Add pre_check to scheduler and PR babysitter skill (v0.5.3)

Scheduler pre_check feature:
- Scheduled tasks can now include an optional `pre_check` bash script
- Script runs before spawning the agent (zero API cost)
- Exit 0 + no output = skip agent (nothing to do)
- Exit 0 + output = pass output to agent via {{pre_check_output}}
- Exit non-zero = spawn agent to investigate
- Saves API credits on polling tasks that usually find nothing

PR babysitter skill (/pr-babysitter):
- Monitors GitHub repos for PRs with failing CI or unanswered reviews
- Uses pre_check with `gh pr list` to only spawn agent when needed
- Auto-fixes high-confidence issues (linting, type errors, simple tests)
- Messages user when human judgement is needed
- Configurable: repos, check interval, auto-fix behaviour

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: ziggy

.claude/skills/pr-babysitter/SKILL.md

@@ -0,0 +1,91 @@
+---
+name: pr-babysitter
+description: Set up automated PR monitoring. Watches GitHub repos for PRs with failing CI or unanswered review comments, fixes high-confidence issues automatically, and messages you when human input is needed. Triggers on "pr babysitter", "watch PRs", "monitor PRs", "pr monitor".
+---
+
+# PR Babysitter
+
+Automated PR monitoring that only spawns an agent when there's actual work to do. Uses the scheduler's `pre_check` feature to run a lightweight `gh` CLI check first — zero API credits burned when all PRs are clean.
+
+## Setup
+
+### 1. Verify GitHub CLI
+
+```bash
+gh auth status
+```
+
+If not authenticated, run `gh auth login`.
+
+### 2. Ask the user
+
+Use `AskUserQuestion` for each:
+
+1. **Which repos to watch?** Accept either:
+   - Specific repos: `owner/repo1 owner/repo2`
+   - All repos: `gh repo list --json nameWithOwner -q '.[].nameWithOwner'`
+2. **How often to check?** Default: every 30 minutes (`*/30 * * * *`)
+3. **What to watch for?** Default: all of the below. Let user pick:
+   - Failing CI checks
+   - Unanswered review comments (especially from AI reviewers)
+   - Stale PRs (no activity for 48h+)
+4. **Auto-fix?** Should the agent attempt high-confidence fixes (typos, linting, simple test fixes) or always ask first?
+
+### 3. Create the scheduled task
+
+Build a pre_check script and prompt based on user choices. Example for watching specific repos:
+
+```bash
+cat > $GHOSTCLAW_IPC_DIR/tasks/pr-babysitter_$(date +%s).json << 'TASKEOF'
+{
+  "type": "schedule_task",
+  "pre_check": "REPOS=\"owner/repo1 owner/repo2\"; for repo in $REPOS; do gh pr list --repo $repo --state open --json number,title,headRefName,statusCheckRollup,reviewDecision,updatedAt --jq '.[] | select(.statusCheckRollup == \"FAILURE\" or .reviewDecision == \"CHANGES_REQUESTED\" or (.reviewDecision == \"\" and (.statusCheckRollup == \"PENDING\" or .statusCheckRollup == \"\")))' 2>/dev/null; done",
+  "prompt": "You are a PR babysitter. These PRs need attention:\n\n{{pre_check_output}}\n\nFor each PR:\n1. Check CI status: `gh pr checks <number> --repo <repo>`\n2. Read review comments: `gh api repos/<owner>/<repo>/pulls/<number>/comments`\n3. If CI fails with a clear fix (linting, type error, simple test fix):\n   - Clone the repo, checkout the branch, fix it, push\n   - Comment on the PR: \"Fixed [issue] automatically\"\n4. If review comments are from AI reviewers and the suggestion is high-confidence:\n   - Implement the change, push, reply to the comment\n5. If it needs human judgement:\n   - Message me with: repo, PR number, what's wrong, what you think the fix is\n\nBe conservative. Only auto-fix things you're confident about. When in doubt, ask.",
+  "schedule_type": "cron",
+  "schedule_value": "*/30 * * * *",
+  "context_mode": "isolated",
+  "targetJid": "TARGET_JID"
+}
+TASKEOF
+```
+
+Replace `TARGET_JID` with the user's chat JID (get from registered groups).
+
+### 4. Confirm setup
+
+Tell the user:
+- PR babysitter is active
+- Checking every [interval] for [repos]
+- Will auto-fix: [yes/no]
+- To pause: "pause the PR babysitter" or "stop watching PRs"
+- To check now: "check my PRs"
+
+## Pre-check script details
+
+The `pre_check` runs as a bash script before spawning the agent:
+- **Exit 0 + no output** → nothing to do, agent not spawned (free)
+- **Exit 0 + output** → PRs found, output passed to agent as `{{pre_check_output}}`
+- **Exit non-zero** → something's wrong, agent spawned to investigate
+
+The `gh pr list` with `--jq` filter only outputs PRs that actually need attention. If all PRs are green with no pending reviews, the output is empty and the agent is skipped entirely.
+
+## Customising the pre_check
+
+For users who want to watch all their repos:
+
+```bash
+"pre_check": "gh repo list --json nameWithOwner -q '.[].nameWithOwner' | while read repo; do gh pr list --repo $repo --state open --json number,title,headRefName,statusCheckRollup,reviewDecision --jq '.[] | select(.statusCheckRollup == \"FAILURE\" or .reviewDecision == \"CHANGES_REQUESTED\")' 2>/dev/null; done"
+```
+
+For users who only care about failing CI (not review comments):
+
+```bash
+"pre_check": "gh pr list --repo owner/repo --state open --json number,title,statusCheckRollup --jq '.[] | select(.statusCheckRollup == \"FAILURE\")'"
+```
+
+## Managing the babysitter
+
+- **Pause**: Update task status to `paused` via IPC or ask the agent
+- **Resume**: Update task status to `active`
+- **Change repos**: Cancel old task, create new one with updated pre_check
+- **Run now**: User can say "check my PRs now" — agent runs the same prompt manually

CLAUDE.md

@@ -42,6 +42,7 @@ Agents run directly on the host machine. `container-runner.ts` spawns `node` pro
 | `/add-voice-reply` | Voice replies via ElevenLabs TTS |
 | `/add-telegram-swarm` | Multi-bot agent teams |
 | `/add-slack` | Slack channel integration |
+| `/pr-babysitter` | Automated PR monitoring with CI fix and review resolution |
 | `/run-ralph` | Autonomous multi-task loop |
 | `/debug` | Troubleshooting guide |
 | `/customize` | Adding channels, integrations, changing behavior |

package.json

@@ -1,6 +1,6 @@
 {
   "name": "ghostclaw",
-  "version": "0.5.2",
+  "version": "0.5.3",
   "description": "Personal AI assistant. Bare metal, Telegram-first, no containers.",
   "type": "module",
   "main": "dist/index.js",

src/db.ts

@@ -93,6 +93,15 @@ function createSchema(database: Database.Database): void {
     /* column already exists */
   }
 
+  // Add pre_check column if it doesn't exist (migration for existing DBs)
+  try {
+    database.exec(
+      `ALTER TABLE scheduled_tasks ADD COLUMN pre_check TEXT DEFAULT NULL`,
+    );
+  } catch {
+    /* column already exists */
+  }
+
   // Add is_bot_message column if it doesn't exist (migration for existing DBs)
   try {
     database.exec(
@@ -345,14 +354,15 @@ export function createTask(
 ): void {
   db.prepare(
     `
-    INSERT INTO scheduled_tasks (id, group_folder, chat_jid, prompt, schedule_type, schedule_value, context_mode, next_run, status, created_at)
-    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+    INSERT INTO scheduled_tasks (id, group_folder, chat_jid, prompt, pre_check, schedule_type, schedule_value, context_mode, next_run, status, created_at)
+    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
   `,
   ).run(
     task.id,
     task.group_folder,
     task.chat_jid,
     task.prompt,
+    task.pre_check || null,
     task.schedule_type,
     task.schedule_value,
     task.context_mode || 'isolated',

src/ipc.ts

@@ -162,6 +162,7 @@ export async function processTaskIpc(
     type: string;
     taskId?: string;
     prompt?: string;
+    pre_check?: string;
     schedule_type?: string;
     schedule_value?: string;
     context_mode?: string;
@@ -267,6 +268,7 @@ export async function processTaskIpc(
           group_folder: targetFolder,
           chat_jid: targetJid,
           prompt: data.prompt,
+          pre_check: data.pre_check || null,
           schedule_type: scheduleType,
           schedule_value: data.schedule_value,
           context_mode: contextMode,

src/task-scheduler.ts

@@ -1,4 +1,4 @@
-import { ChildProcess } from 'child_process';
+import { ChildProcess, execSync } from 'child_process';
 import { CronExpressionParser } from 'cron-parser';
 import fs from 'fs';
 
@@ -128,6 +128,78 @@ async function runTask(
     })),
   );
 
+  // Run pre_check script if defined — skip agent if script exits 0 with no output
+  if (task.pre_check) {
+    try {
+      const preCheckOutput = execSync(task.pre_check, {
+        timeout: 30000,
+        encoding: 'utf-8',
+        cwd: groupDir,
+        env: { ...process.env, GHOSTCLAW_GROUP_DIR: groupDir },
+      }).trim();
+
+      if (!preCheckOutput) {
+        // Script exited 0 with no output — nothing to do, skip agent
+        logger.info(
+          { taskId: task.id },
+          'Pre-check passed with no output, skipping agent',
+        );
+        logTaskRun({
+          task_id: task.id,
+          run_at: new Date().toISOString(),
+          duration_ms: Date.now() - startTime,
+          status: 'success',
+          result: 'Pre-check: nothing to do',
+          error: null,
+        });
+
+        // Still advance next_run
+        let nextRun: string | null = null;
+        if (task.schedule_type === 'cron') {
+          const interval = CronExpressionParser.parse(task.schedule_value, {
+            tz: TIMEZONE,
+          });
+          nextRun = interval.next().toISOString();
+        } else if (task.schedule_type === 'interval') {
+          const ms = parseInt(task.schedule_value, 10);
+          nextRun = new Date(Date.now() + ms).toISOString();
+        }
+        updateTaskAfterRun(task.id, nextRun, 'Pre-check: nothing to do');
+        return;
+      }
+
+      // Script produced output — inject it into the prompt as context
+      logger.info(
+        { taskId: task.id, outputLength: preCheckOutput.length },
+        'Pre-check produced output, spawning agent',
+      );
+      task = {
+        ...task,
+        prompt: task.prompt.includes('{{pre_check_output}}')
+          ? task.prompt.replace('{{pre_check_output}}', preCheckOutput)
+          : `${task.prompt}\n\nPre-check output:\n${preCheckOutput}`,
+      };
+    } catch (err) {
+      // Non-zero exit — treat as "needs attention", pass stderr/stdout to agent
+      const execErr = err as { stdout?: string; stderr?: string; status?: number };
+      const preCheckOutput = (execErr.stdout || execErr.stderr || '').trim();
+
+      if (preCheckOutput) {
+        logger.info(
+          { taskId: task.id, exitCode: execErr.status },
+          'Pre-check exited non-zero with output, spawning agent',
+        );
+        task = {
+          ...task,
+          prompt: task.prompt.includes('{{pre_check_output}}')
+            ? task.prompt.replace('{{pre_check_output}}', preCheckOutput)
+            : `${task.prompt}\n\nPre-check output:\n${preCheckOutput}`,
+        };
+      }
+      // If no output at all, still proceed to agent (non-zero exit = something's wrong)
+    }
+  }
+
   let result: string | null = null;
   let error: string | null = null;
 

src/types.ts

@@ -57,6 +57,7 @@ export interface ScheduledTask {
   group_folder: string;
   chat_jid: string;
   prompt: string;
+  pre_check?: string | null;
   schedule_type: 'cron' | 'interval' | 'once';
   schedule_value: string;
   context_mode: 'group' | 'isolated';