The source says CogSend is MIT-licensed, runs in the user’s Cloudflare account, and currently supports Mastodon, Bluesky, LinkedIn, Threads, and X. It also says a post can have platform-specific overrides, while long copy can be split into a thread according to the tightest selected platform’s counting rules. These are implementation claims from the project’s author, not an independent evaluation.
The mechanism is per-destination orchestration: the scheduler records a result for each platform, lets a failure remain local to that destination, retries retryable errors with backoff, and exposes failed destinations for selective repair and retry. Credentials are described as encrypted at rest; access is limited to one admin account with an authenticator app, while an API key connects scripts, Shortcuts, and cron jobs. This combines a human interface with a small automation surface rather than treating publishing as one opaque bulk action.
why_now: social publishing workflows increasingly span platforms with conflicting formats and unreliable APIs, while solo operators want fewer recurring SaaS dependencies and more control over credentials. The project’s release is timely because it packages this need into a deployable Cloudflare-based instance rather than a bespoke script.
The first implication is that a solo publisher can make platform routing, copy variation, and recovery from partial delivery failures part of infrastructure they control. A second implication is that the API key may make CogSend useful as a final delivery layer for writing agents or scheduled internal workflows, provided the operator accepts the operational burden of account connections and platform policy changes.
evidence_boundary: the available evidence is a single primary announcement, so it establishes the author’s stated capabilities and deployment model but does not establish production reliability, security quality, maintenance cadence, API compliance, or real-world cost. The author’s statement that Cloudflare’s free plan is sufficient applies to a personal instance; R2 still requires a payment method on file.
The first watch condition is whether the public project gains independent deployment reports, active maintenance, and evidence that its per-platform retry and credential-handling model remains reliable as APIs change. Also watch whether documentation makes the required Workers, D1, and R2 setup reproducible without repository-level troubleshooting, and whether proposed Instagram, TikTok, and YouTube integrations become supported releases rather than roadmap ideas.