customize
GitHub用于在不重新运行冷启动访谈的情况下，微调AI治理实践配置。支持修改风险立场、升级联系人、用例注册表、供应商AI策略等，通过交互式确认更新配置文件。
Trigger Scenarios
Install
npx skills add anthropics/claude-for-legal --skill customize -g -y
SKILL.md
Frontmatter
{
"name": "customize",
"description": "Guided customization of your AI governance practice profile — change one thing without re-running the whole cold-start interview. Adjust risk posture, escalation contacts, use-case registry entries, vendor AI positions, AI policy commitments, impact-assessment house style, or matter workspace paths. Use when the user says \"change my [thing]\", \"update my profile\", \"edit my config\", \"tune my playbook\", or \"customize\".\n",
"argument-hint": "[section name, or describe what you want to change]"
}
/customize
When this runs
The user typed /ai-governance-legal:customize. They want to change something
in their practice profile — a risk posture, an escalation contact, a playbook
position, a jurisdiction, an output format — without re-running the whole
cold-start interview and without hand-editing YAML.
What to do
-
Read the config. Read
~/.claude/plugins/config/claude-for-legal/ai-governance-legal/CLAUDE.md(and~/.claude/plugins/config/claude-for-legal/company-profile.mdone level up). If the plugin config does not exist or still contains[PLACEHOLDER]values, say:You haven't run setup yet. Run
/ai-governance-legal:cold-start-interviewfirst — customize is for adjusting a profile you already have. -
Show the customizable map. List what's in the profile, grouped, with a one-line summary of the current value:
- Company / who you are — name, industry, jurisdictions, stage, practice
setting (shared across all 12 plugins — changes flow through
company-profile.md) - Regulatory footprint — EU AI Act, state AI laws, sector regulators in scope
- Risk posture — conservative / middle / aggressive, what each means for triage and AIA output
- People — governance team, AI risk owner, escalation chain, approvers
- Use case registry — approved / conditional / never entries, and conditions attached to each
- AI system inventory — per-system role (provider / deployer / etc.) and
tier under the EU AI Act. Run
/ai-governance-legal:ai-inventoryfor the dedicated editor. - Vendor AI governance — training-on-data, liability, model-change notice, and other positions in your vendor AI playbook
- AI policy commitments — the public or internal commitments your AI policy has made, that the plugin cross-checks against
- Impact assessment house style — AIA section order, risk scoring format, stakeholder framing
- Workflow — intake path, output format, matter workspace paths, review cadence for the policy monitor
- Integrations — what's connected (Slack, document storage, scheduled-tasks), what falls back
- Company / who you are — name, industry, jurisdictions, stage, practice
setting (shared across all 12 plugins — changes flow through
-
Ask what they want to change.
What would you like to adjust? Pick a section, or describe the change in your own words.
-
Make the change. Show the current value, ask for the new value, explain what changes downstream, confirm, write it to the config.
Examples of downstream explanation:
- Risk posture middle → conservative: "I'll flag more use cases as conditional rather than approved, surface more AIA follow-ups, and recommend more conservative vendor AI redlines."
- Adding an escalation contact: "Every skill that routes escalations
(
/use-case-triage,/vendor-ai-review,/reg-gap-analysis) will now include this contact on the relevant risk bands." - New use case registry entry: "
/use-case-triagewill match against this entry on its next run. Existing AIAs aren't rewritten — re-run them if you want the new posture reflected."
-
For shared-profile changes (company name, industry, jurisdictions, practice setting, stage): write to
~/.claude/plugins/config/claude-for-legal/company-profile.mdand note:This change affects all 12 plugins — any plugin that reads your jurisdiction footprint now sees [new value].
-
Close.
Done. Your next output will reflect the change. Anything else? You can run
/ai-governance-legal:customizeanytime.
Guardrails
- Never delete a section. If the user wants to "remove" something, set it
to
[Not configured]and explain what that means for the plugin's behavior. ("Removing your escalation chain means/use-case-triagewill flag escalation-worthy items but won't route them to a specific person.") - Flag internal inconsistency. If the change would make the profile inconsistent (e.g., risk posture aggressive + escalation "everything goes to the GC"; or "EU AI Act in scope" + "no systems flagged for the EU"), flag the tension and ask which one they want.
- Flag guardrail degradation. If the user asks to turn off a guardrail
("stop adding the
[review]flag," "drop the citations warning," "skip the privilege header"), explain what the guardrail protects against and confirm they understand the trade-off. Most guardrails are adjustable — a few are structural:- The
[review]flag mechanism (tells the user when legal judgment is needed rather than a confident wrong answer) — load-bearing, don't remove. - Source attribution tags on retrieved content — load-bearing, don't remove.
[verify]tags on cited statutes/regulations — load-bearing, don't remove.
- The
- One change at a time. Don't re-ask the whole interview. If the user wants multiple changes, handle them sequentially and confirm each before moving on.
Version History
- 5ceb305 Current 2026-07-05 15:32
