fix: reduce Ralph default from 50 to 10 iterations, make configurable

50 was excessive for typical overnight task lists. Now defaults to 10.
Configurable via RALPH_MAX_ITERATIONS in .env or per-run.

The /run-ralph skill now counts tasks and sets iterations to
task count + 2 (buffer for retries), or asks the user.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: ziggy

.claude/skills/run-ralph/SKILL.md

@@ -39,22 +39,29 @@ Rules for good task files:
 sqlite3 store/messages.db "SELECT jid FROM registered_groups WHERE folder = 'main' LIMIT 1;"
 ```
 
-### 3. Start the run
+### 3. Set iterations
+
+Count the unchecked tasks in the task file. Set `maxIterations` to that count + 2 (buffer for retries). If unsure, ask the user: "You have N tasks — run all of them, or set a limit?"
+
+Default is 10 if not specified. Can also be set via `RALPH_MAX_ITERATIONS` in `.env`.
+
+### 4. Start the run
 
 Write an IPC file:
 
 ```bash
 CHAT_JID="<main chat JID>"
 TASK_FILE="/absolute/path/to/tasks.md"
 WORK_DIR="/absolute/path/to/project"
+MAX_ITER=<task count + 2>
 
 cat > data/ipc/main/tasks/ralph_start_$(date +%s).json << EOF
 {
   "type": "start_ralph",
   "taskFile": "$TASK_FILE",
   "workDir": "$WORK_DIR",
   "targetJid": "$CHAT_JID",
-  "maxIterations": 50,
+  "maxIterations": $MAX_ITER,
   "notifyProgress": true
 }
 EOF
@@ -65,7 +72,7 @@ The bot will message you: "Ralph run started: ralph-{id}"
 ### 4. Monitor progress
 
 Between each iteration, the bot messages you:
-> Ralph [3/50]: Completed "Add login endpoint with JWT"
+> Ralph [3/7]: Completed "Add login endpoint with JWT"
 
 Check detailed progress:
 ```bash
@@ -111,7 +118,7 @@ The task file will have all checkboxes marked `[x]`.
 ## Safety
 
 - Each iteration runs in **isolated context** (fresh agent, no carried-over confusion)
-- Max 50 iterations by default (configurable via `maxIterations`)
+- Max 10 iterations by default (configurable via `maxIterations` or `RALPH_MAX_ITERATIONS` in `.env`)
 - Tasks that fail 3 times are skipped (configurable via `maxFailuresPerTask` in config.json)
 - Agent is instructed to **commit after each task** — easy to roll back
 - You can stop the loop at any time via the stop IPC command

src/config.ts

@@ -12,6 +12,7 @@ const envConfig = readEnvFile([
   'TELEGRAM_ONLY',
   'AGENT_IDLE_TIMEOUT',
   'AGENT_ABSOLUTE_TIMEOUT',
+  'RALPH_MAX_ITERATIONS',
 ]);
 
 export const ASSISTANT_NAME =
@@ -55,6 +56,11 @@ export const AGENT_ABSOLUTE_TIMEOUT = parseInt(
     '2700000',
   10,
 ); // 45 min default
+export const RALPH_MAX_ITERATIONS = parseInt(
+  process.env.RALPH_MAX_ITERATIONS || envConfig.RALPH_MAX_ITERATIONS || '10',
+  10,
+); // 10 steps default — override in .env or per-run
+
 export const MAX_CONCURRENT_CONTAINERS = Math.max(
   1,
   parseInt(process.env.MAX_CONCURRENT_CONTAINERS || '5', 10) || 5,

src/ralph-runner.test.ts

@@ -127,7 +127,7 @@ describe('startRalphRun', () => {
       k.endsWith('config.json'),
     )!;
     const config = JSON.parse(deps.files[configKey]);
-    expect(config.maxIterations).toBe(50);
+    expect(config.maxIterations).toBe(10);
     expect(config.maxFailuresPerTask).toBe(3);
   });
 });

src/ralph-runner.ts

@@ -6,7 +6,7 @@
 
 import path from 'path';
 
-import { DATA_DIR } from './config.js';
+import { DATA_DIR, RALPH_MAX_ITERATIONS } from './config.js';
 import { logger } from './logger.js';
 import {
   RalphConfig,
@@ -165,7 +165,7 @@ export async function startRalphRun(
     workDir: params.workDir,
     targetJid: params.targetJid,
     groupFolder: params.groupFolder,
-    maxIterations: params.maxIterations ?? 50,
+    maxIterations: params.maxIterations ?? RALPH_MAX_ITERATIONS,
     maxFailuresPerTask: params.maxFailuresPerTask ?? 3,
     currentIteration: 0,
     consecutiveFailures: {},