thevoidforge-methodology 21.0.0 → 23.1.0

package/docs/methods/SUB_AGENTS.md

@@ -10,7 +10,7 @@ Parallelize development across multiple Claude Code sessions. Each session runs
 
 **All orchestration uses the build journal.** Every delegation, handoff, and resolution is logged to `/logs/handoffs.md`. Agents read journal files to recover context. See `/docs/methods/BUILD_JOURNAL.md`.
 
-**Full character roster: `/docs/NAMING_REGISTRY.md`** — 260+ named characters across 9 universes. No duplicates allowed.
+**Full character roster: `/docs/NAMING_REGISTRY.md`** — 263 named characters across 9 universes. No duplicates allowed. All agents materialized as subagent definitions in `.claude/agents/`.
 
 ---
 
@@ -174,6 +174,45 @@ Users can create project-specific sub-agents that carry domain knowledge. Define
 
 **When to create a custom agent:** When a project has a domain-specific pattern that the built-in agents miss repeatedly. If the same lesson appears 3+ times in LESSONS.md about the same technology, that's a signal for a custom agent rather than more method doc checklist items.
 
+## Subagent Definitions (ADR-044)
+
+All 263 agents are materialized as `.claude/agents/{name}.md` files. Commands now use `subagent_type: {agent-id}` instead of inline prompts. Each definition contains the agent's identity, behavioral directives, domain expertise, and output format.
+
+### The `.claude/agents/` Directory
+
+Each file is a standalone subagent definition that Claude Code's native subagent system can invoke. File naming convention: `{lowercase-name}.md` (e.g., `picard.md`, `batman.md`, `red-hood.md`).
+
+### Model Tiering
+
+| Tier | Model | Count | When |
+|------|-------|-------|------|
+| Lead | `inherit` (Opus) | ~20 | Domain leads who orchestrate, write fixes, make decisions |
+| Specialist | `sonnet` | ~200 | Domain experts who analyze, review, report findings |
+| Scout | `haiku` | ~43 | Lightweight reconnaissance, grep patterns, quick checks |
+
+Leads inherit the main session's model (Opus). Specialists run on Sonnet for cost efficiency without sacrificing analysis quality. Scouts run on Haiku for fast, cheap reconnaissance.
+
+### Tool Restrictions
+
+| Profile | Tools | Agents |
+|---------|-------|--------|
+| Builder | Full (Read, Edit, Write, Bash, Grep, Glob) | Leads who write code, fixes, ADRs, logs |
+| Reviewer | Read, Bash, Grep, Glob | Specialists who analyze and report but don't write |
+| Scout | Read, Grep, Glob | Lightweight agents that only read |
+| Adversarial | Read, Bash, Grep, Glob | Red-team agents who probe and test |
+
+### Description-Driven Dispatch
+
+Commands no longer use static dispatch tables (the old ADR-042 Cross-Domain Triggers). Instead, when Opus processes a command, it:
+
+1. Scans `git diff --stat` to identify changed files
+2. Matches changed file paths against the `description` fields of all 263 agents
+3. Launches matching specialists automatically alongside core agents
+
+This means a security audit that touches database migrations automatically picks up Spock (schema) without a hardcoded trigger table. The descriptions in `.claude/agents/*.md` ARE the dispatch rules.
+
+See `docs/AGENT_CLASSIFICATION.md` for the full classification manifest and `docs/adrs/ADR-044-subagent-materialization.md` for the architecture.
+
 ## Naming Rule
 
 When spinning up agents, check NAMING_REGISTRY.md AND `docs/CUSTOM_AGENTS.md` (if it exists). First claim wins. No duplicates across sessions. Log active names.

package/docs/methods/SYSTEMS_ARCHITECT.md

@@ -23,6 +23,14 @@
 
 **Need more?** Pull from Star Trek pool: Riker, Worf, Sisko, Janeway, Seven, O'Brien, Pike. See NAMING_REGISTRY.md.
 
+## Dynamic Dispatch (ADR-044)
+
+Agent dispatch is now description-driven. When Opus processes a command, it scans `git diff --stat` and matches changed files against the `description` fields of all 263 agents in `.claude/agents/`. Matching specialists launch automatically alongside core agents. No static dispatch tables needed.
+
+See `docs/AGENT_CLASSIFICATION.md` for the full classification and `docs/adrs/ADR-044-subagent-materialization.md` for the architecture.
+
+**Promoted agent:** **Riker** runs on every ADR written in any command — challenges trade-offs, prevents rubber-stamped decisions.
+
 ## Goal
 
 Ensure architecture matches product needs. Identify structural risks and scaling cliffs before they're expensive. Decide, don't defer — one clear path, not a menu.

package/docs/patterns/financial-transaction.ts

@@ -300,12 +300,35 @@ async function idempotentAppend<T>(
 // └── reports/
 //     └── 2026-03.json          # monthly summary
 
+// Global treasury paths — used by vault, totp, and backup (which stay user-scoped).
+// Per-project code should use getTreasuryDir(projectDir) and derived path functions.
 const TREASURY_DIR = join(homedir(), '.voidforge', 'treasury');
 const SPEND_LOG = join(TREASURY_DIR, 'spend-log.jsonl');
 const REVENUE_LOG = join(TREASURY_DIR, 'revenue-log.jsonl');
 const PENDING_OPS = join(TREASURY_DIR, 'pending-ops.jsonl');
 const BUDGETS_FILE = join(TREASURY_DIR, 'budgets.json');
 
+/**
+ * Get treasury directory for a project, or the global default.
+ * Per-project: {projectDir}/cultivation/treasury/
+ * Global (no arg): ~/.voidforge/treasury/
+ */
+function getTreasuryDir(projectDir?: string): string {
+  return projectDir ? join(projectDir, 'cultivation', 'treasury') : TREASURY_DIR;
+}
+function getSpendLog(projectDir?: string): string {
+  return join(getTreasuryDir(projectDir), 'spend-log.jsonl');
+}
+function getRevenueLog(projectDir?: string): string {
+  return join(getTreasuryDir(projectDir), 'revenue-log.jsonl');
+}
+function getPendingOps(projectDir?: string): string {
+  return join(getTreasuryDir(projectDir), 'pending-ops.jsonl');
+}
+function getBudgetsFile(projectDir?: string): string {
+  return join(getTreasuryDir(projectDir), 'budgets.json');
+}
+
 // ── Number Formatting (§9.15.4) ───────────────────────
 
 function formatCurrency(cents: Cents, detail: boolean = false): string {
@@ -339,4 +362,5 @@ export {
   atomicWrite, appendToLog, idempotentAppend,
   formatCurrency, formatRoas, formatPercentage,
   TREASURY_DIR, SPEND_LOG, REVENUE_LOG, PENDING_OPS, BUDGETS_FILE,
+  getTreasuryDir, getSpendLog, getRevenueLog, getPendingOps, getBudgetsFile,
 };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "thevoidforge-methodology",
-  "version": "21.0.0",
+  "version": "23.1.0",
   "description": "VoidForge methodology — agents, commands, methods, patterns.",
   "license": "MIT",
   "files": [
@@ -9,6 +9,7 @@
     "docs/methods/",
     "docs/patterns/",
     "docs/NAMING_REGISTRY.md",
+    "docs/AGENT_CLASSIFICATION.md",
     "HOLOCRON.md",
     "VERSION.md",
     "CHANGELOG.md",