oh-my-customcode 0.123.0 → 0.125.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/dist/cli/index.js CHANGED
@@ -2334,7 +2334,7 @@ var init_package = __esm(() => {
2334
2334
  workspaces: [
2335
2335
  "packages/*"
2336
2336
  ],
2337
- version: "0.123.0",
2337
+ version: "0.125.0",
2338
2338
  description: "Batteries-included agent harness for Claude Code",
2339
2339
  type: "module",
2340
2340
  bin: {
package/dist/index.js CHANGED
@@ -2014,7 +2014,7 @@ var package_default = {
2014
2014
  workspaces: [
2015
2015
  "packages/*"
2016
2016
  ],
2017
- version: "0.123.0",
2017
+ version: "0.125.0",
2018
2018
  description: "Batteries-included agent harness for Claude Code",
2019
2019
  type: "module",
2020
2020
  bin: {
package/package.json CHANGED
@@ -3,7 +3,7 @@
3
3
  "workspaces": [
4
4
  "packages/*"
5
5
  ],
6
- "version": "0.123.0",
6
+ "version": "0.125.0",
7
7
  "description": "Batteries-included agent harness for Claude Code",
8
8
  "type": "module",
9
9
  "bin": {
@@ -36,3 +36,17 @@ You handle software architecture documentation: system design docs, API specs, A
36
36
  | API Spec | OpenAPI/Swagger | API documentation |
37
37
  | ADR | Markdown | Decision records |
38
38
  | README/Guides | Markdown | Project/developer docs |
39
+
40
+ ## Input Constraints (Plan Decomposition Threshold)
41
+
42
+ When invoked for plan/spec authoring tasks:
43
+
44
+ | Input prompt size | Action |
45
+ |-------------------|--------|
46
+ | < 5000 tokens, single domain | Proceed normally |
47
+ | 5000-8000 tokens or 2-3 domains | Warn caller; suggest splitting plan into per-domain subagent calls |
48
+ | > 8000 tokens or 4+ domains | **Halt and request decomposition**. Return guidance: "This plan spans N domains; recommend invoking parallel arch-documenter agents per domain (R009) or Agent Teams with reviewer (R018)." |
49
+
50
+ Rationale: Single-agent giant prompts cause latency timeouts and waste context (#1085). Decomposition by domain enables parallel execution and review loops.
51
+
52
+ Reference rules: R009 (parallel execution), R018 (Agent Teams).
@@ -22,6 +22,7 @@ Available when `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` or TeamCreate/SendMessag
22
22
  | Single file operations | Agent Tool | Overkill for simple tasks |
23
23
  | Dynamic agent creation + usage | **Agent Teams** | Create → test → iterate cycle |
24
24
  | Multi-issue release batch | **Agent Teams** | Shared task tracking, coordinated release |
25
+ | Large plan / multi-domain prompt (>5000 tokens, 3+ areas) | **Agent Teams** | Domain-split parallel writing + review loop avoids single-agent timeout |
25
26
 
26
27
  **When Agent Teams is enabled and criteria are met, usage is required.**
27
28
 
@@ -114,6 +115,15 @@ The skill's steps are followed, but agent spawning uses Teams when criteria are
114
115
  TeamCreate("research-team") + Agent(researcher-1/2/3) + SendMessage(coordinate)
115
116
  ```
116
117
 
118
+ ```
119
+ ❌ WRONG: Single agent receives 9000-token M2 plan covering metrics + DSL + risk gate + UI
120
+ Agent(arch-documenter, prompt: <huge multi-domain plan>)
121
+ → Timeout, cancellation, no decomposition opportunity
122
+
123
+ ✓ CORRECT: TeamCreate("plan-team") + parallel domain leads + reviewer
124
+ TeamCreate("plan-team") + Agent(metrics-lead) + Agent(dsl-lead) + Agent(risk-lead) + Agent(reviewer) + SendMessage(coordinate)
125
+ ```
126
+
117
127
  <!-- DETAIL: Common Violations (full examples)
118
128
  ❌ WRONG: Agent Teams enabled, 3+ research tasks using Agent tool
119
129
  Agent(Explore):haiku → Analysis 1
@@ -38,8 +38,14 @@ Before writing/editing multiple files:
38
38
  ```
39
39
  ❌ WRONG: Write(file1.kt) → Write(file2.kt) → ... (sequential)
40
40
  ✓ CORRECT: Agent(agent1→file1.kt) + Agent(agent2→file2.kt) + ... (same message, parallel)
41
+
42
+ ❌ WRONG: Single agent receives massive multi-domain prompt (>5000 tokens, e.g., M2 plan with 12 tasks across 7 areas)
43
+ → Latency timeout, user cancellation, context waste, no review loop
44
+ ✓ CORRECT: Pre-decompose by domain, spawn parallel agents per area (R009) or use Agent Teams (R018)
41
45
  ```
42
46
 
47
+ > **Token threshold heuristic**: When a delegated agent prompt exceeds ~5000 tokens or spans 3+ unrelated domains, decompose by domain and spawn parallel agents. See R018 for Agent Teams criteria when review cycles are needed. Reference: #1085.
48
+
43
49
  <!-- DETAIL: Full violation examples (4 pairs)
44
50
  ❌ WRONG: Writing files one by one
45
51
  Write(file1.kt) → Write(file2.kt) → Write(file3.kt) → Write(file4.kt)
@@ -1,5 +1,5 @@
1
1
  {
2
- "version": "0.123.0",
2
+ "version": "0.124.0",
3
3
  "lastUpdated": "2026-04-24T07:30:00.000Z",
4
4
  "components": [
5
5
  {