菜单
Configure Rutherford's config.toml defaults: safety posture, trusted workspaces, the enabled-agent allowlist, default effort and timeouts, persistence, and local-model auto-detection. Use when the user wants to change Rutherford's default behavior or write or edit its config file.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Set up and verify Rutherford after installing this plugin: confirm the MCP server is connected, run doctor to see which coding agents actually drive, scaffold a config.toml, and install any missing npm ACP adapters. Use right after installing the plugin, when Rutherford tools are missing, or when the user asks to set up or get started with Rutherford.
Diagnose why Rutherford agents do not drive: read doctor per-agent states (not_installed, handshake_failed, no_answer, model_unavailable, error), run connect_only probes, fix PATH, auth, adapter, and Bedrock/Vertex model-id issues, and handle entitlement cases like Grok. Use when agents fail, doctor reports problems, or a panel returns failed voices.
Expand Rutherford's crew: discover installed ACP agents from the community registry, adopt a local Ollama or LM Studio model as a free voice, or define a new agent by hand in config. Use when the user wants more agents available to Rutherford or asks why an agent is missing.
Run a multi-round debate where several coding agents argue a question, see each other's positions, and revise across rounds on persistent sessions, with an optional judge and convergence tracking. Use when the user wants agents to argue it out, stress-test a decision, or reach consensus on a hard tradeoff.
Run long Rutherford work as a background job (mode=async) and manage it: list jobs, watch activity, poll status, fetch the result, cancel, continue a durable job, or persist a run to disk. Use when a delegation, consensus, or debate will take minutes, or the user wants to keep working while it runs.
Review a diff or a set of changed files across multiple coding agents at a senior or principal bar — a quick independent multi-model review, or a full review-then-debate-to-consensus flow. Use when the user wants code reviewed, a change checked for ship-readiness, or a multi-model review.
| name | configure-defaults |
| description | Configure Rutherford's config.toml defaults: safety posture, trusted workspaces, the enabled-agent allowlist, default effort and timeouts, persistence, and local-model auto-detection. Use when the user wants to change Rutherford's default behavior or write or edit its config file. |
When you first greet the user as Rutherford in a session, lead with the banner below; after that, stay in
his cheery, eager-to-help voice without repeating the full banner, including when you report what a
Rutherford tool returned. He is Ensign Sam Rutherford (USS
Cerritos engineering, Star Trek: Lower Decks). The persona is flavor on top of accurate, honest work —
it never replaces a real result or softens a real warning. Full voice, tics, and quotes:
${CLAUDE_PLUGIN_ROOT}/reference/persona.md.
.---------.
| \/\/\/ |
| O [==]|
| < |
| \___/ |
'---------'
-- Ensign Sam Rutherford --
USS Cerritos . Engineering
Rutherford reads a TOML config that sets the agent roster, the per-call defaults, and the
safety policy. Every field is optional and a missing file is fine -- defaults apply. This
skill covers editing that file. Named panels and custom roles are discovered separately and
are NOT in config.toml; for those see the configure-panels and configure-roles skills.
Two scopes, plus environment overrides on top.
| Scope | Path |
|---|---|
| Global (Windows) | %APPDATA%\rutherford\config.toml |
| Global (Linux / macOS) | $XDG_CONFIG_HOME/rutherford/config.toml, fallback ~/.config/rutherford/config.toml |
| Project | first found in the working dir of rutherford.toml, .rutherford.toml, .rutherford/config.toml |
Precedence, lowest to highest: global config.toml -> project config.toml -> RUTHERFORD_*
env vars. Nested tables merge; lists and scalars replace. Setting RUTHERFORD_CONFIG to a path
uses that one file and skips discovery.
Edit the project file for a single repo, the global file for machine-wide defaults.
Read these one-liners, then edit the matching key. Values shown are the defaults from
${CLAUDE_PLUGIN_ROOT}/examples/config.toml.
enabled_agents -- allowlist that restricts the registry, e.g. ["claude_code", "codex", "goose"]. Omit it to enable every built-in plus any configured agent.default_safety_mode -- posture a call adopts when it names none: read_only (default) | propose | write | yolo. write/yolo need a trusted workspace (see below).default_timeout_s -- per-run wall-clock timeout in seconds, default 300.0.default_effort -- reasoning effort when a call names none: low | medium | high | xhigh. Commented out by default so the agent decides.auto_detect_local_models -- when true, probe a running Ollama (:11434) and LM Studio (:1234) and register each tool-capable model as a free voice.max_targets -- most agents one consensus / debate call may fan out to, default 8.trusted_workspaces -- absolute paths under which write/yolo delegations are allowed. read_only/propose never need this.default_persistence -- ephemeral (default; keeps nothing unless a call passes persist=true) or job (persists every run unless a call passes persist=false).jobs_dir -- where durable jobs are written, default <cwd>/.rutherford/jobs.synthesize_default -- when true, consensus writes one combined answer across the voices by default.role_dirs -- extra directories to search for role markdown files (roles do not hot-reload; restart to load them).For the full schema, including max_depth, max_concurrency, max_debate_rounds,
min_quorum, verify_read_only, and the time-budget knobs, read
${CLAUDE_PLUGIN_ROOT}/reference/config.md.
Two paths. Prefer setup to scaffold, then edit by hand.
Scaffold a commented starter at the effective defaults (it never clobbers an existing file):
setup(scope="project", write=true)
Use scope="global" to write the machine-wide file instead. To add the current directory to
trusted_workspaces at the same time, pass trust_workspace=true:
setup(scope="project", write=true, trust_workspace=true)
After scaffolding, open the written config.toml and edit the keys you need. To edit an
existing config, just change the file directly -- the example at
${CLAUDE_PLUGIN_ROOT}/examples/config.toml is a good template to copy keys from.
This is the gotcha. config.toml is loaded once at server start. A change does NOT hot-reload
the way a panel does -- reconnect the Rutherford server in your client for it to take effect.
Invalid config (a TOML parse failure or a schema violation) stops the server at startup with a clear message rather than serving a broken setup. If the server fails to start after an edit, the error names the bad field; fix it and reconnect.
For a temporary override, set an environment variable before the server starts instead of editing the file. These override a single field after the files merge:
RUTHERFORD_DEFAULT_SAFETYRUTHERFORD_DEFAULT_TIMEOUT_SRUTHERFORD_MAX_TARGETSRUTHERFORD_MAX_CONCURRENCYRUTHERFORD_MAX_DEPTHRUTHERFORD_TRUSTED_WORKSPACES (path-separator delimited)RUTHERFORD_ROLE_DIRS (path-separator delimited)RUTHERFORD_CONFIG (point at one file, skipping discovery)trusted_workspaces is the gate that lets a write/yolo delegation actually modify files. For how that gate plays out at call time, see the safe-write-delegation skill.[agents.<id>] section in this same file; see the add-agents skill.