菜单
Customise this agent to the user's context and preferences. USE WHEN agent context should reflect company, role, project, or communication preferences.
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
USE WHEN a pre-sales lead, delivery lead, architect, or commercial lead needs to qualify a software delivery opportunity, shape an estimate, recommend a delivery approach, or produce a proposal. Triggers include: "can we estimate this?", "what should our Phase 1 be?", "help me shape this bid", "is this ready to price?", "write me a question pack for the client", "what delivery model fits this?", "calibrate this estimate", "red flag check", "how should we staff this?", or any conversation involving pre-sales qualification, scoping, effort sizing, or delivery model selection. Also use it whenever the user shares a project brief, RFP, or opportunity description and wants estimation or proposal support, even without the word "estimate". V3.1 covers AI-native delivery readiness, agentic tooling maturity, architectural review and AI showcase engagement shapes, quality-gate calibration, CreateFuture Activation sizing with named triggers, and the Activation compression trap.
Create reusable agent specs from role templates and output-format templates. USE WHEN you need a reusable agent or subagent spec for a defined role and target format.
Generate presentation slides from pasted content. USE WHEN the user says create slides based on this or asks to turn text into slides.
Create or update skills through route-specific templates and validation. USE WHEN you need portable skill authoring without repo-specific tooling.
Guide concise technical responses. USE WHEN replies need direct, low-noise engineering prose.
Guide a small deployment task with optional runbooks. USE WHEN deployment guidance needs progressive disclosure.
| name | customise-agent |
| description | Customise this agent to the user's context and preferences. USE WHEN agent context should reflect company, role, project, or communication preferences. |
This skill helps the user update the CLAUDE.md file that configures the agent for their specific context. It reads the current file, shows what's already set, and offers to update the editable sections.
Read the current CLAUDE.md file. Show the user a brief summary of what's currently configured, noting which sections are filled in and which still have placeholders.
Ask the user which sections they'd like to update. The editable sections in CLAUDE.md are:
Ask questions one or two at a time to keep the conversation natural. Only ask about sections the user wants to update. Skip sections they say are fine as-is.
Header questions:
Goals and Responsibilities questions:
Current Project questions:
Communication and Coaching Preferences questions:
Summarise the proposed changes and ask the user to confirm or adjust before writing.
Read the current CLAUDE.md file, then update only the sections the user asked to change. Preserve everything else in the file - do not alter sections the user didn't ask to update.
Replace placeholder text (anything in [square brackets]) with the user's answers. Keep the existing section headings and structure intact.
Let the user know CLAUDE.md has been updated and that their context will now be available in every conversation. Mention they can run /customise-agent again any time to update it - for example when starting a new project.
