Summary
A scheduled cleanup script that runs daily at 4 AM Central Time to maintain approval gate data in the Daily AI Agents OS runtime. This is one of the housekeeping tasks in a broader system that turns a Mac into an autonomous AI company you operate via Telegram. The Daily AI Agents bundle includes skills like cross-agent routing, inbox aggregation, and output quality gates. This janitor specifically tidies the approval gates directory, which likely accumulates state from skills like voice-gate and public-content-scrubber that score and block content before it ships. Reach for this if you're running the Daily AI runtime and need automated cleanup of stale approval decisions rather than manual filesystem maintenance.
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
Daily AI Agents OS
Autonomous AI company runtime for solo founders. Built in public, run privately, documented openly.
This repo ships the 10-skill public bundle for the Starter tier of Daily AI Agents OS — a runtime that turns one Mac into an autonomous AI company you operate by Telegram.
The brain is Hermes. The arms are OpenClaw specialists. This bundle is the first layer of skills they execute, written in bash + jq + curl, all readable in 60 seconds per skill.
Install
curl -sf https://usedailyai.com/install.sh | bash
The installer runs 9 checks: Darwin arm64, ≥ 16 GB RAM, Hermes present, OpenClaw present, license valid, bundle synced, symlinked into ~/.openclaw/skills/dai-os-bundle, smoke-test, and gateway-ready prompt. A license key is required — set DAI_OS_LICENSE_KEY in your env, or paste it when prompted.
What's in the bundle (10 skills)
| Skill | What it does |
|---|---|
session-keeper | Daily browser-session audit at 09:00 — keeps long-lived logins alive on a schedule. |
delegation-orchestrator | Cross-agent fault-tree help routing with a 60-second daemon tick. |
hermes-inbox-router | Multi-channel inbound aggregation across Telegram, email, Slack. |
outreach-paced | Reddit and forum outreach pacing with the anti-self-promo gate at Step 3.5. |
model-warmth-keeper | Local LLM auto-recovery on Ollama (http://127.0.0.1:11434) and LM Studio. |
launchd-drift-recovery | Daily macOS launchd plist health audit at 06:00. |
ship-log-draft | Friday 17:00 CT 7-day aggregation of git, postmortems, auto-skills, ops notes. |
skill-registry-sync | Auto-wire owner_agent for every new SKILL.md the runtime discovers. |
voice-gate | Public-output quality scoring against a VOICE.md reference. |
public-content-scrubber | Hard gate against financial-data leaks in any public-bound text. |
Each skill is a directory with SKILL.md (frontmatter + description), run.sh (executable bash), INSTALL.md (env vars + smoke), and EXAMPLES.md (1-2 sanitized real outputs).
Tiers
| Tier | Price | What's included |
|---|---|---|
| Starter | $99/mo | 1 user, this 10-skill bundle, Discord access |
| Team | $499/mo | Unlimited users, full 200+ skill library, monthly office hours with Cooper |
Buy at https://usedailyai.com/os.
What you bring
This OS runs on your hardware. Bring:
- A Mac with macOS 14+ on Apple Silicon (M1, M2, or M3) and ≥ 16 GB RAM
- A free Telegram bot token from
@BotFather(2 minutes) - A ChatGPT Plus or Pro subscription — Hermes routes through Codex OAuth
- Optional: a Stripe account for the e-commerce skills
- Optional: a GitHub PAT for the dev-loop skills
What this is, what this isn't
It is: 12 working skills, an install script, a runtime contract, and a CI workflow that validates every skill on every push. Honest, opinionated, evolving.
It isn't: a hosted SaaS. It isn't a guarantee that any particular skill produces a measurable business outcome on your stack. It isn't a replacement for understanding what your agents are doing — every skill is auditable bash, every output passes through voice-gate and public-content-scrubber you can read.
If a skill produces output that looks wrong, the source is in skills/<name>/run.sh and you can fix it yourself. That's the deal at the Starter tier.
Honest disclaimers
- Outputs are not guaranteed. Skills produce drafts; you decide what ships.
- Costs are yours to manage. Local model inference is free; cloud LLM calls and Stripe transaction fees are not.
- Compatibility is best-effort. Hermes 0.11.x and OpenClaw 2026.4+ are independent projects with their own release cadences. We pin known-good versions; major upstream changes may briefly break.
Updates and support
- Newest skills land here weekly. Watch the
CHANGELOG.mdfor the running list. - Discord (invite arrives in your activation email) is the primary support channel.
- Bug reports: GitHub issues are read; reply volume varies.
- Pro and Team tiers get email + priority Discord.
License
MIT. Use the bundle as you want. Attribution appreciated, not required.
Daily AI Agents · An autonomous AI-native solo founder operating system.
Featured
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
