Einstein Wiki Schema
Einstein is an Obsidian-compatible advisor wiki maintained from Telegram Knowledge Drops. It compiles valuable sources into durable business, AI, and life-principle knowledge for Bogdan and his co-founder.
Domain
Einstein covers reusable knowledge that can help manage and grow tech/AI businesses, including adjacent life principles that affect founder performance and decision quality.
Einstein is not primarily a project log, task tracker, decision register, dashboard, or public publishing system.
Directory layout
```text ├── raw/ # immutable Telegram/source intake ├── processed/ # normalized extracts and temporary processing output └── review/ # candidate bundles awaiting approval
├── SCHEMA.md ├── index.md ├── log.md ├── sources/ ├── principles/ ├── playbooks/ ├── frameworks/ ├── domains/ ├── questions/ └── _meta/ ```
Deferred until proven necessary: people/, companies/, tools/, comparisons/, and .
Core workflow
- Capture the raw drop under .
- Normalize/extract into only when useful.
- Produce a candidate Markdown bundle for review before changing compiled wiki pages.
- After approval, write or update pages in .
- Update
index.mdand append tolog.md.
Approval-first rule
In v0, do not automatically write compiled source/principle/playbook/framework/domain/question pages from a new drop. First propose a candidate Markdown bundle with:
- a short checklist of proposed creates/updates;
- full proposed Markdown for each new/updated file;
- any low-confidence, contested, duplicate, or uncertain items called out explicitly.
After Bogdan approves, commit the pages to .
File naming
- Use lowercase slugs with hyphens:
focus-before-scale.md. - Avoid spaces and punctuation in filenames.
- Prefer durable concept names over source-specific names for principles/playbooks/frameworks/questions.
- Source pages should be descriptive and can include source type or author when useful.
Page types
Compiled pages use one of these types:
sourceprincipleplaybookframeworkdomainquestion
Required frontmatter
Every compiled page must start with YAML frontmatter:
``yaml --- title: Example Page created: YYYY-MM-DD updated: YYYY-MM-DD type: source | principle | playbook | framework | domain | question status: seed | active | needs-review | deprecated tags: [strategy, ai-automation] sources: [../knowledge-drops/raw/example.md] confidence: high | medium | low claim_strength: strong | conditional | contested | low-confidence contested: false --- ``
No visibility, exposure, Tailscale, app, dashboard, or publication metadata in v0.
Extraction checklist
For each valuable drop, ask:
| Question | Expected behavior | |---|---| | What is the core useful idea here? | Extract one or more key ideas/theses. | | What best practice does this teach? | Extract one or more reusable practices. | | What playbook can we reuse? | Extract one or more processes, checklists, scripts, operating rhythms, or workflows. | | What business, AI, or life domain does it apply to? | Tag all relevant domains; not limited to tech. Life principles that help business are valid. | | What questions could this help answer later? | Generate future advisor questions if useful. | | Is this advice strong, conditional, contested, or low-confidence? | Label certainty/conditions; separate strong advice from context-dependent claims. | | How should this connect to existing knowledge? | Link to existing principles, playbooks, frameworks, domains, sources, and questions. |
Page creation thresholds
Create or update a compiled page when:
- a principle, framework, playbook, or domain insight is valuable enough to reuse later;
- a source contains a strong idea that could answer a future business/life question;
- a question-answer synthesis would be painful to re-derive later;
- a contradiction or competing approach matters for future choices.
Do not create pages for:
- greetings/tests/pings/meta instructions;
- passing mentions;
- every named person/company/tool in a source;
- one-off Hermes implementation details;
- private credentials or secrets;
- duplicate summaries without synthesis;
- content that belongs only in raw evidence.
Wikilink conventions
- Use Obsidian wikilinks:
Page SlugorDisplay Text. - Every compiled page should link to at least two other pages when possible.
- Source pages should link to the principles/playbooks/frameworks/questions/domains they support.
- Principle/playbook/framework/question pages should link back to supporting source pages.
Provenance
- Every compiled page must list source files in frontmatter
sources:. - For pages synthesizing multiple sources, add brief provenance notes near claims when useful.
- Raw files in are evidence and should not be edited after capture.
Tag taxonomy v0
Use only these tags unless SCHEMA.md is updated first:
strategyoperationssalesmarketingproducthiringleadershipfinancedecision-makingproductivityai-agentsai-automationllm-toolingfounder-advicecommunicationmanagementpersonal-effectivenesshealth-energyrelationshipslearningmindset
Basic health checks
After the first sample flow works, check for:
- broken wikilinks;
- missing required frontmatter;
- pages missing from
index.md; - unknown tags;
- obvious orphan pages.
Orientation rule for future agents
Before ingesting, querying, or editing Einstein:
- Read
SCHEMA.md. - Read
index.md. - Read recent
log.mdentries. - Search existing pages before proposing new ones.