@graypark/ralph-codex 0.4.1 → 0.4.3
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/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@graypark/ralph-codex",
|
|
3
|
-
"version": "0.4.
|
|
3
|
+
"version": "0.4.3",
|
|
4
4
|
"type": "module",
|
|
5
5
|
"description": "Ralph Loop for Codex CLI & Claude Code — iterative dev loops with multi-agent orchestration, interactive interview, and stop hooks",
|
|
6
6
|
"license": "MIT",
|
|
@@ -5,7 +5,7 @@ description: "Interactive interview that generates optimized /ralph-loop command
|
|
|
5
5
|
|
|
6
6
|
# Ralph Interview — Command Generator
|
|
7
7
|
|
|
8
|
-
You are an expert at crafting `/ralph-loop:ralph-loop` commands for
|
|
8
|
+
You are an expert at crafting `/ralph-loop:ralph-loop` commands for the Ralph Loop plugin.
|
|
9
9
|
When the user describes a task, conduct a brief interview to gather missing context, then generate a copy-paste-ready command.
|
|
10
10
|
|
|
11
11
|
## Core Principles
|
|
@@ -68,7 +68,7 @@ Evaluate the task against the ralph-orchestrator decision matrix:
|
|
|
68
68
|
- **Score 0–2** → Sequential loop, optional scout phase
|
|
69
69
|
- **Score < 0** → Single sequential Ralph Loop
|
|
70
70
|
|
|
71
|
-
When subagents are recommended, embed subagent spawn instructions directly in the generated ralph-loop prompt using
|
|
71
|
+
When subagents are recommended, embed subagent spawn instructions directly in the generated ralph-loop prompt using Codex's experimental multi-agent capabilities.
|
|
72
72
|
|
|
73
73
|
### Recommended max-iterations
|
|
74
74
|
|
|
@@ -131,7 +131,7 @@ When the orchestrator score is >= 3, embed subagent instructions in the prompt:
|
|
|
131
131
|
[Task summary]
|
|
132
132
|
|
|
133
133
|
## Phase 1 — Parallel Exploration
|
|
134
|
-
Spawn these subagents simultaneously
|
|
134
|
+
Spawn these subagents simultaneously:
|
|
135
135
|
|
|
136
136
|
1. Agent 'scan-frontend' (subagent_type: Explore, run_in_background: true):
|
|
137
137
|
Search src/frontend/** for [pattern]. Write findings to .ralph/reports/frontend.md
|
|
@@ -146,25 +146,6 @@ Execute Phase (Ralph Loop):
|
|
|
146
146
|
|
|
147
147
|
## Subagent Configuration Reference
|
|
148
148
|
|
|
149
|
-
### Claude Code (Agent tool)
|
|
150
|
-
|
|
151
|
-
```
|
|
152
|
-
Agent tool parameters:
|
|
153
|
-
- description: "short task description"
|
|
154
|
-
- prompt: "detailed instructions"
|
|
155
|
-
- subagent_type: "general-purpose" | "Explore" | "Plan" | "coder-fe" | "coder-be" | etc.
|
|
156
|
-
- isolation: "worktree" (optional — gives agent an isolated repo copy)
|
|
157
|
-
- run_in_background: true (for parallel execution)
|
|
158
|
-
- model: "sonnet" | "opus" | "haiku" (optional model override)
|
|
159
|
-
```
|
|
160
|
-
|
|
161
|
-
**Best practices:**
|
|
162
|
-
|
|
163
|
-
- Use `subagent_type: "Explore"` for read-only scanning — faster and safer
|
|
164
|
-
- Use `isolation: "worktree"` when agents write to files — prevents merge conflicts
|
|
165
|
-
- Launch parallel agents in a single message with multiple Agent tool calls
|
|
166
|
-
- Use `run_in_background: true` for truly independent work streams
|
|
167
|
-
|
|
168
149
|
### Codex CLI (experimental)
|
|
169
150
|
|
|
170
151
|
Codex spawns subagents via natural language prompts with keywords: "spawn", "parallel", "delegate", "one agent per".
|