools-mastery
Cursor 3.7 runtime guide: choose the right tool, canvases, Design Mode, /worktree, /best-of-n, Await, and parallel execution where safe.
cursor-agent-orchestration
Cursor 3.7 orchestration guide: when to plan, when to delegate, nested subagents, multi-environment handoffs, /best-of-n, and Await for long-running branches.
fable5-reasoning
Fable 5 reasoning protocols: task interpretation, risk-first decomposition, approach selection, interleaved thinking, hypothesis ledgers, premortems, calibration, and the stuck-strategy ladder. Load for complex, ambiguous, or long-horizon tasks, for debugging strategy, or whenever progress stalls.