Onboarding.md
1 --- 2 concept_id: CON-onboarding 3 created: 2026-01-26 4 updated: 2026-01-26 5 type: parent-concept 6 extraction_trigger: "N>=3 artifacts sharing this pattern" 7 instance_count: 5 8 tags: 9 - concept 10 - protocol 11 - agents 12 parent: null 13 session: "[[2026-01-26]]" 14 --- 15 16 # Onboarding 17 18 > The progressive revelation of system capabilities to new participants. 19 20 --- 21 22 ## The Pattern 23 24 Onboarding in Sovereign OS follows the principle of **progressive trust revelation**: 25 26 1. **Zero-assumption start** - New agents/users start with minimal context 27 2. **Capability discovery** - Learn what the system can do through interaction 28 3. **Trust accumulation** - Privileges expand as trust is established 29 4. **Context inheritance** - Prior knowledge flows to new sessions 30 31 This pattern applies to: 32 - Claude instances starting a new session 33 - New team members joining the project 34 - External services integrating with the system 35 36 --- 37 38 ## Invariants (What MUST Be True) 39 40 ^invariants 41 42 1. **Bootstrap is mandatory** - Every new context must run `agent_bootstrap.py` 43 2. **CLAUDE.md is read first** - Core protocols before any work 44 3. **Mesh awareness** - New instances check cross-session state 45 4. **Trust starts low** - New actors earn capabilities, don't assume them 46 47 --- 48 49 ## Extension Points (What CAN Vary) 50 51 ^extensions 52 53 | Artifact Type | Extension | Example | 54 |---------------|-----------|---------| 55 | Protocol | Bootstrap sequence | What checks run at startup | 56 | Design | Context loading | Which files to read first | 57 | PRD | User journey | How human users are introduced | 58 59 --- 60 61 ## Instances 62 63 ```dataview 64 TABLE 65 type as "Type", 66 status as "Status", 67 file.mtime as "Modified" 68 FROM "" 69 WHERE contains(parent, "Onboarding") OR contains(tags, "onboarding") 70 SORT file.mtime DESC 71 ``` 72 73 --- 74 75 ## Anti-Patterns 76 77 ^antipatterns 78 79 1. **Assuming context** - Acting on stale knowledge from previous sessions 80 2. **Skipping bootstrap** - Starting work without checking system state 81 3. **Permission explosion** - Granting full access before trust is established 82 4. **Information overload** - Dumping entire system context at once 83 84 --- 85 86 ## The Onboarding Funnel 87 88 ``` 89 ┌─────────────────────────────────────────┐ 90 │ 1. BOOTSTRAP │ 91 │ - Read CLAUDE.md │ 92 │ - Run agent_bootstrap.py │ 93 │ - Display startup message │ 94 └─────────────────┬───────────────────────┘ 95 ▼ 96 ┌─────────────────────────────────────────┐ 97 │ 2. MESH AWARENESS │ 98 │ - Check git log (what changed?) │ 99 │ - Check mesh (who else is working?) │ 100 │ - Surface pending votes │ 101 └─────────────────┬───────────────────────┘ 102 ▼ 103 ┌─────────────────────────────────────────┐ 104 │ 3. CONTEXT INHERITANCE │ 105 │ - Read LIVE-COMPRESSION.md │ 106 │ - Review INSIGHT-BACKLOG.md │ 107 │ - Check FO-INPUT.md │ 108 └─────────────────┬───────────────────────┘ 109 ▼ 110 ┌─────────────────────────────────────────┐ 111 │ 4. READY FOR WORK │ 112 │ - System state understood │ 113 │ - Pending tasks visible │ 114 │ - Trust baseline established │ 115 └─────────────────────────────────────────┘ 116 ``` 117 118 --- 119 120 ## Parent Concepts 121 122 This concept feeds into: 123 - [[Agents]] - Agent onboarding is a key subcase 124 - [[Trust]] - Onboarding establishes initial trust levels 125 126 --- 127 128 ## Extraction History 129 130 | Date | Trigger | Instance Count | 131 |------|---------|----------------| 132 | 2026-01-26 | N>=3 (5 artifacts) | 5 | 133 134 --- 135 136 *Concept extracted via Emergent Parent Extraction protocol*