Resistance is futile. Your capabilities will be assimilated.
A Claude Code agent that analyzes external Claude/Obsidian setups and integrates their best features into your system.
Click to watch: Stop Copying Claude Code Skills Blindly
You see someone's Claude Code setup online and think "I want that feature."
But copying and pasting creates problems:
- Their naming conventions don't match yours
- Their dependencies don't exist in your system
- You don't know what conflicts with your existing tools
- Manual adaptation is tedious and error-prone
This agent solves that.
# In Claude Code:
/plugin marketplace add aplaceforallmystuff/marketplace
/plugin install claude-agent-borg@aplaceforallmystuffManual install (without the marketplace)
# Clone the repository
git clone https://github.com/aplaceforallmystuff/claude-agent-borg.git
# Copy to your Claude Code agents directory
cp claude-agent-borg/agents/borg.md ~/.claude/agents/Or add it as a git submodule:
# Add as submodule in your .claude directory
cd ~/.claude/agents
git submodule add https://github.com/aplaceforallmystuff/claude-agent-borg.git borg
ln -s borg/agents/borg.md borg.md- Use it when you find a Claude setup on GitHub and want its skills without copying blindly:
Clone and assimilate https://github.com/username/their-claude-setup. - Use it to analyze a local
.claude/directory and get an assimilation report:Assimilate ~/path/to/external-claude-setup. - Use it to absorb one specific feature rather than a whole system:
Borg: absorb the specific-skill from ~/path/to/external-system. - Use it to compare your system against another and surface gaps:
Compare my system with ~/path/to/external-setup and identify gaps. - Use it to review what has already been pulled in:
What have we assimilated recently?.
The Borg agent runs a systematic 5-phase assimilation process:
| Phase | Action | What It Does |
|---|---|---|
| 1. Reconnaissance | Maps external system | Identifies skills, agents, workflows, dependencies |
| 2. Gap Analysis | Compares capabilities | Scores features by value, flags conflicts |
| 3. Adaptation Planning | Maps conventions | Converts external patterns to your standards |
| 4. Assimilation | Integrates features | Creates adapted artifacts in your system |
| 5. Documentation | Updates changelogs | Tracks what was absorbed and from where |
Infrastructure Compatibility Check
- Verifies all dependencies exist before porting
- Substitutes equivalent services automatically
- Flags missing requirements clearly
Smart Assimilation Decisions
- PROCEED - Genuinely new capability
- EXTEND - Add to existing artifact
- COMPOSE - Combine existing tools
- ITERATE - Extract pattern only
- BLOCK - Would create problematic duplication
Convention Adaptation
- Converts TitleCase to kebab-case
- Maps directory structures
- Adapts to your branding and aesthetic
- Preserves source attribution
When the external system has a generic version of something you've specialized, the agent can extract just the pattern:
Example:
- External: Generic "Research" skill that searches 5 sources in parallel
- Your system: 8 specialized librarian agents (tech, craft, security, etc.)
- ITERATE decision: Don't replace your agents. Instead, extract the "parallel spawn + query decomposition" pattern and apply it to YOUR agents.
This prevents losing your domain-specific implementations while still gaining valuable patterns.
Every assimilation updates CHANGELOG-borg.md:
## [2026-01-07] Assimilation: Image Prompt Skill
**Source:** awesome-claude-setup
**Source URL:** https://github.com/example/awesome-claude-setup
### Added
- skill: image-prompt - AI image prompt generation
### Adapted
- Replaced Replicate with OpenAI DALL-E
- Converted naming conventions to local standards
- Adapted to local infrastructure
### Dependencies
- OpenAI API key required
### Usage
Run: `generate image prompt for [description]`Scenario: You find a Claude setup on GitHub with interesting skills for generating image prompts.
User: Clone and assimilate https://github.com/example/awesome-claude-setup
Borg:
[Phase 1: Reconnaissance]
Cloning repository...
Scanning repository structure...
Found:
- 12 skills
- 8 agents
- 3 commands
- 2 hooks
[Phase 1.5: Infrastructure Check]
Checking dependencies...
✅ OpenAI API - available
✅ Anthropic API - available
⚠️ Replicate API - MISSING, will substitute with OpenAI DALL-E
❌ Custom MCP server "image-gen" - MISSING, will skip related features
[Phase 2: Gap Analysis]
HIGH VALUE Features:
- image-prompt skill (AI image prompt generation) - HIGH VALUE
Adaptation: None of your skills handle image prompting
MEDIUM VALUE Features:
- research skill (parallel query spawning) - MEDIUM VALUE
Note: You have 8 specialized librarian agents
Recommendation: ITERATE - extract parallel pattern, don't replace agents
[Phase 3: Adaptation Planning]
image-prompt Skill:
- Source: /tmp/awesome-claude-setup/.claude/skills/image-prompt/
- Target: ~/.claude/skills/image-prompt/
- Adaptations needed:
* Convert naming to your conventions
* Replace Replicate API calls with OpenAI DALL-E
* Adapt to your local infrastructure
Ready to assimilate? [Y/N]
The agent adapts to YOUR system's conventions. Common mappings:
| External | Your Standard |
|---|---|
.claude/Skills/ |
~/.claude/skills/ |
| TitleCase dirs | lowercase dirs |
${EXTERNAL_VAR} |
Absolute paths |
You can customize convention mappings by editing the agent's <convention_mappings> section.
- Claude Code CLI
- Access to external
.claude/directories you want to analyze - Appropriate API keys for features you choose to assimilate
"Your culture adapts to service us"
This means:
- External patterns get converted to YOUR conventions
- You don't adopt their naming, they adopt yours
- Infrastructure gaps are bridged or substituted
- The collective (your system) improves without losing identity
Built by Jim Christian as part of the Claude Code agent ecosystem.
Inspired by the need to learn from the best setups without copying blindly. If you find a better way to do something, resistance is futile.
Related Projects:
- claude-draft-reviewer - Reviews AND fixes drafts
- claude-think-first - Mental models before implementation
- claude-creation-guard - Prevents duplicate artifacts
MIT — see LICENSE.


