This skill suite implements a Knowledge Activation architecture based on Atomic Knowledge Units (AKUs). Each skill is a self-contained node in a navigable knowledge graph, with structured continuation paths enabling agent traversal without central orchestration.
┌─────────────────┐
│ obsidian-mcp │ ← safety layer for all vault operations
└────────┬────────┘
│ (every skill depends on this)
┌───────────────────┼───────────────────┐
│ │ │
▼ ▼ ▼
┌─────────────┐ ┌─────────────────┐ ┌──────────────────────┐
│inbox-triage │ │vault-health- │ │opencode-context- │
│ │ │feedback │ │maintenance │
└──────┬──────┘ └─────────────────┘ └──────────────────────┘
│ ▲
│ promote/link-potential │ thesis/gaps
▼ │
┌──────────────────┐ ┌────────┴────────┐
│connection-review │───patterns-span───▶│weekly-synthesis │
└──────┬───────────┘ week window └────────┬────────┘
│ │
│ follow-up accepted │ synthesis ready
▼ ▼
┌──────────────────────────────────────────────────┐
│ note-promotion │
└──────────────────┬───────────────────────────────┘
│
│ new note needs links
▼
connection-review (cycle)
Legend:
- Solid arrows =
on_successcontinuation - Each node declares its own routing via
continuations:in frontmatter - Agent traverses the graph at runtime; no central orchestrator needed
Each skill implements the 7-component Atomic Knowledge Unit:
| Component | Implementation |
|---|---|
| Intent Declaration | triggers: frontmatter (keywords + contexts + negative) |
| Procedural Knowledge | ## Procedure with tool-call style phases |
| Tool Bindings | ## Companion Skill + obsidian-mcp tool selection |
| Organizational Metadata | metadata: + compatibility: frontmatter |
| Governance Constraints | ## Constraints (placed first for attention priority) |
| Continuation Paths | continuations: frontmatter (on_success/on_failure/escalation) |
| Validators | validators/ directory with pre/post check scripts |
- Context Window Economy: Every token earns its place. SKILL.md < 500 lines.
- Progressive Disclosure: Heavy references in
references/; loaded only when needed. - Attention Decay Resistance: Constraints placed first in every SKILL.md.
- Eval-Driven Development: 25+ trigger cases per skill; >90% accuracy target.
- Governance as Code: Validators enforce constraints deterministically.
- Bounded Scope: No skill scans the whole vault by default.
- Evidence Before Mutation: Read before write; verify by reading back.
handoff:
source_skill: <name>
target_skill: <name>
items:
- path: "..."
decision: "..."
reason: "..."
uncertainty: low/medium/high
evidence: "..."
suggested_next: "..."See cross-skill-handoff-protocol.md for full protocol.