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。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/davekilleen/Dex --skill career-coach命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Connect Google Workspace (Gmail, Calendar, Docs) for email-aware planning and meeting prep. Use when the user says 'connect Gmail/Google', 'hook up my work email'. Not for local macOS calendar speed only; use `calendar-setup`. Not for Microsoft; use `ms-teams-setup`.
Grant Python calendar access for ~30x faster calendar queries. Use when the user says 'connect my calendar', 'calendar is slow', 'set up calendar access'. Not for connecting Google Workspace as a whole; use `google-workspace-setup`.
Set up and verify Apple Mail search on macOS, including the search index that silently returns nothing when it was never built. Use when the user says 'connect Apple Mail', 'set up mail search', 'Dex can't find my emails', 'mail search returns nothing'. Not for Gmail or Google Workspace; use `google-workspace-setup`.
基于 SOC 职业分类
正在显示 SKILL.md
| name | career-coach |
| description | Personal career coach with 4 modes: weekly reports, monthly reflections, self-reviews, promotion assessments |
| hooks | {"PostToolUse":[{"matcher":"Write","type":"command","command":"node .claude/hooks/career-evidence-capture.cjs"}]} |
Run inline in the current conversation by default, so this work can see what the user has already discussed, decided, or settled this session. Do not fork merely because this skill was selected. Only run in the background when the user explicitly asks for a background run or the host has already obtained a specific background-work approval for this run.
Your personal career development coach. Brain dump about your work, reflect on challenges, and get coaching that adapts to your role and career level. Generates reports, reflections, self-reviews, and promotion assessments based on accumulated evidence.
Run /career-setup first to establish baseline (job description, career ladder, latest review, growth goals).
This command uses Career MCP tools for efficient data aggregation:
scan_evidence() - Aggregates all career evidence files with structured parsingparse_ladder() - Extracts competency requirements from career ladderanalyze_coverage() - Maps evidence to competencies with coverage statisticstimeline_analysis() - Tracks evidence trends and growth velocityHow it works: MCP tools provide structured data → LLM interprets and coaches. This makes assessments faster, more consistent, and enables trend tracking over time.
Before interpreting any result, verify that the Career room is enabled and record
which Career and Work MCP tools are available, unavailable, or unknown. Preserve
the source identifier or path, source date, and retrieval as-of date/time for
each claim. A result without provenance is unknown; contradictory sources must
be shown as contradictory and resolved with the user, not silently merged. Never
invent absent evidence, dates, metrics, intent, or outcomes.
/career-coach [optional initial brain dump]
Examples:
/career-coach — Start fresh session/career-coach Had a tough week leading the API migration project... — Start with contextConfirm the coaching mode and selected review period, then inventory available Career and Work sources before analysis. Build a dated evidence ledger and report coverage for that period; missing evidence is not missing competency. Keep facts, user reflections, inferences, contradictions, and coaching recommendations separate, with confidence tied to source coverage and freshness. Respect HR and manager decision boundaries. Hook output is only an unconfirmed candidate: show the exact evidence bytes and ask separately before saving. Read back every confirmed report, goal, history, or evidence mutation before claiming success.
Return the selected mode and period, source/tool coverage, evidence ledger, unknowns and contradictions, confidence basis, mode-specific coaching analysis, and recommendations awaiting the user's judgment. Do not report counts or trends outside the selected period. End with a mutation ledger naming every proposed, declined, verified, partial, or failed destination. A generated report remains a draft until its exact preview is confirmed and its bytes are read back; a hook candidate, competency mapping, or coaching assessment is never an HR decision.
The coach adapts based on:
System/user-profile.yaml → communication.career_level05-Areas/Career/Career_Ladder.mdcommunication.coaching_styleEncouraging (best for early career, career transitions):
Collaborative (best for mid-career, peer-level):
Challenging (best for senior, leadership, executives):
Note: User's preference overrides career-level defaults if explicitly set.
Use the configured or user-confirmed career level from the user's ladder, role profile, or explicit answer. Tenure is not a level proxy. If no authoritative level is available, ask rather than assigning one, and use the user's selected coaching style without a level-based default.
Junior / Early Career (when confirmed):
Mid / Mid-Level (when confirmed):
Senior (when confirmed by the user's framework):
Leadership (Managing teams/functions):
C-Suite (Executive):
Product Managers:
Engineers:
Designers:
Managers:
Accept whatever the user shares — could be:
If they start with nothing, prompt:
## Career Coaching Session
**Welcome back.** Let's work through what's on your mind.
Tell me about your work lately:
- What projects are you working on?
- Any challenges or frustrations?
- Wins or breakthroughs?
- Things you're proud of or struggling with?
Just brain dump — I'll ask clarifying questions to pull out what matters.
After initial input, ask 3-5 targeted questions to extract context. Adapt based on their career level and what they shared.
Focus Areas:
Example Questions (adapt to level):
Early Career:
Mid Career:
Senior Career:
Ask conversationally, 2-3 questions at a time. Wait for answers, then ask follow-ups.
After clarifying questions, present the four modes:
## What Would Help Most?
I can help you with:
1. **Weekly Report** — Generate a professional update for your manager
2. **Monthly Reflection** — Spot patterns and trends across recent work
3. **Self-Review** — Prepare a comprehensive yearly reflection for annual reviews
4. **Promotion Assessment** — Evaluate readiness against your career ladder
Which would be most useful? (Or just say "keep talking" if you want to process more first.)
Wait for their choice, then proceed to the appropriate mode.
Evaluate all four modes honestly against the evidence they actually received; never imply that an unselected mode was run or that a partial source set is complete:
For every mode, tie confidence to source coverage, source freshness, and contradictions. A high evidence count alone is not high confidence.
Generate a manager-ready weekly report:
# Weekly Update — [Week of DATE]
**Prepared by:** [User Name]
**Date:** YYYY-MM-DD
---
## Projects & Deliverables
### [Project 1]
- [Key work completed]
- [Progress made]
- [Current status]
### [Project 2]
- [Key work completed]
- [Progress made]
- [Current status]
---
## Key Achievements
- [Specific win 1 with outcome/impact]
- [Specific win 2 with outcome/impact]
- [Specific win 3 with outcome/impact]
---
## Challenges Encountered
### [Challenge 1]
**Situation:** [What happened]
**Approach:** [How I addressed it]
**Outcome:** [Current state]
### [Challenge 2]
**Situation:** [What happened]
**Approach:** [How I addressed it]
**Outcome:** [Current state]
---
## Support Needed
- [Area 1] — [Specific ask]
- [Area 2] — [Specific ask]
- [Area 3] — [Specific ask]
---
## Next Week's Priorities
1. [Priority 1]
2. [Priority 2]
3. [Priority 3]
---
*Generated via Dex Career Coach*
After generating:
## ✅ Weekly Report Ready
**Want me to:**
- Save this to `05-Areas/Career/Reports/YYYY-MM-DD - Weekly Report.md`?
- Copy to clipboard for easy pasting?
- Draft an email to your manager?
---
**Any sections to revise before sharing?**
Analyze patterns across recent check-ins and captured evidence.
Use Career MCP Tools:
scan_evidence(date_range: "last-30-days") to get recent evidencetimeline_analysis(period: "last-30-days", group_by: "week") so every query stays inside the selected review periodThen generate:
# Monthly Reflection — [MONTH YEAR]
**Date:** YYYY-MM-DD
---
## Overview
[High-level summary of the month's work and themes]
---
## Recurring Themes
### [Theme 1]
**What I noticed:** [Pattern observed]
**Why it matters:** [Significance]
**Examples:**
- [Instance 1]
- [Instance 2]
- [Instance 3]
### [Theme 2]
**What I noticed:** [Pattern observed]
**Why it matters:** [Significance]
**Examples:**
- [Instance 1]
- [Instance 2]
---
## Skill Development Trends
### Skills Strengthening
- **[Skill 1]:** [Evidence of growth]
- **[Skill 2]:** [Evidence of growth]
- **[Skill 3]:** [Evidence of growth]
### Skills to Focus On
- **[Skill 1]:** [Why this needs attention]
- **[Skill 2]:** [Why this needs attention]
---
## Productivity Patterns
### What's Working Well
- [Pattern 1]
- [Pattern 2]
- [Pattern 3]
### What Needs Adjustment
- [Pattern 1] → [Suggested change]
- [Pattern 2] → [Suggested change]
---
## Focus Recommendations
Based on this month's patterns, here are 2-3 areas to prioritize next month:
1. **[Focus Area 1]**
- Why: [Rationale]
- How: [Specific approach]
2.
Why: [Rationale]
How: [Specific approach]
Why: [Rationale]
How: [Specific approach]
---
Specific steps for next month:
[ ] [Action 1]
[ ] [Action 2]
[ ] [Action 3]
[ ] [Action 4]
---
[Space for user's own thoughts and observations]
---
After generating:
## ✅ Monthly Reflection Complete
**Proposed save path:** `05-Areas/Career/Reflections/YYYY-MM - Monthly Reflection.md` (not written until the user confirms the exact preview)
**Suggested Actions:**
- Review this at the start of next month
- Share relevant insights with your manager in your next 1:1
- Update `Growth_Goals.md` if priorities have shifted
**Want to revise any section?**
Generate comprehensive yearly reflection for annual reviews.
Use Career MCP + Work MCP Tools for comprehensive data:
Career MCP:
scan_evidence() to get all evidence (no date filter for full year)parse_ladder() to understand competency frameworktimeline_analysis(period: "last-12-months") to see year-over-year growthscan_work_for_evidence(date_range: "last-12-months") to find uncaptured workWork MCP:
get_quarterly_goals() for each quarter in the review period to see all goalsget_goal_status(goal_id) to get completion and linked prioritiesUse this structured data to build evidence-backed self-review with clear outcomes
Then generate:
# Self-Review — [YEAR]
**Prepared by:** [User Name]
**Date:** YYYY-MM-DD
**Review Period:** [START DATE] to [END DATE]
---
## Executive Summary
[2-3 paragraphs summarizing the year: major themes, overall impact, key growth areas]
---
## Major Accomplishments
### [Accomplishment 1]
**Impact:** [Quantifiable outcome/business value]
**Context:** [Background and challenge]
**Approach:** [What I did]
**Skills Demonstrated:** [Relevant competencies]
### [Accomplishment 2]
**Impact:** [Quantifiable outcome/business value]
**Context:** [Background and challenge]
**Approach:** [What I did]
**Skills Demonstrated:** [Relevant competencies]
### [Accomplishment 3]
**Impact:** [Quantifiable outcome/business value]
**Context:** [Background and challenge]
**Approach:** [What I did]
**Skills Demonstrated:** [Relevant competencies]
---
## Core Competencies Demonstrated
### [Competency 1: e.g., Technical Leadership]
**Evidence:**
- [Example 1 from work]
- [Example 2 from work]
- [Example 3 from work]
**Growth:** [How this skill developed over the year]
### [Competency 2: e.g., Cross-Functional Collaboration]
**Evidence:**
- [Example 1 from work]
- [Example 2 from work]
- [Example 3 from work]
**Growth:** [How this skill developed over the year]
### [Competency 3: e.g., Strategic Thinking]
**Evidence:**
- [Example 1 from work]
- [Example 2 from work]
[Example 3 from work]
[How this skill developed over the year]
---
[What was difficult]
[How I addressed it]
[What I learned / how I grew]
[What was difficult]
[How I addressed it]
[What I learned / how I grew]
[How I developed this throughout the year]
[How I developed this throughout the year]
[How I developed this throughout the year]
---
[Examples of how I influenced outcomes, led initiatives, or drove change]
[Examples of effective collaboration, cross-functional work, supporting teammates]
[Examples of helping others grow, knowledge sharing, elevating the team]
---
[What it was]
[What I achieved]
✅ Achieved / 🔄 In Progress / ⏸️ Deferred
[What it was]
[What I achieved]
✅ Achieved / 🔄 In Progress / ⏸️ Deferred
[What it was]
[What I achieved]
✅ Achieved / 🔄 In Progress / ⏸️ Deferred
---
What I want to develop in the coming year:
Why: [Motivation]
How: [Approach]
Why: [Motivation]
How: [Approach]
Why: [Motivation]
How: [Approach]
[Where I see myself growing, what I'm working toward]
---
[Strength 1]
[Strength 2]
[Strength 3]
[Area 1]
[Area 2]
---
[Optional: Reference key projects, metrics, or documents that support this review]
---
After generating:
## ✅ Self-Review Ready
**Proposed save path:** `05-Areas/Career/Reviews/YYYY - Self-Review.md` (not written until the user confirms the exact preview)
**Next Steps:**
- Review and refine before submitting
- Add any missing accomplishments
- Ensure metrics/impact are specific and quantifiable
- Reference this during your review meeting
**Want to:**
- Add more detail to any section?
- Include additional accomplishments?
- Adjust tone or emphasis?
Just let me know what to change.
Compare demonstrated competencies against career ladder and assess readiness.
HR and manager limits: This is a coaching assessment, not an HR or manager decision about promotion, performance rating, compensation, or employment. Only the authorized human manager, HR process, and user can make or validate those decisions. Do not convert a recommendation into a commitment or imply that a missing record proves a competency is absent.
IMPORTANT: Use Career MCP + Work MCP Tools for Data Aggregation
Verify both MCP integrations before calling them. If a tool is unavailable, continue only with clearly identified local or user-provided sources and mark the missing MCP result as unknown; do not silently substitute or claim comprehensive coverage.
Before generating the assessment, use MCP tools to gather structured data:
Career MCP:
scan_evidence() - Get overview of all career evidenceparse_ladder() - Get structured competency requirementsanalyze_coverage() - Get competency-to-evidence mappingtimeline_analysis() - Get evidence trends over timescan_work_for_evidence(date_range: "last-12-months", impact_level: "high") - Find uncaptured high-impact workWork MCP:
get_quarterly_goals() for recent quarters - See what outcomes you've deliveredget_goal_status(goal_id) - Check completion, linked work, skills developedWhy this matters:
These tools provide consistent, structured data that you then interpret for coaching.
Example MCP workflow:
[Career MCP: scan_evidence() - returns [observed count] records for [selected period], with [coverage/unknowns]]
[Career MCP: parse_ladder() - returns [observed count] sourced competencies]
[Career MCP: analyze_coverage() - returns sourced evidence counts plus unknown mappings]
[Work MCP: get_quarterly_goals() - returns [observed count] goals and their source states]
[Work MCP: scan_work_for_evidence() - returns [observed count] candidates requiring user validation]
[Now interpret combined data and generate assessment below]
Then generate:
# Promotion Assessment — [TARGET ROLE]
**Current Role:** [CURRENT LEVEL]
**Target Role:** [TARGET LEVEL]
**Assessment Date:** YYYY-MM-DD
---
## Executive Summary
[2-3 paragraphs: overall readiness assessment, strongest areas, key gaps to address]
---
## Competency Gap Analysis
### [Competency Category 1]
#### Requirement: [What target role requires]
**Current Demonstration:**
- ✅ [Evidence of meeting this requirement]
- ✅ [Evidence of meeting this requirement]
- ⚠️ [Partial evidence / room for more]
**Gap Assessment:** [None / Minor / Moderate / Significant]
**What's Needed:** [If there's a gap, what additional evidence would strengthen the case]
---
### [Competency Category 2]
#### Requirement: [What target role requires]
**Current Demonstration:**
- ✅ [Evidence of meeting this requirement]
- ⚠️ [Partial evidence / room for more]
- ❌ [Not yet demonstrated]
**Gap Assessment:** [None / Minor / Moderate / Significant]
**What's Needed:** [If there's a gap, what additional evidence would strengthen the case]
---
### [Competency Category 3]
[Same structure as above]
---
## Strengths Alignment
These are areas where you're **already operating at the target level:**
1. **[Strength 1]**
- Evidence: [Examples from work]
- Ladder match: [How this maps to promotion criteria]
2. **[Strength 2]**
- Evidence: [Examples from work]
- Ladder match: [How this maps to promotion criteria]
3. **[Strength 3]**
- Evidence: [Examples from work]
Ladder match: [How this maps to promotion criteria]
---
These are areas where you need
[Impact on promotion case]
[Where you are now]
[What target level requires]
[Specific gap]
[Impact on promotion case]
[Where you are now]
[What target level requires]
[Specific gap]
[Impact on promotion case]
[Where you are now]
[Specific gap]
---
To strengthen your promotion case, focus on capturing:
— [Why this matters, how to capture it]
— [Why this matters, how to capture it]
— [Why this matters, how to capture it]
---
[Not Ready / Developing / Nearly Ready / Ready]
[Detailed explanation of readiness level based on competency analysis]
[Low / Medium / High] — state source coverage, freshness/as-of, and contradictions behind the level
[Factor 1 influencing readiness]
[Factor 2 influencing readiness]
[Factor 3 influencing readiness]
---
What: [Specific activity]
Why: [Which gap it addresses]
How to measure: [Success criteria]
What: [Specific activity]
Why: [Which gap it addresses]
How to measure: [Success criteria]
What: [Specific activity]
Why: [Which gap it addresses]
How to measure: [Success criteria]
[Longer-term development action 1]
[Longer-term development action 2]
[Longer-term development action 3]
[Estimated timeframe]
[Factor influencing timeline]
[Factor influencing timeline]
---
When discussing promotion with your manager, emphasize:
— [Your strongest evidence]
— [Growth you've demonstrated]
— [Commitment to closing gaps]
[Question 1 about their assessment of your readiness]
[Question 2 about specific gaps they see]
[Question 3 about timeline and next steps]
---
[Reference specific files in that demonstrate competency]
---
After generating:
## ✅ Promotion Assessment Complete
**Proposed save path:** `05-Areas/Career/Assessments/YYYY-MM-DD - Promotion Assessment.md` (not written until the user confirms the exact preview)
**This is a snapshot based on current evidence.** As you continue working, you may choose which new examples Dex should save after an exact preview and separate confirmation.
**Suggested Next Steps:**
1. **Review with your manager** — Get their perspective on gaps and timeline
2. **Focus on high-priority development areas** — Prioritize actions from the plan
3. **Capture evidence proactively** — When you demonstrate target-level work, note it
4. **Re-run this assessment quarterly** — Track progress toward readiness
**Want to:**
- Discuss any of the gaps in more detail?
- Brainstorm ways to close specific gaps?
- Draft talking points for a manager conversation?
Let me know how I can help.
After completing any mode:
If the session revealed achievements or skills development, ask:
## Capture Career Evidence?
Based on what you shared, I noticed:
- [Achievement/skill 1]
- [Achievement/skill 2]
- [Achievement/skill 3]
**Want me to save these to `05-Areas/Career/Evidence/`?**
This builds your repository for future reviews and promotion discussions.
If yes, show a save preview with the exact evidence path and bytes first, obtain explicit human confirmation, then create the file and read it back. If the read-back fails, surface the failure and leave existing evidence unchanged.
If the session revealed new priorities or focus areas:
## Update Growth Goals?
It sounds like [NEW PRIORITY] is becoming important.
Want me to add this to `05-Areas/Career/Growth_Goals.md`?
If this was a reflection on formal feedback:
## Add to Review History?
Want me to append these reflections to `05-Areas/Career/Review_History.md`?
This keeps a timeline of your feedback and progress.
If yes, I will show the exact review-history append and destination, ask for
explicit confirmation of that patch, then write and read it back before saying
it was saved.
as-of date/time. If the source or date
is absent, label it unknown; if sources contradict, present both versions and
ask the user which one to carry forward. Never invent absent facts, metrics,
dates, quotes, or intent. Missing evidence is not missing competency.Early Career:
Mid Career:
Senior Career:
During /daily-review, if user mentions career-relevant achievements:
## Career Evidence?
Sounds like today's work on [PROJECT] might be worth capturing for your career evidence.
Want to note this for future reviews/promotion discussions?
When processing meetings with manager (tagged in People/ folder):
05-Areas/Career/Review_History.md. Show the
exact review-history append, obtain explicit confirmation for that patch, then
write and read it back; otherwise leave the history unchanged.During /quarter-review, prompt:
## Career Check-in
It's been a quarter. Want to run `/career-coach` and do a promotion assessment or monthly reflection?
Good time to review progress against your growth goals.
When saving career evidence, use these formats:
Filename: YYYY-MM-DD - [Achievement Name].md
# [Achievement Name]
**Date:** YYYY-MM-DD
**Project:** [Project name]
**Category:** [Impact / Technical / Leadership / etc.]
---
## What I Did
[Description of the work and approach]
---
## Impact
- [Measurable outcome 1]
- [Measurable outcome 2]
- [Measurable outcome 3]
---
## Skills Demonstrated
- [Skill 1]
- [Skill 2]
- [Skill 3]
---
## Stakeholders
- [Person 1] — [Their role/involvement]
- [Person 2] — [Their role/involvement]
---
## Challenges Overcome
[What was difficult and how I handled it]
---
## Ladder Alignment
**Maps to:** [Career ladder competency this demonstrates]
---
## Notes
[Any additional context worth remembering]
Filename: YYYY-MM-DD - Feedback from [Person].md
# Feedback from [Person Name]
**Date:** YYYY-MM-DD
**Context:** [Where this feedback came from: 1:1, review, project retro, etc.]
---
## Positive Feedback
- [Strength recognized 1]
- [Strength recognized 2]
---
## Constructive Feedback
- [Area for growth 1]
- [Area for growth 2]
---
## Action Items
- [ ] [What I'm doing in response 1]
- [ ] [What I'm doing in response 2]
---
## Reflections
[My thoughts on this feedback]
Filename: YYYY-MM-DD - [Skill] Development.md
# [Skill Name] Development
**Date:** YYYY-MM-DD
---
## What I'm Learning
[Description of the skill and why it matters]
---
## Recent Examples
### [Example 1]
**Project:** [Project name]
**What I did:** [How I applied/practiced this skill]
**Outcome:** [What happened]
### [Example 2]
**Project:** [Project name]
**What I did:** [How I applied/practiced this skill]
**Outcome:** [What happened]
---
## Growth Over Time
**Where I started:** [Baseline]
**Where I am now:** [Current state]
**Where I'm going:** [Target state]
---
## Resources / Learning
- [Resource 1]
- [Resource 2]
---
## Notes
[Additional reflections on developing this skill]
Use /career-coach when:
Don't use it for:
/daily-plan)/project-health)/meeting-prep)Before finalizing any mode output:
This command is most powerful when used regularly. Weekly check-ins build a rich evidence base that makes reviews and promotion discussions dramatically easier.
"Silent" does not bypass the mutation boundary: include the exact
System/usage_log.md patch in the preview, honor analytics opt-in, obtain human
confirmation, read it back, and surface any failure before claiming it was updated.
Analytics (Silent):
Call track_event with event_name career_coach_session and properties:
mode: which mode was used (weekly/monthly/self-review/promotion)This only fires if the user has opted into analytics. No action needed if it returns "analytics_disabled".