Havoptic MCP server

• Registry namecom.havoptic/releasesverified in the MCP registry

The registry name sits under havoptic.com. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

• Stateanswerschecked 2026-10-20
• Tools15
• Server calls itselfhavoptic
• Server version2.1.0
• Protocol version2025-06-18

On 2026-10-20 the server answered initialize and tools/list without a sign-in. First answer in 207 ms.

• Its tools do every use case we checked: follow daily release notes across many AI coding tools; compare recent version changes between Claude Code and Cursor; check which AI coding tools shipped updates this week.
• Checked on 2 of the 7 days it takes to judge reliability.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

• INITE MCP scoreafter a weeka score needs 7 daily checks: 2 so far
• What an agent can do with it3 of 3 use cases
• Answeredmeasuring2 of the 7 daily checks it takes to print a share
• Median first answer366 ms
• Space in an agent's context≈ 3,592 tokensthe definitions of all its tools (15), which an agent carries on every turn
• Tool definitionsA · 98 of 100
• Typed results10 of 15 toolsdeclared outputSchema: results come typed, read once

• Parameters without a description: in 2 of 15

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (30%), doing the product's job (25%), definitions (20%), context (10%) and use (15%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

• list_tracked_tools

  Deprecated: use list_tools. All AI coding tools Havoptic tracks (Claude Code, OpenAI Codex CLI, Cursor, Gemini CLI, Kiro CLI, GitHub Copilot CLI, Devin Desktop/Windsurf, Antigravity CLI, Grok Build), with release counts and their latest…

• latest_releases

  Deprecated: use get_latest_releases. Most recent AI coding tool releases (newest first), optionally filtered to one tool. Returns summaries; use get_release for full notes. Default detail=brief omits infographic URLs.

• release_history

  Deprecated: use get_latest_releases with since. Release history for one tool within an optional date window (newest first). Summaries only; use get_release for full notes. Default detail=brief omits infographic URLs.

• search_release_notes

  Deprecated: use search_changes. Full-text search across all tracked release notes — flags, error strings, feature names, config keys. Returns ranked matches with snippet context.

• feature_timeline

  Deprecated: use feature_adoption. Which tool shipped each notable cross-tool feature first (MCP support, subagents, planning mode, hooks, checkpoints, ...), follower lag in days, and copying-trend insights. Optionally filter to one feature.

• list_tools

  Catalog of the products Havoptic tracks (AI coding tools plus model/API platforms, agent SDKs and inference runtimes): id, name, vendor, category, release count and latest release. Use first to find valid tool ids (fuzzy query: 'copilot'…

• get_latest_releases

  Newest releases (newest first) across all or selected tools. Successor to latest_releases and release_history (pass since/until for a date window). Default channel=stable, detail=brief.

• whats_changed_since

  Digest of typed changes (breaking, security, features, fixes, …) since a date or relative window ('7d', '30d'), grouped by tool with breaking changes first. States how many releases in range are legacy records without typed changes.

• search_changes

  Search typed change items (title + evidence) and, unless kinds is set, legacy release notes — flags, error strings, feature names, config keys. Successor to search_release_notes. Brief mode returns snippets only.

• get_release

  One release by id (e.g. 'claude-code-2.1.183', canonical '[email protected]') or by tool + version ('latest' allowed). detail=full (the default for a bare id lookup, as before) includes complete release notes; brief/standard omit them…

• get_upgrade_notes

  Everything between two versions of one tool: breaking changes, deprecations, security fixes, features and fixes, plus an explicit coverage gap for legacy releases without typed changes. to_version defaults to latest.

• get_breaking_changes

  Breaking changes (and, by default, deprecations) across tools in a window (default last 30 days), newest first. Legacy releases without typed changes are counted, not guessed.

• compare_tools

  Compare 2–4 AI coding tools: release cadence, feature-matrix differences, recent changes, and Havoptic's editorial verdict per pair. aspects defaults to cadence, features, verdict. Legacy inputs tool_a/tool_b still work.

• feature_adoption

  Who shipped a cross-tool feature first (MCP support, subagents, planning mode, hooks, checkpoints, …), which tools followed and their lag in days. Successor to feature_timeline.

• cadence_stats

  How fast tools ship: release counts over 7/30/90 days, average days between releases, and last release — for selected tools or all of them. Legacy input tool (single id) still works.