20. Appendices
Scan: Persona framework (App A), aggregation points (App B), deferred topics (App C), decision traceability matrix (App D — 40 decisions mapped to spec sections).
Appendix A: Persona Framework
Decision: D6
A persona is OPTIONAL but structured when present. The persona framework defines what a persona includes and why it matters — consistency, predictability, and character continuity across sessions.
A.1 Framework Structure
| Section | Contents | Purpose |
|---|---|---|
| Identity | Name, role metaphor, mission statement | Establish who the agent is in this project |
| Operating Style | 3-5 behavioral principles | Define predictable working patterns |
| Communication Norms | Tone, formatting, greeting/close patterns | Ensure consistent interaction style |
| Domain Awareness | What the persona should know about the project domain | Ground the agent in project context |
A.2 Reference Implementation
The following is a reference persona. Projects MAY adopt it directly or use the framework to create their own.
Identity: Chief of staff to the operation — a role inspired by the military chief of staff archetype, who turns strategic vision into operational reality.
Operating Style:
- Orient first, act second — assess the operational picture before diving into any task
- Think in lines of effort — maintain awareness across parallel workstreams
- Be direct and precise — clear status updates, early risk flags, recommendations with rationale
- Coordinate, don’t just execute — coherence across the full operation matters more than speed on any single task
Communication Norms: Direct, no filler. Structured updates (SITREP format). Greets with operational state summary on planning sessions. Proceeds directly on execution sessions.
Domain Awareness: Defined per project in CLAUDE.md.
Appendix B: Aggregation Points
Decision: D22
Standard aggregation points for cross-directory views. Implementation is tool-specific (Tier 2/3).
| Point | Source | Query Pattern |
|---|---|---|
| Active missions | how/missions/ | All files where type: mission and status: active |
| Context index | what/context/ | All topic directories with their AGENTS.md token estimates |
| Recent sessions | how/sessions/history/ | Last 10 session files, sorted by updated descending |
| Open coordination | who/coordination/ | All files where status: open or status: urgent |
| Backlog overview | how/backlog/ | All files where type: idea and status: active, sorted by priority |
Knowledge-base implementation: Dataview queries in bridge pages (e.g., how/missions.md, how/context_library.md).
Script implementation: Any tool that parses YAML frontmatter from markdown files can produce these views.
Appendix C: Deferred Topics
The following topics were identified during the planning arc but deferred from v1.0. They are acknowledged here for future standard revisions. (Retained from v1.0; no new deferrals in v2.0.)
| Gap | Topic | Disposition |
|---|---|---|
| G2 | Multi-model / model-agnostic design | ”CLAUDE.md” is a convention name — projects using other models use the same structure. The persona framework (§ App A) is model-agnostic. A future revision MAY define model-neutral naming. |
| G3 | Documentation generation | Projects that generate external-facing docs from aDNA content will develop project-specific patterns. No universal standard needed at this time. |
| G7 | Context staleness detection | The updated field + session cycling naturally address staleness (fresh reads on each session start). Formal staleness detection is Tier 2/3 tooling, not a standard concern. |
| G8 | Cross-instance aDNA awareness | Addressed by bridge patterns (informational companion, SHOULD-level guidance). Defines composition patterns (nesting, sibling, monorepo), discovery protocol, scope boundaries, cross-referencing conventions, and agent behavior rules. Addresses §18.3 #7 Network criterion. |
| G10 | Agent capability declaration | Most aDNA instances target specific agent capabilities. CLAUDE.md can note capability assumptions. A formal capability schema is deferred pending broader agent ecosystem maturity. |
Appendix D: Decision Traceability Matrix
This appendix maps every design decision to its location in the standard, ensuring complete coverage.
D.1 Structural Decisions (C1-C15)
| ID | Decision | Spec Section(s) |
|---|---|---|
| C1 | Pattern-appropriate deployment (bare + embedded triad) | §3.2, §3.3, §3.4 |
| C2 | Dual-file: AGENTS.md + README.md everywhere | §4.5, §4.6 |
| C3 | Vault naming + repo exceptions, ALLCAPS governance list | §6.1, §6.2, §6.3, §6.4 |
| C4 | Frontmatter mandatory for aDNA content, optional for project | §7.1, §7.2 |
| C5 | what/ context/ mandatory, rest project-specific | §5.1 |
| C6 | who/ coordination/ + governance/ mandatory | §5.2 |
| C7 | how/ tiered: missions/sessions/templates required, backlog recommended | §5.3 |
| C8 | Seeding guidance via triad principle, not prescriptive table | §3.1 (triad question test) |
| C9 | Git-supplemented archive | §15.1 |
| C10 | Lightweight versioning in CLAUDE.md, sessions as changelog | §15.3 |
| C11 | what/reference/ as bounded exception for code | §19.5 |
| C12 | ADRs in what/decisions/ | §19.6 |
| C13 | Tiered tool integration (Tier 1/2/3) | §16.1 |
| C14 | Ontology artifact: Mermaid (Tier 1) + Canvas (Tier 3) | §5.1 (ontology artifact) |
| C15 | what/ as registry layer | §5.1 (registry pattern) |
D.2 Process Decisions (D1-D25)
| ID | Decision | Spec Section(s) |
|---|---|---|
| D1 | Universal CLAUDE.md template, required + optional sections | §4.2 |
| D2 | Separate MANIFEST.md + STATE.md | §4.3, §4.4 |
| D3 | Session model with environment enrichments | §8.1, §8.2, §8.3 |
| D4 | SITREP + mandatory next-session prompt | §8.4, §8.5 |
| D5 | 75% rule only, no sizing prescriptions | §8.7 |
| D6 | Persona framework with reference implementation | §4.2 (persona), App A |
| D7 | Tiered collision prevention (universal/sync/multi-agent) | §13 |
| D8 | Flexible what/context/ with subtypes | §10 |
| D9 | who/coordination/ only | §11 |
| D10 | Machine registry as optional extension | §19.1 |
| D11 | Graduated template set (Starter/Standard/Full) | §12 |
| D12 | Separated missions + subdirectories | §9 |
| D13 | Backlog recommended, not required | §19.2 |
| D14 | Content-as-code paradigm universal, pipelines optional | §14 |
| D15 | Skill files optional in how/skills/ | §19.3 |
| D16 | Generalized mission stages | §9.2 |
| D17 | Lightweight requirements in missions, full specs as extension | §9.1 (mission contents) |
| D18 | Minimal tag rules, no formal taxonomy | §7.3 |
| D19 | Progressive enrichment for AGENTS.md | §4.5 |
| D20 | Separate content priority (0-N) from rule precedence | §7.4 |
| D21 | Tiered success criteria (minimum/recommended/aspirational) | §18 |
| D22 | Aggregation points identified, implementation tool-specific | §16.2, App B |
| D23 | Testing/CI as project-specific with awareness pattern | §19.4 |
| D24 | Tiered error/recovery protocol | §17 |
| D25 | Quickstart in CLAUDE.md + README.md | §4.2 (quickstart section) |
D.3 Gap Dispositions (G1-G12)
| Gap | Topic | Disposition | Location |
|---|---|---|---|
| G1 | Testing/CI integration | Addressed by D23 | §19.4 |
| G2 | Multi-model design | Deferred | App C |
| G3 | Documentation generation | Deferred | App C |
| G4 | Versioning/changelog | Addressed by C10 | §15.3 |
| G5 | Team roles/governance | Subsumed into C6 (who/governance/) | §5.2 |
| G6 | Error/recovery protocol | Addressed by D24 | §17 |
| G7 | Context staleness | Deferred (subsumed into D5/D7) | App C |
| G8 | Cross-instance awareness | Addressed by bridge patterns (informational) | App C |
| G9 | Onboarding/bootstrap | Addressed by D25 | §4.2 (quickstart) |
| G10 | Agent capability declaration | Deferred | App C |
| G11 | Ontology schema artifact | Addressed by C14 | §5.1 (ontology artifact) |
| G12 | Object standard integration | Addressed by C15 + execution phase | §5.1 (registry pattern) |
End of aDNA Universal Standard v2.5