external-agent 0.3.0
Delegate to another coding agent (Codex by default; Gemini CLI, Ollama or LM Studio models, or Claude Code) from Claude Code, sharing the same conversation: /agent, /agent-review, and a tool Claude can call like a sub-agent.
Install
/plugin marketplace add samuelbfernandes/moddeck /plugin install external-agent@moddeck
Run both lines inside Claude Code.
Hooks into
- Status line
- Tool calls
- Slash commands
- Model tools
Also in samuelbfernandes/moddeck
- repro-log 0.1.0 Keeps a reproducibility recipe for data work: the analysis commands that succeeded and the files you changed. /repro writes it to .moddeck/repro/ with git commit, versions, lock files, seeds and a runnable command list. Secrets are never written. 1
- data-guard 0.1.0 Asks before a git commit or git add takes in large files, data files, personal-looking table columns or Jupyter notebooks with outputs. Only names are ever shown, never file contents. 1
- concise 0.1.1 While switched on, every prompt carries rules that shape Claude's replies so you can act on them: the next action first, numbered steps, state restated each turn, no tangents, specific time estimates. Off until you type /concise on. 1
- auto-test 0.1.1 Runs the project's tests once at the end of a turn in which Claude edited files, and shows a green or red result. Off until you type /auto-test on. 1
- big-file-guard 0.1.0 Saves tokens: when the model reads a large data or log file whole, it gets the first lines and a note, unless it asks for a range itself. 1
- branch-janitor 0.4.5 Watches your git branches by itself (nothing to run) and notices when a branch's work is done (merged, squash-merged, pull request merged), even if you merged in a plain terminal. Checks it is safe to delete and offers to delete it. Never deletes without a click; the remote copy is a second, sepa... 1
- checkpoint 0.1.0 One-click safety snapshots of your working tree (untracked files too) that never touch your branch, index or stash. /checkpoint, list, diff, restore, drop. Restore shows what would be lost and asks first. 1
- command-guard 0.3.3 Holds risky Bash commands (rm -rf, git reset --hard, force push, DROP TABLE ...), shows exactly what they would destroy, and asks first. A safety net, not a permission system. 1
- decision-log 0.1.0 /decision <text> [-- rationale] adds a dated row to the project's docs/DECISIONS.md (the agents-md-scaffold table); /decision-log shows the last 20. Writes only that file, only when asked. 1
- diagram 0.1.1 Draws the last mermaid diagram from Claude's answer as box art in a pane: run /diagram to open it. Flowcharts, sequence diagrams and pie charts; offline, no model tokens. 1
More status line mods
- jev-auto-mode 0.1.1 A permission layer driven by a JSON policy: allow, ask or deny every tool call (Bash, edits, web, MCP, subagents, skills), every slash command or skill you type, and every skill preloaded into a subagent. Rules decide first (deny > ask > allow, compound Bash commands split); what no rule covers c... 33.4k
- cache-clock 0.1.1 A prompt-cache line under your status line, from Claude Code's own cache figures: time left while warm, hit rate and misses, and how many tokens the next message re-caches once cold. /cache-clock setup adds it. 473
- secrets-veil 0.0.11 Masks secret values in tool results before the model reads them: named vendor variables, value-shape patterns, and high-entropy tokens. No reveal path. 302
- lemo-meter 0.2.0 用量横条:输入框上方显示上下文、5 小时额度、花费和 git 分支;状态栏显示消息编号(已有自己的状态栏时默认关闭);点「检查」才联网查新版本 / Usage meter: context, 5-hour quota, cost and git branch above the prompt; a status line with the message number (off by default if you have your own); checks for updates online only when you press Check 71
- glowup 1.2.0 mod + skill A glow-up for Claude Code: a cockpit pane for changes, subagents and context, an activity band, and shareable themes. 41