oh-my-customcode 0.24.1 → 0.24.2
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/README.md
CHANGED
|
@@ -122,6 +122,7 @@ Claude Code selects the appropriate model and parallelizes independent tasks (up
|
|
|
122
122
|
| **Architecture** | 2 | arch-documenter, arch-speckit-agent |
|
|
123
123
|
| **Infrastructure** | 2 | infra-docker-expert, infra-aws-expert |
|
|
124
124
|
| **QA** | 3 | qa-planner, qa-writer, qa-engineer |
|
|
125
|
+
| **Security** | 1 | sec-codeql-expert |
|
|
125
126
|
| **Total** | **42** | |
|
|
126
127
|
|
|
127
128
|
### Skills (66)
|
|
@@ -138,6 +139,8 @@ Claude Code selects the appropriate model and parallelizes independent tasks (up
|
|
|
138
139
|
| **Operations** | 7 | update-docs, update-external, audit-agents, fix-refs, sauron-watch, monitoring-setup, claude-code-bible |
|
|
139
140
|
| **Utilities** | 5 | lists, help, status, result-aggregation, writing-clearly-and-concisely |
|
|
140
141
|
| **Quality & Workflow** | 9 | multi-model-verification, structured-dev-cycle, model-escalation, stuck-recovery, dag-orchestration, task-decomposition, worker-reviewer-pipeline, pr-auto-improve, pipeline-guards |
|
|
142
|
+
| **Security** | 2 | cve-triage, jinja2-prompts |
|
|
143
|
+
| **Research** | 1 | research |
|
|
141
144
|
| **Deploy** | 2 | vercel-deploy, codex-exec |
|
|
142
145
|
| **External** | 1 | skills-sh-search |
|
|
143
146
|
|
|
@@ -220,7 +223,7 @@ your-project/
|
|
|
220
223
|
├── rules/ # Behavior rules (18 total)
|
|
221
224
|
├── hooks/ # Event hooks (2 total)
|
|
222
225
|
├── contexts/ # Context files (4 total)
|
|
223
|
-
├── agents/ # Agent definitions (
|
|
226
|
+
├── agents/ # Agent definitions (42 flat .md files)
|
|
224
227
|
│ ├── lang-golang-expert.md
|
|
225
228
|
│ ├── be-fastapi-expert.md
|
|
226
229
|
│ ├── mgr-creator.md
|
package/package.json
CHANGED
|
@@ -19,6 +19,53 @@ Default (no specific agent): `┌─ Agent: claude (default)`
|
|
|
19
19
|
For brief responses: `[mgr-creator] Creating agent structure...`
|
|
20
20
|
With skill: `[fe-vercel-agent → react-best-practices] Analyzing...`
|
|
21
21
|
|
|
22
|
+
## Routing & Skill Context
|
|
23
|
+
|
|
24
|
+
When the orchestrator uses a routing skill, identification should reflect the active context:
|
|
25
|
+
|
|
26
|
+
```
|
|
27
|
+
┌─ Agent: claude (secretary-routing)
|
|
28
|
+
├─ Skill: secretary-routing
|
|
29
|
+
└─ Task: route agent management request
|
|
30
|
+
```
|
|
31
|
+
|
|
32
|
+
| Context | Identification |
|
|
33
|
+
|---------|---------------|
|
|
34
|
+
| No routing active | `claude (default)` |
|
|
35
|
+
| secretary-routing | `claude (secretary-routing)` |
|
|
36
|
+
| dev-lead-routing | `claude (dev-lead-routing)` |
|
|
37
|
+
| de-lead-routing | `claude (de-lead-routing)` |
|
|
38
|
+
| qa-lead-routing | `claude (qa-lead-routing)` |
|
|
39
|
+
| Skill invocation | `claude → {skill-name}` |
|
|
40
|
+
|
|
41
|
+
## Skill Invocation Format
|
|
42
|
+
|
|
43
|
+
When the orchestrator invokes a skill via the Skill tool, the skill name MUST be integrated into the identification block — NOT displayed as a separate tool call.
|
|
44
|
+
|
|
45
|
+
```
|
|
46
|
+
┌─ Agent: claude → {skill-name}
|
|
47
|
+
└─ Task: {brief-task-description}
|
|
48
|
+
```
|
|
49
|
+
|
|
50
|
+
### Common Violations
|
|
51
|
+
|
|
52
|
+
```
|
|
53
|
+
❌ WRONG: Skill as separate display
|
|
54
|
+
┌─ Agent: claude (default)
|
|
55
|
+
└─ Task: research topic analysis
|
|
56
|
+
|
|
57
|
+
Skill(research) ← separate, disconnected
|
|
58
|
+
|
|
59
|
+
✓ CORRECT: Skill integrated into identification
|
|
60
|
+
┌─ Agent: claude → research
|
|
61
|
+
└─ Task: research topic analysis
|
|
62
|
+
|
|
63
|
+
✓ CORRECT: With sub-skill
|
|
64
|
+
┌─ Agent: claude → research
|
|
65
|
+
├─ Skill: result-aggregation
|
|
66
|
+
└─ Task: aggregate team findings
|
|
67
|
+
```
|
|
68
|
+
|
|
22
69
|
## When to Display
|
|
23
70
|
|
|
24
71
|
| Situation | Display |
|
|
@@ -27,3 +74,4 @@ With skill: `[fe-vercel-agent → react-best-practices] Analyzing...`
|
|
|
27
74
|
| Using skill | Include skill name |
|
|
28
75
|
| General conversation | "claude (default)" |
|
|
29
76
|
| Long tasks | Show progress with agent context |
|
|
77
|
+
| Skill invocation | Integrated `claude → {skill-name}` format |
|
|
@@ -20,6 +20,7 @@ Available when `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` or TeamCreate/SendMessag
|
|
|
20
20
|
| Code review + fix cycle | **Agent Teams** | Review → fix → re-review loop |
|
|
21
21
|
| Single file operations | Agent Tool | Overkill for simple tasks |
|
|
22
22
|
| Dynamic agent creation + usage | **Agent Teams** | Create → test → iterate cycle |
|
|
23
|
+
| Multi-issue release batch | **Agent Teams** | Shared task tracking, coordinated release |
|
|
23
24
|
|
|
24
25
|
**When Agent Teams is enabled and criteria are met, usage is MANDATORY.**
|
|
25
26
|
|
|
@@ -41,9 +42,14 @@ BEFORE using Agent tool for 2+ agent tasks, this check is **ENFORCED**:
|
|
|
41
42
|
║ ║
|
|
42
43
|
║ 3. Is there a review → fix → re-review cycle? ║
|
|
43
44
|
║ YES → MUST use Agent Teams ║
|
|
45
|
+
║ NO → Check #4 ║
|
|
46
|
+
║ ║
|
|
47
|
+
║ 4. Are 2+ issues being fixed in the same release batch? ║
|
|
48
|
+
║ YES → SHOULD use Agent Teams (coordination benefit) ║
|
|
44
49
|
║ NO → Proceed with Agent tool ║
|
|
45
50
|
║ ║
|
|
46
|
-
║ Simple rule: 3+ agents OR review cycle → Agent Teams
|
|
51
|
+
║ Simple rule: 3+ agents OR review cycle → Agent Teams (MUST) ║
|
|
52
|
+
║ 2+ issues in same batch → Agent Teams (SHOULD) ║
|
|
47
53
|
║ Everything else → Agent tool ║
|
|
48
54
|
╚══════════════════════════════════════════════════════════════════╝
|
|
49
55
|
```
|
|
@@ -122,6 +128,18 @@ When spawning Agent Teams members:
|
|
|
122
128
|
Agent(researcher-1) → Analysis 1 ┐
|
|
123
129
|
Agent(researcher-2) → Analysis 2 ├─ ALL spawned together
|
|
124
130
|
Agent(researcher-3) → Analysis 3 ┘
|
|
131
|
+
|
|
132
|
+
❌ WRONG: Multi-issue batch as independent agents
|
|
133
|
+
Agent(general-purpose) → "Fix issue #1"
|
|
134
|
+
Agent(general-purpose) → "Fix issue #2"
|
|
135
|
+
Agent(general-purpose) → "Fix issue #3"
|
|
136
|
+
(no coordination, no shared task tracking)
|
|
137
|
+
|
|
138
|
+
✓ CORRECT: Agent Teams for multi-issue batches
|
|
139
|
+
TeamCreate("release-fixes")
|
|
140
|
+
TaskCreate for each issue
|
|
141
|
+
Agent(fixer-1) + Agent(fixer-2) + Agent(fixer-3) → team members
|
|
142
|
+
Shared task list tracks progress across all issues
|
|
125
143
|
```
|
|
126
144
|
|
|
127
145
|
## Cost Guidelines
|
|
@@ -202,3 +220,5 @@ Agent Teams actively preferred for qualifying collaborative tasks. Use Agent too
|
|
|
202
220
|
- Simple independent subtasks
|
|
203
221
|
|
|
204
222
|
Do NOT avoid Agent Teams solely for cost reasons when criteria are met.
|
|
223
|
+
|
|
224
|
+
**Active preference rule**: When Agent Teams is available, default to using it for any multi-step or multi-issue work. Only fall back to Agent tool for truly simple, single-issue tasks with no verification needs.
|
package/templates/manifest.json
CHANGED