It reads before it writes.
Dodger researches the market on its own: competitors, reports, reviews. Connect your tools and it also works from what only your team has: what your users do, what your team knows, and where the finished work should land.
What users do
Product analytics ground every answer and every spec. Dodger reads what actually happened, not what anyone remembers.
Funnels, retention, and what changed after every release.
Traffic, sources, and where visitors drop off.
On the wayUsage patterns across your features.
On the way
AmplitudeOn the wayCohorts and long-run retention.
On the wayWhat your team knows
Decisions live in docs, meetings, and support threads. Dodger reads them where they are.
Specs, decks, and research docs, read where they live.
Meeting notes become evidence Dodger can cite.
Dodger joins the meeting, listens, and answers in the room.
What customers ask when something is unclear.
On the waySupport threads from your busiest channels.
On the wayWhere the work lands
Dodger's work shows up in the tools your team already ships with.
Ask Dodger anything and approve its actions without leaving the channel.
Epics and stories arrive in the backlog, linked to their evidence.
Pull requests with checks, review, and a recording of the feature working.
Wireframes arrive as editable frames.
On the wayAnd anything that speaks MCP.
Dodger runs an MCP server. Claude Code, Cursor, Codex, and your own software call it for the context behind any task: why it exists, what was decided, and how the feature is doing with real users. Keys are scoped to your workspace, and every caller sees only what it's allowed to.
Connect your stack.
Dodger works from the first connection. Each tool you add gives every answer, spec, and action more of what only your team knows.