ClaudeMod
Back to browse
Harnesses

Claude Agentic Framework

dralgorhythm's lightweight agentic framework for Claude Code — composable pipelines, tool chaining, and structured output validation for reliable code generation.

dralgorhythm58 starsAdded 1 months ago

Claude Agentic Framework

A drop-in template for Claude Code projects. Adds coordinated multi-agent swarms, specialized commands, 67 reusable skills, and safety hooks — all configured through a single install command.

Install

Run this inside your project directory:

cd your-project
curl -sSL https://raw.githubusercontent.com/dralgorhythm/claude-agentic-framework/main/scripts/init-framework.sh | bash -s .

The script will:

  • Copy .claude/ (commands, skills, rules, hooks, agents, templates)
  • Copy .mcp.json (MCP server configuration)
  • Copy CLAUDE.md and AGENTS.md (project instructions)
  • Create an artifacts/ directory for planning documents
  • Set up .gitignore entries
  • Install hook dependencies
  • Initialize Beads issue tracking (required for swarm coordination)

Beads Setup

Beads is the issue tracker that coordinates swarm workers — it's how agents claim tasks, track progress, and avoid conflicts. Install it before running the init script:

curl -sSL https://raw.githubusercontent.com/steveyegge/beads/main/scripts/install.sh | bash

The init script will then run bd init in your project automatically.

The script prompts before overwriting any existing files. Re-run it to pull in framework updates.

After Install

  1. Edit CLAUDE.md — Add your build/test commands and project context
  2. Edit .claude/rules/tech-strategy.md — Configure your tech stack (this is required — the framework enforces whatever you put here)
  3. Start Claude Code and try: /architect hello

What You Get

Commands

Single-agent expert modes, invoked via slash commands:

CommandRole
/architectSystem design, ADRs
/builderImplementation, debugging, testing
/qa-engineerTest strategy, E2E, accessibility
/security-auditorThreat modeling, security audits
/ui-ux-designerInterface design, visual assets
/code-checkSOLID, DRY, consistency audit

Swarm Orchestrators

Multi-agent commands that fan work out across parallel workers:

CommandWhat It Does
/swarm-planLaunches 3-6 explorer agents to research patterns, dependencies, and constraints — produces a decomposed plan
/swarm-executePicks up planned work, fans out across builder agents (up to 8 parallel), each running quality gates
/swarm-reviewLaunches 5 parallel reviewers (security, performance, architecture, tests, quality) — run 2-3 times
/swarm-researchDeep multi-source investigation with verification tiers

The Full Cycle

/architect <feature>  →  /swarm-plan  →  /swarm-execute  →  /swarm-review (2-3x)  →  PR

One agent thinks. Many agents build. Many agents review.

Workers

Six specialized agent types tuned for cost and capability:

WorkerModelUse
worker-explorerHaikuFast codebase search, dependency mapping
worker-builderSonnetImplementation, testing, refactoring
worker-reviewerOpusCode review, security analysis
worker-researcherSonnetQuick web research, API docs
worker-researchOpusDeep multi-source investigation
worker-architectOpusComplex design decisions, ADRs

Skills

67 skills across 7 categories — auto-suggested based on keywords in your prompt:

Architecture · Engineering · Product · Security · Operations · Design · Languages & Frameworks

Covers everything from designing-systems and debugging to react-patterns, terraform, and application-security. See docs/skills.md for the full list.

Safety Hooks

Pre-configured hooks that run automatically:

  • Secret detection — blocks commits containing API keys, tokens, private keys
  • Protected files — prevents accidental modification of .env, .mcp.json, .beads/
  • Push blocking — stops direct pushes to main/master
  • Dangerous command guard — warns on rm -rf, force push, terraform destroy
  • File locking — prevents concurrent edits in multi-agent swarms

MCP Servers

Four servers pre-configured in .mcp.json:

ServerPurpose
Sequential ThinkingStructured multi-step reasoning
Chrome DevToolsBrowser testing, performance profiling
Context7Up-to-date library documentation
FilesystemFile operations beyond workspace

Customization

Everything is designed to be extended:

  • Add commands → .claude/commands/your-command.md
  • Add skills → .claude/skills/category/your-skill/SKILL.md
  • Add rules → .claude/rules/your-rule.md
  • Add hooks → .claude/hooks/your-hook.sh
  • Add workers → .claude/agents/worker-yourtype.md

Templates for each are in .claude/templates/.

See docs/customization.md for details.

Docs

More Harnesss

Harnesses

claude-simone

Structured project management harness for Claude Code — define milestones and sprints in markdown, then run execution loops that drive Claude through each task.

project-managementmilestonessprints+2
by Helmi
GitHub
Harnesses

Claude Squad

Multi-agent tmux orchestrator — spawn N isolated Claude Code sessions in parallel, each in its own git worktree, then merge the best results back to main.

multi-agentparalleltmux+3
by SMTG AI
GitHub
Harnesses

Claude Swarm

Distributed Claude agent swarms — agents communicate via MCP, share tool access, and collaborate on complex tasks with configurable topology and role specialization.

swarmdistributedmulti-agent+3
by parruda
GitHub

Command Palette

Search for a command to run...