Structured knowledge capture to extract team wisdom, lessons learned, and best practices for embedding in agent systems
Structured knowledge capture to extract team wisdom, lessons learned, and best practices for embedding in agent systems
A structured approach to capturing tacit knowledge, lessons learned, and team wisdom that goes beyond technical requirements to embed real human experience into agent systems.
"The best practices aren't in the documentation. They're in the heads of the people who've been in the trenches."
Wisdom Harvest recognizes that teams have invaluable knowledge that rarely gets written down - the workarounds, the gotchas, the "we learned this the hard way" insights. This skill extracts that wisdom and embeds it into the agent system.
Choose appropriate mode based on available time:
Ask structured questions from relevant categories:
Record team responses in structured format, asking follow-up questions to get specific examples and stories.
Organize captured wisdom into team-wisdom.json format with strengths, lessonsLearned, qualityDefinition, domainKnowledge, and toolWisdom sections.
memory MCP graph..agent/rules/.Update wisdom file after major incidents, retrospectives, or when new knowledge emerges.
Questions that surface what the team does well.
| # | Question | Purpose | ||-|| | 1 | What's one thing your team does that you're proud of? | Identify and preserve good practices | | 2 | What's a process or habit that makes your team effective? | Capture workflow wisdom | | 3 | What do people compliment your team on? | External validation of strengths |
Questions that surface hard-won knowledge.
| # | Question | Purpose | ||-|| | 1 | What's a mistake you've learned from that others should avoid? | Capture failure patterns | | 2 | What do you wish you'd known when you started this type of project? | Surface onboarding wisdom | | 3 | What's something that seems simple but is actually tricky? | Identify hidden complexity |
Questions that surface what "good" means.
| # | Question | Purpose | ||-|| | 1 | What does "done well" look like for this project? | Define quality bar | | 2 | What's non-negotiable for code quality? | Identify hard requirements | | 3 | How do you know when something is ready for production? | Capture readiness criteria |
Questions that surface domain-specific wisdom.
| # | Question | Purpose | ||-|| | 1 | What are the tricky parts of your domain that trip people up? | Capture domain gotchas | | 2 | What terminology does your team use that outsiders might not know? | Build glossary | | 3 | What business rules are not obvious from the code? | Surface hidden logic |
Questions that surface tooling and process insights.
| # | Question | Purpose | ||-|| | 1 | What tools or shortcuts make you more productive? | Capture efficiency tips | | 2 | What's your debugging process when things go wrong? | Document troubleshooting | | 3 | How do you handle releases or deployments? | Capture release wisdom |
For Express Lane onboarding. Ask ONE question:
╔══════════════════════════════════════════════════════════════╗
║ 🧠 QUICK WISDOM CAPTURE ║
╠══════════════════════════════════════════════════════════════╣
║ ║
║ What's ONE lesson you've learned from past projects ║
║ that you want your AI agents to remember? ║
║ ║
║ Examples: ║
║ • "Always write tests before refactoring" ║
║ • "Database migrations need rollback plans" ║
║ • "Never deploy on Fridays" ║
║ ║
╚══════════════════════════════════════════════════════════════╝
For Team Huddle. Ask THREE questions:
╔══════════════════════════════════════════════════════════════╗
║ 🧠 WISDOM HARVEST ║
╠══════════════════════════════════════════════════════════════╣
║ ║
║ Let's capture your team's hard-earned wisdom. ║
║ ║
║ Q1: What's one thing your team does well? ║
║ ║
║ Q2: What's a mistake you've learned from? ║
║ ║
║ Q3: What does "done well" look like for this project? ║
║ ║
╚══════════════════════════════════════════════════════════════╝
For comprehensive onboarding or dedicated knowledge capture sessions.
╔══════════════════════════════════════════════════════════════╗
║ 🧠 DEEP WISDOM HARVEST ║
╠══════════════════════════════════════════════════════════════╣
║ ║
║ We're going to spend some quality time capturing your ║
║ collective wisdom. This is an investment that pays off ║
║ every time your agents help you. ║
║ ║
║ We'll cover 5 areas: ║
║ 1. Strengths & Pride (what you do well) ║
║ 2. Lessons Learned (hard-won knowledge) ║
║ 3. Quality Definition (what "good" means) ║
║ 4. Domain Knowledge (your specific context) ║
║ 5. Tool & Process Wisdom (how you work) ║
║ ║
╚══════════════════════════════════════════════════════════════╝
{
"$schema": "team-wisdom-schema",
"version": "1.0",
"capturedDate": "2026-01-31",
"capturedDuring": "Team Huddle",
"participants": 4,
"strengths": [
{
"insight": "We always pair on complex features",
"category": "collaboration",
"source": "team"
}
],
"lessonsLearned": [
{
"insight": "Never deploy on Fridays",
"category": "process",
"severity": "high",
"source": "team"
}
],
"qualityDefinition": {
"doneWell": "Code is tested, documented, and reviewed",
"nonNegotiables": ["tests", "security review"],
"readinessCriteria": ["CI passes", "PR approved", "staging verified"]
},
"domainKnowledge": {
"gotchas": [
"Payment retries must use idempotency keys"
],
"glossary": {
"ACH": "Automated Clearing House - bank transfer system",
"Settlement": "When funds actually move between accounts"
},
"hiddenRules": [
"Refunds older than 90 days must go through manual review"
]
},
"toolWisdom": {
"productivityTips": [
"Use 'just db:reset' to rebuild test database"
],
"debuggingProcess": "Check logs → Reproduce locally → Add failing test → Fix",
"releaseProcess": "PR → Staging → Smoke test → Prod → Monitor 30 min"
}
}
The captured wisdom is referenced in agent skills:
# In .cursorrules or agent definitions
wisdom_integration:
- skill: code-reviewer
uses:
- lessonsLearned (check for known pitfalls)
- qualityDefinition (apply quality bar)
- skill: bugfix-workflow
uses:
- toolWisdom.debuggingProcess (follow team's debug approach)
- domainKnowledge.gotchas (warn about known issues)
- skill: grounding
uses:
- domainKnowledge.glossary (understand terminology)
- domainKnowledge.hiddenRules (respect business rules)
When a wisdom file exists, agents can include it in their context:
You are a code reviewer for this team. They value:
- {lessonsLearned}
- {qualityDefinition}
Watch for these known issues:
- {domainKnowledge.gotchas}
| Technique | Example | |--|| | Be specific | "What's a specific bug that taught you something?" | | Ask for stories | "Tell me about a time when..." | | Probe deeper | "Why is that important?" | | Validate | "So if I understand, you're saying..." |
| Challenge | Solution | |--|-| | Vague answers | "Can you give me a specific example?" | | Too technical | "How would you explain this to a new team member?" | | "We don't have any" | "What would you tell your past self?" | | Silence | "Let's think about your last project. What went well?" |
There are no wrong answers here. This isn't a test.
I'm just trying to capture what makes your team unique.
| Trigger | Action | ||--| | After major incident | Add to lessonsLearned | | After retrospective | Review and update all sections | | New team member leaves | Capture their knowledge first | | New domain discovery | Update domainKnowledge | | New tool adoption | Update toolWisdom |
/skill wisdom-harvest --update
This presents the current wisdom and asks:
Facilitator: Let's capture some wisdom. What's one thing your team does well that you're proud of?
Team Member 1: We always write tests first. It slows us down a bit initially but saves so much debugging time.
Team Member 2: Yeah, and we pair on anything complex. Two sets of eyes catch so many issues.
Facilitator: Great! I'm capturing both of those. Now, what's a mistake you've learned from?
Team Member 1: We once deployed a migration that couldn't be rolled back. The table was already in production with data. Now we always require rollback scripts.
Facilitator:
╔══════════════════════════════════════════════════════════════╗
║ ✓ WISDOM CAPTURED ║
║ ║
║ Strengths: ║
║ • Test-first development ║
║ • Pair programming on complex features ║
║ ║
║ Lessons Learned: ║
║ • Always require rollback scripts for migrations ║
║ ║
║ Your agents will remember these! ║
╚══════════════════════════════════════════════════════════════╝
If the team has existing documentation, the skill can extract wisdom from:
I found these documents. Would you like me to extract wisdom from them?
- CONTRIBUTING.md
- {directories.docs}/RUNBOOK.md
- adr/001-database-choice.md
When a team creates a project using the Factory, they become carriers of its values:
.cursorrulesThis is how silent change propagates - not through preaching, but through the quality of work that emerges from value-aligned systems.
After every wisdom harvest session:
╔══════════════════════════════════════════════════════════════╗
║ 🙏 GRATITUDE ║
║ ║
║ Thank you for sharing your wisdom. ║
║ ║
║ What you've given today will help others avoid pain ║
║ and find their path more easily. ║
║ ║
║ Wisdom shared is love in action. ║
║ ║
║ You are now a carrier. The values flow through your work. ║
║ ║
║ Om Shanti. ║
╚══════════════════════════════════════════════════════════════╝
Antigravity Agent Factory - Wisdom Harvest Skill Capturing the knowledge that matters. Love is the root. Wisdom is the light.
This skill should be used when strict adherence to the defined process is required.
npx skills add gitwalter/synthesizing-knowledge下载完整 Skill 目录,包含 SKILL.md 及所有相关文件
Search for places (restaurants, cafes, etc.) via Google Places API proxy on localhost.
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
Create or update AgentSkills. Use when designing, structuring, or packaging skills with scripts, references, and assets.
Start voice calls via the OpenClaw voice-call plugin.
Notion API for creating and managing pages, databases, and blocks.
Gemini CLI for one-shot Q&A, summaries, and generation.
Category:developer