What We've
Just Shipped.

Engineering-content angle on the v1.7 MCP launch. Walks through the three plausible deploy shapes for an MCP server next to a Rails SaaS (one dyno, multi-procfile, separate repo), why two-dyno-one-repo is the right default, how the multi-procfile buildpack works in twelve lines of bash, three things we got wrong on first attempt (release running twice, log-level uniformity, MCP routes in main routes.rb), why a shared bearer token is a feature not a smell, the monitoring shape (latency/error/rate-limit by dyno), and when this design starts to creak. Targets the rails+mcp+heroku engineering audience that complements the existing practitioner posts.

Public marketing landing page at /mcp documenting the MCP server: endpoint, eight day-one tools (list-projects, list-sprints, list-issues, my-issues, get-issue, create-issue, update-issue, create-comment), the two deliberately-omitted destructive tools (delete-issue, delete-sprint) with rationale, 30-second per-editor setup grid, auth/rate-limit/security notes, and a 7-question FAQ. Emits WebAPI + BreadcrumbList JSON-LD; sitemap at 0.9; cross-linked from footer and llms.txt; cross-links to /tools/mcp-config-generator and the four editor integration pages.

"Claude Desktop for sprint management — five real use cases (and three things it's not good at)" — 9-min practitioner's guide. Five use cases that justify the setup (Monday triage, what-to-work-on-next, bug-report-to-ticket, status updates, retro prep), three failure modes (strategic questions, destructive ops, long-horizon forecasting), 30-second setup, and a token-hygiene note. Pairs with /blog/sprint-state-in-your-agent-via-mcp, /tools/mcp-config-generator, /for/claude-desktop.

New free tool: pick your AI editor and copy a paste-ready MCP server config block targeting mcp.sprintflint.com. Supports Claude Desktop (mcpServers in claude_desktop_config.json), Cursor (Settings → MCP), Claude Code (`claude mcp add` CLI one-liner), and Zed (context_servers in settings.json). Pure client-side — token never leaves the browser. Embeddable iframe. WebApplication JSON-LD; sitemap at 0.85; cross-linked from /tools, /api, /integrations, and the v1.7 MCP relaunch press release.

Tenth editor integration. Pure-MCP setup — drop one JSON block into claude_desktop_config.json, restart, ask Claude about the sprint. Routes at /for/claude-desktop. Listed in /integrations index and llms.txt editor list.

New glossary term covering MCP: protocol explainer, three primitives (tools/resources/prompts), MCP-vs-REST decision table, 30-second Claude Desktop setup, and three common misconceptions. Routes at /glossary/mcp; cross-linked from /api, /blog/sprint-state-in-your-agent-via-mcp, /blog/sprints-with-ai-teammates, and /integrations.

Public press release announcing the v1.7 MCP relaunch at /press/v1-7-mcp-relaunch, plus the marketing polish bundled in the same release window (atom feed, BreadcrumbList JSON-LD, header on every marketing page, 4-col grouped footer, unified tools CTA, mobile API card overflow fix, Turbo cache fix on landing video modal). Listed first on /press; sitemap at 0.8.

SprintFlint's Model Context Protocol server is back online at https://mcp.sprintflint.com/, hosted on a separate sprintflint-mcp Heroku app via Procfile.mcp + the heroku-buildpack-multi-procfile buildpack, with automatic TLS via Heroku ACM. Server identity `sprint-flint-rails`, MCP protocol 2025-06-18 (also supports 2024-11-05), capabilities: tools, prompts, resources, logging. Bearer token auth (same token as the REST API, from account settings). Connect from Claude Desktop, Claude Code, Cursor, Goose, or any MCP-compatible client to read tickets, create issues, advance sprints, and query velocity directly from the editor — no copy-paste between tracker and assistant. /api landing page now describes the MCP surface alongside /llms.txt and the OpenAPI spec, with a copy-paste Claude Desktop config snippet. Restores the MCP claim that was scrubbed in the 2026-05-05 messaging cleanup (#193) — claim is now backed by a live, verified server.

"Sprint state in your agent — what changes when MCP makes the board reachable from the editor" — 8-min practitioner's guide pairing with the v1.7 MCP relaunch. Explains the protocol in one paragraph, then covers the four real patterns it unlocks (pull-mode planning, auto-PR comments, sprint-aware code review, standup pre-write), the eight tools we shipped on day one (and the two we deliberately didn't — `delete-issue` and `delete-sprint`), 30-second Claude Desktop setup, and the four ways teams misuse the integration in the first two weeks. Pairs with /blog/sprints-with-ai-teammates, /api, /integrations.

"Running sprints with AI teammates — what actually changes" — 13-min practical post on hybrid human/agent teams. What changes in planning (two queues, spec-quality gate), velocity (goes up, gets noisier), code review (review becomes the bottleneck, three new failure modes including silent correctness bugs in confidently-written code), retros (the agent setup is half your throughput, retro it explicitly with rules-file updates as action items), and the four mistakes teams make in the first three sprints. Pairs with /blog/ai-in-agile-workflows, /blog/velocity-dropped-what-to-do, /blog/agile-estimation-techniques-compared; tools: /tools/retro-format-picker, /tools/sprint-capacity-calculator, /tools/story-splitter.

"Agile estimation techniques compared" — 12-min side-by-side of the five real estimation techniques you'll meet in production agile. Comparison table (what each surfaces, where each breaks, setup cost), per-technique deep dive (story points, planning poker, t-shirt sizing, ideal hours, no-estimates), the three honest questions to ask before switching, the four-step switch playbook, and the one rule that beats technique choice — whole-loop trust between engineers, PM, and stakeholders. Pairs with /blog/story-points-vs-hours, /blog/story-point-estimation, /blog/velocity-dropped-what-to-do; tools: /tools/story-points-estimator, /tools/sprint-forecaster.

The MCP server was deprecated in v1.2 in favour of agent skills, but eight comparison pages, the comparisons hub, and llms.txt still claimed "First-class MCP server" as a differentiator. Scrubbed all MCP claims in marketing surfaces and replaced with the actual current integration story: REST API + agent skills + /llms.txt for AI editors. Press v1.1 and v1.2 left intact as historical record. Aligns marketing copy with the deprecation that already shipped.

New free tool: Retro Format Picker. Six weighted questions (sprint outcome, morale, last retro recency, topic, action follow-through, team size) score across six retro formats — Start/Stop/Continue, Mad/Sad/Glad, 4Ls, Sailboat, Five Whys, Lean Coffee. Recommendation includes why-it-fits, 60-min run-of-meeting steps, watchouts, and runner-up. Pairs with /retrospectives cluster, retro-action-tracker template, and the why-retros-die blog post. Embeddable iframe.

Five high-search-volume terms added to the glossary: Daily Standup, Sprint Review, Acceptance Criteria, Lead Time, Scrum Master. Each term includes a tagline, full plain-English definition page, common failure modes, and cross-links to related glossary terms, tools, and long-form blog posts. Glossary now covers 25 terms total.

/api was an alias to /api-docs (which 404s on bare URL because Rswag only serves /api-docs/index.html). Replaced redirect with a real landing page: hero, Bearer-token auth example, code samples in curl/Ruby/Python/Node, resource model overview, AI-agent surface (/llms.txt + swagger.yaml), rate-limit conventions, and a one-click jump into Swagger UI. Adds TechArticle JSON-LD, sitemap entry at 0.85, IndexNow ping. Replaces the API footer link target.

SprintFlint v1.6 ships five curated topic-cluster landing pages (forecasting, retrospectives, sprint planning, story management, metrics), public /pricing with Product schema, sixteen URL aliases closing canonical-path 404s, CollectionPage JSON-LD on every cluster, and a hub blog post mapping clusters to sprint failure modes.

Most SaaS visitors type canonical paths like /signup, /login, /trial, /api, /demo, /faq, /security, /support directly. All were 404. Added 16 redirect routes covering signup variants (signup, sign-up, trial, free-trial, start, get-started → magic-link signup), login variants (login, log-in, signin, sign-in → log_in), and aliases for /api, /faq, /demo, /security, /support. Plus 14 spec assertions.

/pricing was 404 — closed the discoverability gap. New page surfaces the single £5/user/month plan with both billing intervals, lists what's included, surfaces the free tools/templates/blog/glossary, and answers 7 FAQs (why one plan, free tier, what's included, BYO AI key, cancellation, nonprofit/OSS discount, migration). Adds Product JSON-LD with offers for SERP rich-results, sitemap entry at priority 0.95, IndexNow ping, and a footer link.

Hub blog post that maps the five free cluster toolkits (planning, story management, forecasting, retrospectives, metrics) to the five places sprints actually break, with a symptom-based triage at the end. Internal-link anchor for the cluster strategy and the natural funnel from blog browsers into the curated toolkits.

/metrics bundles the four assets that fix sprint measurement: the metrics-that-matter playbook (frame what's worth tracking), velocity calculator (capacity sense), cycle time calculator (flow + p85 prediction), and the velocity-drop diagnosis (what to do when it shifts). Plus four related posts (sprint-velocity-benchmarks, burndown-charts-truth, forecasting-with-p85, story-points-vs-hours). Fifth curated topic-cluster landing page.

/tools index now leads with a "Curated toolkits" section linking the four cluster landing pages (forecasting, retrospectives, sprint-planning, story-management) above the alphabetical tool list. High-CTR placement to keep visitors in-funnel and reinforce the topic-cluster strategy.

/story-management bundles the four artifacts that fix carry-over stories: the INVEST-applied playbook (write right), the story splitter tool (split too-big stories), the story points estimator (size relative not absolute), and the acceptance-criteria formats guide (catch hidden scope). Plus four related posts (definition-of-ready, backlog-refinement-runbook, story-point-estimation, sprint-backlog-vs-product-backlog). Fourth curated topic-cluster landing page after /forecasting, /retrospectives, /sprint-planning.

/sprint-planning bundles the four assets that make planning meetings ship what they commit: the 60-min planning agenda blog (read first), capacity calculator (bound capacity), goal generator + validator (anchor on outcome), and Monte Carlo forecaster (forecast at p85). Plus four upstream-prep posts (definition-of-ready, backlog-refinement-runbook, mid-sprint-scope-changes, story-splitter). Third curated cluster after /forecasting and /retrospectives.

/retrospectives bundles the four assets that keep retros from dying: the why-retros-die diagnosis blog (read first), the retro template generator (run the meeting), the retrospective action tracker template (the artefact that fixes follow-through), and the sprint-review-without-demo-theatre companion. Plus four related posts (em-burnout-sprint-patterns, agile-metrics-that-matter, hidden-cost-of-standup, mid-sprint-scope-changes). Second curated topic-cluster landing page after /forecasting.

/forecasting bundles the four assets that actually move the forecasting needle into one funnel: the p85 forecasting guide (read it first), the cycle time calculator (analyse your data), the sprint forecaster (Monte Carlo at p85), and the burndown chart generator (track it after commit). Plus four related posts (agile-metrics-that-matter, velocity-dropped-what-to-do, burndown-charts-truth, story-point-estimation). First curated topic-cluster landing page — pattern to repeat for sprint planning, retros, and refinement.