CommaFeed
CommaFeed is patching its way through the attack surface a self-hosted reader inherits
A side-by-side editorial comparison of Kavita and SiYuan — release velocity, themes, recent moves, and the top alternatives to consider.
A self-hosted reading server that spent two releases becoming infrastructure, then paid for it with a CVE.
Kavita ships two to four substantial releases a year with hotfixes trailing each one, and the last two years moved it well past a comic reader: OIDC login, an annotation system, a rebuilt epub reader, a reading-list overhaul, a stats system, and a scanner reported 50x faster. The most recent release is a critical security hotfix carrying CVE-2026-47202 that affects every prior version. Kavita+ remains the paid tier where metadata features land first.
SiYuan stabilises 3.8.1 after a seven-build beta run, all of it widening the agent surface it opened in 3.8.0
SiYuan has just cut 3.8.1 stable, the endpoint of a beta chain that ran six builds in five days. The release groups into three areas: AI integration work - custom editor actions syncing across devices, an Agent panel shortcut, user-level skills read from a shared agents directory, nested rerank API formats and OpenAI Responses API support; editing and database throughput, including virtualised grouped-table rendering and faster document loading; and cross-platform reliability covering HEIF/HEIC previews, PDF export and mobile behaviour. A batch of MCP and publishing-service fixes rounds it out.
Kavita ships two to four substantial releases a year with hotfixes trailing each one, and the last two years moved it well past a comic reader: OIDC login, an annotation system, a rebuilt epub reader, a reading-list overhaul, a stats system, and a scanner reported 50x faster. The most recent release is a critical security hotfix carrying CVE-2026-47202 that affects every prior version. Kavita+ remains the paid tier where metadata features land first.
The arc is a hobby-scale project taking on the obligations of infrastructure — standard auth, an external API contract for third-party clients, sync with other readers, and formal vulnerability disclosure. Each of those decisions expands the attack surface, and the 0.9.x line shows the bill arriving: security hardening now leads the release notes rather than following them. The team has also visibly grown from one maintainer to a named pair, which is what made the parallel reader and foundation rewrites possible.
The next release is likely the delayed scanner-improvement work that 0.8.7 displaced, with continued API hardening rather than new reader surfaces. Expect the CVE details to be published once adoption of 0.9.0.2 is high enough.
SiYuan has just cut 3.8.1 stable, the endpoint of a beta chain that ran six builds in five days. The release groups into three areas: AI integration work - custom editor actions syncing across devices, an Agent panel shortcut, user-level skills read from a shared agents directory, nested rerank API formats and OpenAI Responses API support; editing and database throughput, including virtualised grouped-table rendering and faster document loading; and cross-platform reliability covering HEIF/HEIC previews, PDF export and mobile behaviour. A batch of MCP and publishing-service fixes rounds it out.
This is consolidation of the surface 3.8.0 opened, not an extension of it. The agent, semantic search and MCP endpoint arrived in the 3.8.0 line; 3.8.1 makes them portable and pluggable - skills come from a directory the user controls, editor actions follow the user between devices, and the provider layer now speaks the OpenAI Responses API rather than one bespoke shape. The high beta count against a modest final diff shows a project stabilising a young subsystem rather than building a new one.
The next line most likely pushes on the skills directory - discovery, sharing or marketplace distribution of user-level skills - since that is the piece 3.8.1 made user-editable without giving it a distribution path.
Other Collab products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Kavita or SiYuan.
CommaFeed is patching its way through the attack surface a self-hosted reader inherits
Hive ships in batches, and this one is all planning accuracy and admin control.
Teable ships daily, and the work has moved from grid features to platform governance.
Simpplr publishes the research that names the gap, then ships the product that closes it.
NetNewsWire's 7.1.3 train has moved from rebuilding sync to sweeping up what the rebuild disturbed.
Document360 rebuilt its API for agents; now it's turning the AI inward on authoring.
See all Kavita alternatives → · See all SiYuan alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
Both compete on the same themes — self-hosted — within Collab. SiYuan is currently shipping more aggressively (velocity 6.3 vs 0.0), with 0 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. SiYuan is currently shipping more aggressively (velocity 6.3 vs 0.0), with 0 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Collab products to evaluate alongside.
Top Kavita alternatives in Collab are ranked by recent ship velocity. Browse the "Kavita alternatives" section above for the current picks, or visit /alternatives/kavita for the full list with editorial commentary on each.
Top SiYuan alternatives in Collab are ranked by recent ship velocity. Browse the "SiYuan alternatives" section above for the current picks, or visit /alternatives/siyuan for the full list with editorial commentary on each.