Custom slash commands that fill gaps in Claude Code's built-in feature set. Focused on cross-session context management, architectural decision tracking, and project metadata.
Claude Code has excellent tools for working within a session — /compact, --continue, --resume, /rewind. But it lacks tools for carrying high-signal context across fresh sessions. It also has no built-in support for lightweight architectural decision records or project-level metadata management.
These five commands address those gaps.
| Command | Purpose |
|---|---|
/cts |
Create a context transfer summary for session handoff |
/rehy |
Rehydrate context from latest CTS file into a fresh session |
/adr |
Document, list, and view architectural decisions |
/tag |
Add and manage project tags and metadata |
/tech-stack |
Manage and visualize technology stack across projects |
- Clone this repository:
git clone https://github.com/firstprinciplescg/claude-code-booster-pack.git
cd claude-code-booster-pack- Copy commands to your Claude Code directory:
cp -r commands/* ~/.claude/commands/- Restart Claude Code or reload your configuration.
The core workflow:
Start session -> /rehy -> work -> End session -> /cts
This ensures Claude never repeats work, retries failed approaches, or loses architectural decisions across context resets.
Creates a lean handoff document (30-second skim) capturing:
- Key decisions and WHY they were made
- Approaches tried that failed (and why)
- Current state, blockers, open questions
- Next steps and prerequisites
Excludes code/file contents (they're in git). Focuses on context Claude needs to continue effectively after context reset.
When to use: End of every session, before hitting context limits, or before /clear.
Efficiently restores Claude's working context in a fresh session:
- Checks if CLAUDE.md already in context (token optimization)
- Loads latest CTS file with decisions/next steps
- Provides structured summary of what's accomplished and what remains
Token efficiency: Restores full project context in ~1700 tokens vs ~7000 tokens manually explaining everything.
When to use: Start of every session, after /clear, resuming work after context reset.
Document, list, and view architectural decisions using a lightweight format:
- Date, title, rationale
- Tradeoffs and alternatives considered
- Reversibility assessment
Stores decisions in docs/decisions.md. Prevents context loss on the why behind technical choices.
When to use: Making significant architectural choices, documenting design decisions.
Add and update project tags and metadata for organization across projects:
- Client, GitHub repo, team members, billable status
- Tags for filtering and search
- Integrates with external automation (Make.com, etc.)
When to use: Categorizing projects, adding metadata for search and filtering.
Manage and visualize the technology stack across projects:
- Track technologies by category (Frontend, Backend, Database, DevOps, etc.)
- Record proficiency levels and learning status
- Export to CSV
When to use: Documenting technologies, tracking dependencies, planning migrations.
Previous versions of this booster pack included 17 specialized agents and 4 additional commands. These have been removed because Claude Code now provides them natively:
Agents removed: All 17 specialized agents (architecture-advisor, api-designer, code-review-specialist, debugging-assistant, etc.) are now built-in Claude Code subagent types. Installing custom copies creates duplicates that compete with the native versions.
Commands removed:
/ccc— replaced by built-in/doctor,/config, and the nativeclaude-code-cleanupagent/ship— replaced by built-in/diff,/security-review, and native git workflows/workstation-startand/workstation-end— replaced byclaude --continue,--resume, and built-in session persistence
Contributions are welcome! Please feel free to submit a Pull Request.
- Create a new
.mdfile incommands/ - Follow the existing command format
- Document the command in this README
MIT License - feel free to use and modify these tools for your projects.
Built for the Claude Code community to enhance productivity and workflow efficiency.
Questions or Issues? Open an issue on GitHub or contribute improvements!