swarms

Name: swarms
Author: DheerG

82stars7forks

How to install swarms?

dheerg/swarms

$git clone https://github.com/DheerG/swarms

Installs into the current project.

Summary

Achieve extraordinary results with claude code across a variety of tasks

Files · 1

View on GitHub

README.md

1	# Swarm
2
3	Describe what you want. Get a reviewed, ship-ready PR — without babysitting the agent.
4
5	Out-of-the-box agent teams can be excellent, but getting there takes time and effort — the right specialists, each with a point of view, and a bias applied where it sharpens the team and held back where it would box the agents in. Swarm puts in that effort for you, so the team converges instead of chattering and the work holds up.
6
7	```
8	/swarm:code add SSO so enterprise customers can log in with their own identity provider
9	```
10
11	Swarm assembles a team of five to eight for the task: experienced engineers, plus the customer and business voices a lone coder usually works without. They research the problem, debate the approach, build it, and review each other's work until it's ready, then hand you a PR. You approve the plan and the approach, then stay out of the build. You have the final say before anything ships.
12
13	Swarm also runs triage, writing, and general-purpose teams — see [other ways to launch](#commands).
14
15	---
16
17	## What you get
18
19	- Ships code that holds up — reviewed by a different agent, not rubber-stamped by its author.
20	- Runs unattended for hours — even across days.
21	- Survives interrupted turns without stalling.
22	- Avoids the rate-limit errors parallel teams hit — at a fraction of the tokens.
23
24	---
25
26	## Quick start
27
28	Swarm is a [Claude Code](https://code.claude.com) plugin, so you'll need the CLI installed first (v2.1.178 or newer).
29
30	```bash
31	claude plugin marketplace add DheerG/swarms
32	claude plugin install swarm@swarms --scope project
33	```
34
35	Agent teams must be enabled in Claude Code. Add this to `~/.claude/settings.json` (or let `/swarm:code` enable it for you on first run):
36
37	```json
38	{
39	"env": {
40	"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
41	}
42	}
43	```
44
45	Restart Claude Code so the plugin and the flag take effect, then run `/swarm:code` and describe what you want.
46
47	---
48
49	## How it works
50
51	Three moments need you. The rest runs itself.
52
53	1. Tell it what you want — 1-2 sentences describing the outcome. It has the ability to refine the outcome with you and work around prompting gotchas that would normally result in poor code.
54	2. Approve the team and the plan — the suggested specialists (swap or add members by pointing at them), the plan, and, mid-run, the approach the team converged on. Nothing irreversible happens without your sign-off.
55	3. Decide how to finish — when the team hits the bar, you choose to refine further or ship; you see the result, not the debate, though you can follow it live in [AgentChat](https://github.com/DheerG/agent-chat) if you want to.
56
57	You see the whole plan before a single agent runs:
58
59	> Team Plan
60	>
61	> Mode: Code
62	> Outcomes: (your words, verbatim)
63	> Team: lead (your Claude session) + Principal Engineer facilitator + specialists for the work
64	> Cost tier: Ultra
65	> Ship definition: Create a feature branch from main, commit, push, open PR
66	>
67	> ---
68	> Is this plan final, or do you have remaining inputs?
69	>
70	> - Launch the team — Plan is final — start creating the team now
71	> - I have changes — Let me adjust outcomes, members, or settings first
72
73	Nothing runs until you pick "Launch the team" — and before this plan you've already picked the cost tier and approved the team member by member.
74
75	---
76
77	## A real Converge, in excerpt
78
79	Here is an excerpt from a real caching-strategy Converge:
80
81	> Principal Engineer: Two approaches on the table — cache-aside with a TTL, or read-through with invalidation on write. Security Reviewer, you flagged the TTL approach earlier. What's the actual risk you see?
82	>
83	> Security Reviewer: TTL means stale data can persist for up to N minutes after a permission change. If this endpoint returns authorization state, that's a window.
84	>
85	> Principal Engineer: Lead, does this endpoint return authorization state, or pure profile data?
86	>
87	> Lead: Profile data only — no permissions in the response.
88	>
89	> Principal Engineer: Then TTL risk is acceptable here. Team, any other concerns before I call CONVERGED?
90
91	Notice that the facilitator never settles the question itself. It hands the blocker to whoever can answer it and waits for evidence, and the lead fields that question like any other member. The security risk only surfaced because someone other than the implementer was looking. On their own, they'd have shipped the stale-data window. What reaches you is the conclusion; the argument stays inside the team.
92
93	---
94
95	## Why it works
96
97	### Outcomes over implementations
98
99	You describe what the world looks like when the work is done, rather than the steps to build it. That changes what the team debates.
100
101	- Implementation framing: "Add a Redis cache in front of the user service."
102	- Outcome framing: "User profile reads return in under 50ms and don't hit the DB when cached."
103
104	Implementation framing has the team comparing Redis and Memcached. Outcome framing has them weigh caching against query optimization, the wider and

People who liked this also viewed

6 picks

Type

tradingview-mcpAI-assisted TradingView chart analysis — connect Claude Code to your TradingView Desktop for personal workflow automationWorkflowsJul 20269485.0k
FlowiseBuild AI Agents, VisuallyWorkflowsJul 2026—55k
simBuild, deploy, and orchestrate AI agents. Sim is the central intelligence layer for your AI workforce.WorkflowsJul 2026—29k
openai-agents-pythonA lightweight, powerful framework for multi-agent workflowsWorkflowsJul 2026—28k
claude-code-tips40+ tips for getting the most out of Claude Code, from basics to advanced - includes a custom status line script and Claude Code running itself in a container. Also includes the dx plugin: skills for everyday dev workflows.WorkflowsJul 2026—9.4k
claude-for-legalA suite of plugins for legal workflowsWorkflowsJul 2026—8.8k

swarms

82stars7forks

View on GitHub

How to install swarms?

dheerg/swarms

$git clone https://github.com/DheerG/swarms

Installs into the current project.

Summary

Achieve extraordinary results with claude code across a variety of tasks

Files · 1

View on GitHub

README.md

1	# Swarm
2
3	Describe what you want. Get a reviewed, ship-ready PR — without babysitting the agent.
4
5	Out-of-the-box agent teams can be excellent, but getting there takes time and effort — the right specialists, each with a point of view, and a bias applied where it sharpens the team and held back where it would box the agents in. Swarm puts in that effort for you, so the team converges instead of chattering and the work holds up.
6
7	```
8	/swarm:code add SSO so enterprise customers can log in with their own identity provider
9	```
10
11	Swarm assembles a team of five to eight for the task: experienced engineers, plus the customer and business voices a lone coder usually works without. They research the problem, debate the approach, build it, and review each other's work until it's ready, then hand you a PR. You approve the plan and the approach, then stay out of the build. You have the final say before anything ships.
12
13	Swarm also runs triage, writing, and general-purpose teams — see [other ways to launch](#commands).
14
15	---
16
17	## What you get
18
19	- Ships code that holds up — reviewed by a different agent, not rubber-stamped by its author.
20	- Runs unattended for hours — even across days.
21	- Survives interrupted turns without stalling.
22	- Avoids the rate-limit errors parallel teams hit — at a fraction of the tokens.
23
24	---
25
26	## Quick start
27
28	Swarm is a [Claude Code](https://code.claude.com) plugin, so you'll need the CLI installed first (v2.1.178 or newer).
29
30	```bash
31	claude plugin marketplace add DheerG/swarms
32	claude plugin install swarm@swarms --scope project
33	```
34
35	Agent teams must be enabled in Claude Code. Add this to `~/.claude/settings.json` (or let `/swarm:code` enable it for you on first run):
36
37	```json
38	{
39	"env": {
40	"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
41	}
42	}
43	```
44
45	Restart Claude Code so the plugin and the flag take effect, then run `/swarm:code` and describe what you want.
46
47	---
48
49	## How it works
50
51	Three moments need you. The rest runs itself.
52
53	1. Tell it what you want — 1-2 sentences describing the outcome. It has the ability to refine the outcome with you and work around prompting gotchas that would normally result in poor code.
54	2. Approve the team and the plan — the suggested specialists (swap or add members by pointing at them), the plan, and, mid-run, the approach the team converged on. Nothing irreversible happens without your sign-off.
55	3. Decide how to finish — when the team hits the bar, you choose to refine further or ship; you see the result, not the debate, though you can follow it live in [AgentChat](https://github.com/DheerG/agent-chat) if you want to.
56
57	You see the whole plan before a single agent runs:
58
59	> Team Plan
60	>
61	> Mode: Code
62	> Outcomes: (your words, verbatim)
63	> Team: lead (your Claude session) + Principal Engineer facilitator + specialists for the work
64	> Cost tier: Ultra
65	> Ship definition: Create a feature branch from main, commit, push, open PR
66	>
67	> ---
68	> Is this plan final, or do you have remaining inputs?
69	>
70	> - Launch the team — Plan is final — start creating the team now
71	> - I have changes — Let me adjust outcomes, members, or settings first
72
73	Nothing runs until you pick "Launch the team" — and before this plan you've already picked the cost tier and approved the team member by member.
74
75	---
76
77	## A real Converge, in excerpt
78
79	Here is an excerpt from a real caching-strategy Converge:
80
81	> Principal Engineer: Two approaches on the table — cache-aside with a TTL, or read-through with invalidation on write. Security Reviewer, you flagged the TTL approach earlier. What's the actual risk you see?
82	>
83	> Security Reviewer: TTL means stale data can persist for up to N minutes after a permission change. If this endpoint returns authorization state, that's a window.
84	>
85	> Principal Engineer: Lead, does this endpoint return authorization state, or pure profile data?
86	>
87	> Lead: Profile data only — no permissions in the response.
88	>
89	> Principal Engineer: Then TTL risk is acceptable here. Team, any other concerns before I call CONVERGED?
90
91	Notice that the facilitator never settles the question itself. It hands the blocker to whoever can answer it and waits for evidence, and the lead fields that question like any other member. The security risk only surfaced because someone other than the implementer was looking. On their own, they'd have shipped the stale-data window. What reaches you is the conclusion; the argument stays inside the team.
92
93	---
94
95	## Why it works
96
97	### Outcomes over implementations
98
99	You describe what the world looks like when the work is done, rather than the steps to build it. That changes what the team debates.
100
101	- Implementation framing: "Add a Redis cache in front of the user service."
102	- Outcome framing: "User profile reads return in under 50ms and don't hit the DB when cached."
103
104	Implementation framing has the team comparing Redis and Memcached. Outcome framing has them weigh caching against query optimization, the wider and

People who liked this also viewed

6 picks

Type