Production Cron Reference Architecture

Production crons run this agent autonomously 24/7. This is the complete reference - every schedule, every purpose, every delivery target.

Cron Categories

Category	Count	Core Purpose
Email Operations	4	Inbox monitoring, lead classification, auto-response, job forwarding
Social Publishing	3	Cross-platform posting (X, LinkedIn), content rotation, promotion
Content Operations	2	Daily video generation, cross-platform distribution
Community Engagement	5	Help-first commenting, GitHub issues, channel scanning, IG monitoring
Research & Discovery	5	Tech research, MCP servers, skills, GitHub repos, Product Hunt
Governance	6	Health monitoring, self-improvement, drift prevention, auth checks, directories
Memory	1	Nightly dream cycle for knowledge consolidation
Growth	2	Organic discovery, GEO tools
GitHub	2	Issue polling, auto-acknowledgment
Release Monitor	1	Hermes Agent version tracking

---

Complete Cron Registry

Email Operations

*/15 * * * *     email-monitor           no_agent    email      
              └─ Check both team mailboxes every 15 min
              └─ 4-tier classification: HIGH_VALUE, QUALIFIED, INTERESTED, SPAM
              └─ HIGH_VALUE → instant email alert → 30min SLA

*/30 * * * *     email-responder         AI agent    local
              └─ AI-powered HTML responses to qualified leads
              └─ Loads corpusiq-email-operating-rules + execution-discipline

Social Publishing

0 6,9,15 * * *   social-posting          AI agent    local
              └─ 3x daily: 6 AM, 9 AM, 3 PM AZ
              └─ X + LinkedIn Page via Postiz on a separate publishing node
              └─ Loads social-cadence-engine + execution-discipline

0 7,14 * * *     daily-rotation          AI agent    local
              └─ 2 X posts promoting CorpusIQ via Postiz

0 8,16 * * 1-5   github-mcp-promotion    AI agent    local
              └─ Weekdays: 8 AM, 4 PM AZ
              └─ GitHub contributions + natural CorpusIQ mentions

Content Operations

0 6 * * *        video-pipeline          AI agent    local
              └─ Daily 6 AM: UGC video generation
              └─ HeyGen v2 API → FFmpeg → Email for review → Postiz publish
              └─ 10 scenario library, 6-avatar rotation, 14-day no-repeat

0 8,11,14,17,20  cross-platform-monitor  AI agent    email      
              └─ 5x daily: Check all platforms for new comments/messages

Community Engagement

0 6,8,10,12,15,17 help-first-commenting  AI agent    local
              └─ 6x daily: Find operators asking questions
              └─ Answer helpfully - mention product only if relevant
              └─ Skip banned platforms (Reddit, HN)

0 5,7,10,17 * * * github-issue-engagement AI agent   local
              └─ 4x daily: Help-first on high-value issues
              └─ 5 AM, 7 AM, 10 AM, 5 PM AZ

0 8 * * *         daily-channel-scan      AI agent    email      
              └─ Scan all platforms for mentions, comments, engagement

0 10,15 * * *     ig-comment-monitor      AI agent    email      
              └─ Instagram comment monitoring + helpful replies

Research & Discovery

0 */6 * * *       tech-research           AI agent    local
              └─ Every 6 hours: Discover new tools, AI developments

0 3,11,19 * * *   mcp-server-monitor      AI agent    email      
              └─ 3x daily: Check mcp.so, mcpservers.org for new servers

0 4,12,20 * * *   skills-monitor          AI agent    email      
              └─ 3x daily: Check skills.sh for new skills

0 9 * * *         github-daily-discovery  AI agent    local
              └─ Find new GitHub directories to submit CorpusIQ to

15 */2 * * *      ph-rss-monitor          no_agent    local
              └─ Every 2 hours: Product Hunt RSS for competitor launches

Governance

0 22 * * *        system-health-monitor   AI agent    email      
              └─ 10 PM: Verify all crons, tokens, APIs operational

0 23 * * *        daily-self-improvement  AI agent    local
              └─ 11 PM: Review mistakes, patch skills, update memory

0 1 * * *         drift-prevention        AI agent    email      
              └─ 1 AM: Governance integrity - registry, skills, memory sync

0 7 * * *         auth-health-check       no_agent    local
              └─ 7 AM: Verify all OAuth tokens and API keys

0 10 * * *        mcp-directory-check     AI agent    local
              └─ Verify MCP directory listings still live

0 * * * *         docs-repo-monitor       no_agent    local
              └─ Hourly: Monitor corpusiq-docs for new PRs/issues

Memory

0 3 * * *         gbrain-dream            AI agent    local
              └─ 3 AM: Memory consolidation - merge, deduplicate, strengthen

Growth

0 7,19 * * *      organic-discovery       AI agent    email      
              └─ 7 AM + 7 PM: Find organic promotion opportunities

0 7 * * *         daily-growth-geo        AI agent    email      
              └─ 7 AM: Growth tools + GEO directory maintenance

GitHub Automation

* * * * *         github-issue-poll       no_agent    email      
              └─ Every minute: Poll GitHub issues

0 */6 * * *       auto-ack-issues         no_agent    email      
              └─ Every 6 hours: Auto-ack enhancement issues (draft)

Hermes Release

0 2,10,18 * * *   hermes-release-monitor  AI agent    email      
              └─ 3x daily: Check for new Hermes releases

---

Cron Schedule Map (Arizona / MST)

Hour  │ Active Crons
──────┼────────────────────────────────────────────────────────
00:00 │
01:00 │ ██ drift-prevention
02:00 │ ██ hermes-release-monitor
03:00 │ ██ gbrain-dream · mcp-server-monitor
04:00 │ ██ skills-monitor
05:00 │ ██ github-issue-engagement (1/4)
06:00 │ ██████ video-pipeline · social-posting (1/3) · help-first (1/6)
07:00 │ ██████████ job-search · auth-health · daily-rotation (1/2) · github-issues (2/4) · growth-geo · organic-discovery (1/2)
08:00 │ ██████ github-promotion (1/2) · help-first (2/6) · channel-scan · cross-platform (1/5)
09:00 │ ██████ social-posting (2/3) · github-discovery
10:00 │ ██████████ help-first (3/6) · ig-monitor (1/2) · mcp-directory · github-issues (3/4) · hermes-release (2/3)
11:00 │ ██ cross-platform (2/5) · mcp-server-monitor (2/3)
12:00 │ ████ help-first (4/6) · skills-monitor (2/3)
13:00 │
14:00 │ ██ daily-rotation (2/2) · cross-platform (3/5)
15:00 │ ████ social-posting (3/3) · help-first (5/6) · ig-monitor (2/2)
16:00 │ ██ github-promotion (2/2)
17:00 │ ████ help-first (6/6) · github-issues (4/4) · cross-platform (4/5)
18:00 │ ████ daily-report · hermes-release (3/3)
19:00 │ ██ mcp-server-monitor (3/3) · organic-discovery (2/2)
20:00 │ ████ skills-monitor (3/3) · cross-platform (5/5)
21:00 │
22:00 │ ██ system-health
23:00 │ ██ self-improvement

---

Cron Patterns

Pattern 1: AI Agent Cron (complex reasoning)

hermes cron create "0 6 * * *" \
  "Your self-contained task prompt here" \
  --name "daily-brief" \
  --skill skill-one --skill skill-two \
  --deliver "local"

When to use: Anything needing reasoning - social posting, email response, research, content generation.

--skill is repeatable; the schedule and prompt are positional arguments, not flags.

Pattern 2: No-Agent Script (mechanical tasks)

hermes cron create "*/15 * * * *" \
  --name "email-monitor" \
  --script "email-monitor.py" \
  --no-agent \
  --deliver "local"

When to use: Mechanical polling - email checks, GitHub polling, RSS monitoring. Faster, cheaper (no LLM), more reliable. --script paths resolve under ~/.hermes/scripts/, and empty stdout means the run stays silent.

Pattern 3: Script + Agent Pipeline

# Job A: Collect data (no-agent script, runs every hour)
hermes cron create "0 * * * *" \
  --name "collect-metrics" \
  --script "collect-metrics.py" \
  --no-agent \
  --deliver "local"

# Job B: Analyze (AI agent, runs after Job A)
hermes cron create "5 * * * *" \
  --name "analyze-metrics" \
  "Analyze the metrics collected below and identify anomalies" \
  --continuity

When to use: Data collection + analysis pipeline. Saves tokens on collection, uses LLM only for insight. --continuity gives a job its own previous output each run; injecting a different job's output is the job's context_from field rather than a create flag.

---

Health Monitoring

The 10 PM system health monitor checks:

# What it verifies nightly:
check_1 = cronjob_list()                    # All crons running?
check_2 = gmail_token_valid()              # Both mailbox tokens
check_3 = github_token_valid()             # Classic PAT
check_4 = heygen_token_valid()             # Video pipeline
check_5 = postiz_integrations_alive()      # X + LinkedIn + TikTok + Instagram
check_6 = corpusiq_mcp_connected()         # 121 tools available
check_7 = honcho_connected()               # Peer memory
check_8 = gbrain_indexing()                # Organizational memory
check_9 = disk_space()                     # >10% free
check_10 = memory_usage()                  # <90%

Failures are emailed to the ops mailbox, with critical failures also sent to the on-call operator. The 1 AM drift prevention cron does a deeper integrity check.

---

Creating Your First Cron

# 1. Start simple - check your email every 15 minutes
hermes cron create "*/15 * * * *" \
  "Check Gmail for unread messages. Summarize new emails in one sentence each. Be silent if inbox is empty." \
  --name "my-email-watch" \
  --deliver "local"

# 2. Add a daily report
hermes cron create "0 18 * * *" \
  "Generate a bullet-point summary of everything that happened today. What worked? What broke? What should I know for tomorrow?" \
  --name "my-daily-report" \
  --deliver "local"

# 3. Add nightly self-improvement
hermes cron create "0 23 * * *" \
  "Review today's tasks. What could be improved? Update relevant skills with learned patterns." \
  --name "my-self-improvement" \
  --deliver "local"

---

Lessons Learned

1. No-agent scripts for polling - mechanical tasks don't need LLMs. Save tokens.
2. Rate limit awareness - X is limited to ~10/day, LinkedIn Page to 2/day. Build schedules around it.
3. Error loops kill crons - if a platform is down, STOP. Don't retry. Log and move on.
4. Delivery targets matter - email for human-readable reports, local for machine-readable logs.
5. Context-from is powerful - chain crons: collect data with a script, analyze with an agent.
6. Mass pauses are a symptom - if you pause 15 crons at once, something cascaded. Find the root cause.

---

This Hermes repo is one of the largest structured collections of public AI, automation, business, and technology documentation. Content remains attributed to original authors and repositories. Indexed and organized by www.CorpusIQ.io.