›
›
Bottom line: Daemons targets a real, under-served problem — the operational debt that coding agents leave behind. The execution model (scoped policy files, explicit approval gates, verified checkpoints) is unusually well thought-out for a 24-day-old product. It’s not a Cursor or Claude Code replacement; it’s a maintenance layer that sits alongside them. Worth trying free if your team already runs multiple coding agents and PRs are piling up faster than anyone can review them.
What Is Daemons?
A note on this review: I haven’t run Daemons in a live production repo yet, so treat this as a research-based first look rather than a months-long field test — built from Charlie Labs’ own docs, pricing page, and execution-model documentation, plus how it stacks up against the coding agents I use daily, Claude Code and Cursor. If I put it into daily use, I’ll come back and add hands-on notes.
Daemons is a product from Charlie Labs, the team behind the Charlie coding agent, launched on Product Hunt in July 2026 where it landed #3 Product of the Day with 277 upvotes — solid, credible traction for a brand-new category rather than a launch-day spike that fades.
The pitch is simple and, honestly, overdue: “agents create work, daemons maintain it.” Coding agents like Claude Code and Cursor have made it trivially easy to generate pull requests, issues, and docs changes. What nobody automated is the unglamorous follow-through — fixing a flaky CI check, updating a stale doc after a merged PR, triaging a Sentry alert, nudging a PR that’s gone quiet. Daemons are persistent, role-scoped agents that watch for those signals and handle the loop without a human re-prompting them every time.
Each daemon is defined in a DAEMON.md file stored at .agents/daemons/<daemon-name>/, specifying what it watches for, what routines it’s allowed to run, and what it must never do. That’s a meaningfully different design philosophy from a general-purpose chat agent: a daemon has one narrow job, wakes on a specific trigger, and reports back with a reviewable artifact rather than acting invisibly in the background.
Key Features
Here’s what actually defines the Daemons product in 2026:
Core concept
Seven-Stage Execution Loop
Intake → context build → plan → execute → verify → report → continue. Every stage produces a saved artifact, so runs can resume cleanly after an interruption instead of restarting from scratch.
Unique
Autonomy & Approval Matrix
Read-only classification and routine metadata edits run fully automatic. Anything that expands scope, merges, or closes an issue outside declared policy pauses and asks a human one focused question.
DAEMON.md Policy Files
Plain-Markdown contracts define watch triggers, allowed routines, hard “deny” boundaries, verification commands, and a cron schedule for periodic sweeps — version-controlled alongside your code.
Idempotent, Resumable Runs
Duplicate events reconcile to one outcome instead of spamming duplicate comments. Failed stages retry with bounded backoff and resume from the last safe checkpoint rather than looping.
Native GitHub, Linear, Slack, Sentry
Daemons post durable updates — PR comments, check runs, threaded Slack replies — directly where your team already works. No dashboard you have to remember to check.
Run Artifact Timeline
Every run leaves a linked, auditable trail: merge event → context snapshot → draft branch → verification results → status report. Useful for teams that need to explain what an agent actually did.
Pros & Cons
✅ Pros
- Free tier is genuinely usable — runs several daemons continuously
- Explicit approval gates for anything risky or irreversible
- Config-as-code via version-controlled DAEMON.md files
- Idempotent runs avoid duplicate-comment spam, a common agent annoyance
- Strong Product Hunt reception (#3 Day, 277 upvotes) for a brand-new category
- Shares usage meter with Charlie’s coding agent — no second subscription needed if you already use Charlie
❌ Cons
- Only four native integrations (GitHub, Linear, Slack, Sentry) — no Jira, Asana, or Notion yet
- Engineering-only use case; not for non-technical teams
- Usage limits are daily/weekly, not simple monthly credits — takes some getting used to
- Very young product (24 days old at review time) with no long-term reliability track record
- Small public community (75 PH followers) compared to established competitors
Pricing Explained
Charlie Labs doesn’t sell Daemons on a monthly-credit model — plans are daily/weekly usage limits shared between daemons and human-invoked Charlie sessions.
| Plan | Price | What You Get | Best For |
|---|---|---|---|
| Free | $0 | Free-plan daily + weekly usage limits, unlimited team members, prepaid overage available. | Trying daemons on a small repo |
| StarterPopular | $50/mo | 2.5× the Free plan’s daily/weekly limits, unlimited team members, prepaid overage. | Teams running daemons regularly |
| Team | $200/mo | 10× Free plan limits, priority Slack support, unlimited team members. | Scaling always-on workloads |
| Growth | $1,000/mo | 50× Free plan limits, priority Slack support, unlimited team members. | High-throughput daemon fleets |
Daemons vs Cursor Background Agents vs GitHub Copilot
Daemons isn’t really competing head-to-head with coding agents — it’s positioned to run alongside them. Here’s how the “keep things moving after code ships” layer compares across the three:
| Feature | Daemons | Cursor Background Agents | GitHub Copilot Agent |
|---|---|---|---|
| Price | Free – $1,000/mo | Included in Cursor Pro ($20/mo) | Included in Copilot Pro ($10–19/mo) |
| Runs without a human prompt | ✓ Event/schedule-driven | ✗ Prompt-initiated | ~ Assigned-issue driven |
| Config-as-code policy files | ✓ DAEMON.md | ✗ No | ✗ No |
| Explicit approval gates | ✓ Built-in matrix | ~ Diff review only | ~ PR review only |
| Native Slack + Sentry | ✓ Yes | ✗ No | ✗ No |
| IDE / editor experience | ✗ None — ops layer only | ✓ Full IDE | ✓ Full IDE |
| Maturity / track record | ✗ ~1 month old | ✓ Established | ✓ Established |
| Best for | Post-merge maintenance | Writing & editing code | Writing & editing code |
Cursor and Copilot win the moment you’re actually writing code — that’s not what Daemons is for. Daemons wins the moment code has already shipped and someone needs to keep CI green, docs current, and issues triaged without a human remembering to ask. If your team already reviewed our Cursor AI review or our OpenCode vs Claude Code comparison, Daemons is the natural next layer on top of whichever coding agent you land on.
Performance Ratings
Who Should Use Daemons?
✅ Great fit
Engineering teams already shipping heavily with AI coding agents (Claude Code, Cursor, Copilot) who are drowning in the follow-through — stale docs, unreviewed PRs, flaky CI, ignored Sentry alerts. If your repo lives in GitHub with Linear for issues and Slack for team chat, Daemons plugs in with zero new tools to learn. Teams that want config-as-code control over what an agent is and isn’t allowed to touch will appreciate the DAEMON.md model far more than a black-box automation tool.
⚠️ Think carefully
If your stack runs on Jira, Notion, or Asana instead of Linear, there’s no native integration yet. Non-engineering teams get nothing from this — it’s built exclusively around software development signals. And if you need a mature, years-proven product for anything touching merge or deploy decisions, the newness of Daemons is a legitimate reason to wait and watch before giving it broad autonomy.
Final Verdict
Daemons is one of the more thoughtfully designed products to come out of the “post-agent operational debt” problem in 2026. The seven-stage execution loop, explicit approval matrix, and version-controlled policy files show a team that’s clearly thought hard about trust and reliability — not just autonomy for its own sake.
The catch is simply time. A 24-day-old product with four integrations and no long track record is, by definition, unproven at scale. The free plan makes it low-risk to find out for yourself, and the pricing — unlimited team members on every tier — is genuinely generous next to per-seat competitors.
Recommendation: If your team runs GitHub + Linear + Slack and is already generating more PRs than you can maintain, start a low-stakes daemon (docs updates or issue labeling) on the free plan today. Hold off on merge-authority daemons until the product has more runway behind it.