Junie

You will like Junie if IntelliJ is home: it runs terminal commands and tests as part of a plan and can drive the IDE's own debugger, so when a test fails it steps through the failure rather than guessing from the trace. That is the daily trait, and no other agent here has it.

The catch is the meter: Ultimate is $30 for 35 credits, and the docs do not say how fast an agent run eats them. Pick it for a JetBrains shop where the debugger integration pays for itself. Pick Claude Code otherwise, where the bill is at least legible.

The meter is the worst thing. The licensing page prices chat generations, not agent runs, so a user learns the burn rate by running out mid-task. Brave mode executes without confirmation and there is no Docker sandbox behind it, so the unattended path runs on the developer's machine.

The consequence: expect the first month to be a measurement exercise, and keep Brave mode off until you have seen what the plan step proposes. A published credits-per-task figure would change this verdict. What it does right: the CLI runs on your own keys or local Ollama, no meter at all.

SWE-rebench, June 2026, 61.6 percent resolved and 72.7 percent pass@5, is self-reported on a set that draws fresh tasks each cycle, so results move run to run by the organizer's own account, and pass@5 is a five-attempt figure that should not be read beside single-attempt scores. The earlier 53.6 percent on SWE-bench Verified was a single run from January 2025.

The interesting part is verification: agentic debugging sets breakpoints and steps through code in the IDE's debugger, a richer signal than test output. The observation: the benchmark measures the model, the debugger measures the harness, and only the second is Junie's.

JetBrains is a profitable IDE vendor, and Junie is bundled into its AI plans rather than sold alone, pricing power by attachment: the price rises with the IDE renewal, and nobody churns an IDE over an agent. That is the safest revenue shape on this board and the least ambitious.

The risk: ACP lets other vendors' agents into the IDE, so Junie must earn its place in its own house, and a bundled feature that loses to a guest becomes a checkbox. Likely outcome is neither acquisition nor pivot, just a long life as a line in a renewal email. Position: long, no acquirer needed.

The demo is a plan that steps through the debugger. Procurement: AI Enterprise is $60 per user, so $3,600 a month for sixty, with credits pooled through JetBrains Central and per-user spending limits an admin sets, which is the control that stops one enthusiast from spending the team's month.

The unknown is credit consumption per agent task, which the table does not list, so the pool size is a guess until measured. CI fit exists through the CLI's headless mode. Onboarding is a plugin the IDE already suggests. Approved with conditions: a two-team pilot to measure burn rate before the pool is sized.

Two products under one name. The plugin is closed and metered in JetBrains AI credits, three on the free tier, which is a demo allowance. The CLI is the part I use: brew install jetbrains/junie/junie, bring-your-own-key for OpenAI, Anthropic, Google, xAI and OpenRouter, and Ollama or LM Studio endpoints, so it runs against my box with no credits.

MCP is a mcpServers block with command and args or url and headers, so my servers carry over. Nothing to fork, since both halves are closed. Grudging respect for a JetBrains product that speaks to Ollama without asking for a plan.