Self-hosted bidirectional GitHub mirror
Problem
When GitHub has outages or gets slow, teams and individuals who depend on it for reviews and collaboration are stuck. An HN user asks for a self-hosted GitHub instance that acts as a precise mirror of a real GitHub repo: PRs from GitHub appear locally and reviews made locally sync back to GitHub, so work continues offline and appears online automatically. They end with "Does this exist?"
Opportunity
A sync service/appliance that maintains a live self-hosted Gitea/GitLab mirror of GitHub repositories with bidirectional sync of issues, PRs and reviews, acting as a transparent failover for GitHub outages.
Market analysis
One-way mirroring is a solved problem — Gitea has native repo mirrors, and tools like gitea-mirror and gitmirror already replicate repos, issues, PRs, releases and wikis from GitHub to self-hosted instances. The ask therefore reduces to the genuinely unsolved part: bidirectional sync of reviews and PR state back to GitHub, which is hard, thankless, and monetizes poorly against an episodic pain.
Market · Teams and OSS maintainers dependent on GitHub uptime; outages reliably produce HN anger, but the pain is intermittent and few pay for failover of a free service.
Pricing · The adjacent ecosystem is open-source and free (Gitea, gitmirror); monetization would have to come from hosted sync tiers or support contracts — historically thin.
Pros
- + The bidirectional sync gap is real: no tool found that round-trips local reviews back to GitHub.
- + Self-hosted/OSS audience is reachable for free via HN, Reddit and the Gitea/Forgejo community.
- + Git objects already sync trivially; only the collaboration metadata layer is missing.
Cons
- − Demand spikes only during outages — episodic pain rarely sustains a product.
- − Bidirectional review sync has brutal edge cases: divergent reviews, merges during the outage, identity mapping.
- − GitHub API rate limits and Terms of Service constrain a third-party sync service.
Existing / similar tools
Source
Hacker News (Ask HN)
The honest scoping insight is that “full bidirectional failover” is a distributed-systems nightmare — a review posted on the local mirror during an outage, while the PR author kept pushing to GitHub once it recovered, has no clean reconciliation, and pretending otherwise is how sync tools destroy trust. The shippable version is weaker but real: a read-only live mirror plus an outbound queue, where review comments typed locally are held and replayed to GitHub through its API the moment connectivity returns. Nobody loses work, no conflict semantics are promised, and the 95% case (reading PRs and queuing comments during a brownout) is fully covered. Sell it as outage insurance, not as a second GitHub.