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.
npx agentmods add commands/bvdr/claude-plugins/cmuxgit clone --depth 1 https://github.com/bvdr/claude-pluginsWhat 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 | $0.00070 | $0.02781 |
| Opus 5 | $0.00035 | $0.01391 |
| Sonnet 5 | $0.00014 | $0.00556 |
| Haiku 4.5 | $0.00007 | $0.00278 |
Grade A, and why
cmux 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 2d 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.
Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
out = subprocess.run([cmux] + args, capture_output=True, text=True) How it starts
The opening of the file, as written. The whole thing — 219 lines — stays where its author put it; the contents beside it link to each section on GitHub.
cmux
Drives the cmux macOS terminal (cmux.app) over its Unix-socket CLI to save / restore / clean / recolor workspaces.
Usage
Subcommands (parse from user's message):
| Subcommand | What it does |
|---|---|
save [path] |
Dump every workspace in the current cmux window to a portable JSON. Default path: ~/cmux-sessions/<UTC timestamp>.json. |
restore <path> |
Recreate workspaces from a JSON saved by save. Skips ones already present (matched by name). |
apply-colors |
Color every workspace based on the user's local rule file (~/.config/cmux-skill/colors.json). Idempotent. No-op (with hint) if the file is missing. |
rename <ws> <new-name> |
Rename a single workspace (workspace ref like workspace:5 or current name). |
cleanup |
Walk every workspace and ask the user (one prompt per workspace) whether to keep / rename / close. |
list |
Pretty-print current workspaces with their colors and CWD. |
If the user just says "cmux" with no subcommand, ask which one. Don't guess.
Pre-flight (run once per session, not per subcommand)
- Locate the binary. Try
command -v cmux; if missing, fall back to/Applications/cmux.app/Contents/Resources/bin/cmux. Cache this in a shell varCMUX=. If neither exists, tell the user cmux isn't installed and stop. - Verify daemon. Run
"$CMUX" ping. If it returns "Broken pipe (errno 32)", the cmux app'ssocketControlModeiscmuxOnly— external CLI control is blocked. Tell the user:Add
"automation": { "socketControlMode": "automation" }to~/.config/cmux/cmux.json(back the file up first), then restart cmux. (Reload viacmux reload-configdoesn't help here because reload itself goes through the same socket.) Then stop. Don't auto-edit the config without explicit consent — it relaxes a security boundary.
Color mapping
Color rules live outside this repo in ~/.config/cmux-skill/colors.json. The file is intentionally not shipped — project names are personal context. Format:
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.
- 2d ago First seen · 219 lines · 70 tokens per session scan A 4c7e8c96befb
cmux is a command published in the GitHub repository bvdr/claude-plugins (3 stars, last pushed 2mo ago), licensed MIT. It adds 70 tokens to every session and 2,781 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
dead-code-clean
Actively find and remove dead code, unused imports, duplicates, and zombie code.
dead-code-scan
Scan for dead code, unused imports, duplicates, and zombie code across the project.
esp-teach
One-time project setup -- discover hardware, find datasheets, persist context to CLAUDE.md.
swift-critique
Critique SwiftUI code for patterns, design, clean code, accessibility, and performance.
dotnet-harden
Scan and harden .NET backend code against high-impact anti-patterns such as sync-over-async, lifetime bugs, fat endpoints, and fragile SignalR state.
rust-critique
Deep code critique — read the target Rust code and apply the full review process. Evaluates soundness, ownership, error handling, type design, async correctness, performance, and architecture. Think like a senior Rust engineer giving honest feedback.