Retool
Retool is retiring standalone Assist while folding the same capability into the app builder.
A side-by-side editorial comparison of gh and Merge — release velocity, themes, recent moves, and the top alternatives to consider.
| Feature | gh | Merge |
|---|---|---|
| Sector | Infra & APIs | Infra & APIs |
| Velocity score | 2.5 | 6.3 |
| Sparks · 30d | 0 | 1 |
| Top themes | github-api, http-client, pagination, token-handling | unified api, agent handler, mcp connectors, ai gateway |
| Last editorial update | 5d ago | 20h ago |
| Website | Visit → | — |
gh spent its last two releases making failures and interruptions recoverable.
The GitHub API client rebuilt on httr2 in 1.4.0, then took a security-driven breaking change in 1.5.0: response headers are no longer stored in returned objects, because they can carry sensitive information. The 1.6.0 release made interrupted pagination recoverable and downgraded personal access token format validation from an error to a warning.
Merge is turning its weekly integration digest into an agent control plane — the news is always at the bottom.
Merge publishes one dated digest a week, and the structure is consistent: unified Accounting, ATS, CRM, File Storage and HRIS reliability work up top, then Agent Handler and Gateway at the end, where the directional changes live. The feed now also carries per-product breakout entries (Gateway, Unified, Agent Handler) that restate the same week's items in more detail rather than adding new ones. The Agent Handler catalog carries hundreds of generic MCP connectors alongside Merge's own, on shared authentication and policy, and Gateway has been accumulating model coverage, routing controls and guardrails over the same period.
The GitHub API client rebuilt on httr2 in 1.4.0, then took a security-driven breaking change in 1.5.0: response headers are no longer stored in returned objects, because they can carry sensitive information. The 1.6.0 release made interrupted pagination recoverable and downgraded personal access token format validation from an error to a warning.
The direction is tolerance for partial and unusual outcomes rather than new endpoint coverage. Interrupting a paginated call now raises a classed condition carrying the records already fetched; a 304 Not Modified returns an empty response with headers intact instead of erroring; an unrecognised token format warns and proceeds. A fake_github_app() built on webfakes ships for testing, and is offered to other package authors.
Token format handling is now configurable rather than fixed, so the next likely work is following GitHub's credential formats as they change, not expanding the client surface.
Merge publishes one dated digest a week, and the structure is consistent: unified Accounting, ATS, CRM, File Storage and HRIS reliability work up top, then Agent Handler and Gateway at the end, where the directional changes live. The feed now also carries per-product breakout entries (Gateway, Unified, Agent Handler) that restate the same week's items in more detail rather than adding new ones. The Agent Handler catalog carries hundreds of generic MCP connectors alongside Merge's own, on shared authentication and policy, and Gateway has been accumulating model coverage, routing controls and guardrails over the same period.
The through-line is that Merge is repositioning from data plumbing to the layer agents pass through. Each week adds either reach (more connectors, more tools per connector, more models) or control (guardrails, per-project policy, access configuration). This window is reach-and-efficiency: the GitHub connector expanded to 145 tools, Outlook payloads were cut substantially, and Gateway guardrails became configurable per project rather than per account.
Expect the per-project granularity applied to guardrails to spread to the rest of Gateway's controls, and the connector catalog to keep absorbing hosted third-party MCP servers the way Axiom was added.
Other Infra & APIs 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 gh or Merge.
Retool is retiring standalone Assist while folding the same capability into the app builder.
WPML made machine translation the default, and its point releases keep chasing WordPress and page builders.
A forest plot package that keeps handing users control of one more graphical detail.
Interval-valued data plotting, spending 2026 making its function names and examples survive CRAN.
A microbiome network model that got itself un-archived by deleting the dependency that killed it.
Three releases in ten days, every one of them a CRAN reviewer's correction rather than a code change.
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Merge is currently shipping more aggressively (velocity 6.3 vs 2.5), with 1 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. Merge is currently shipping more aggressively (velocity 6.3 vs 2.5), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Infra & APIs products to evaluate alongside.
Top gh alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "gh alternatives" section above for the current picks, or visit /alternatives/gh for the full list with editorial commentary on each.
Top Merge alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Merge alternatives" section above for the current picks, or visit /alternatives/merge-dev for the full list with editorial commentary on each.