● 89% vs 68% on 66 blind tasks with claude-sonnet-5-5

Your coding agent learned last year's APIs.

stalemap finds the dependencies that changed after your model's training cutoff and tells the agent what changed: before it starts, after every edit, and whenever it asks.

❯ npx stalemap

Install for your agent

Using an AI agent? Ask it to "install stalemap from stalemap.com". Instructions for agents: stalemap.com/install.md

stale (what the model learned) → installed

// ai 7 · AI SDK
- streamText({ maxSteps: 5 })
+ streamText({ stopWhen: isStepCount(5) })

// zod 4
- z.string({ required_error: "…" })
+ z.string({ error: "…" })

// next 16
- middleware.ts → middleware()
+ proxy.ts      → proxy()

// mongoose 9
- pre("save", function (next) {…})
+ pre("save", async function () {…})

Three ways the agent finds out

stalemap compares each dependency's version the model knew with the one you've installed: type declarations, release notes and upgrade guides. It then puts what changed in front of the agent, ranked by what your code actually uses.

At session start

Notes

A short, ranked block in CLAUDE.md or AGENTS.md: removed and renamed APIs, new required options, changed defaults, each with its replacement.

After each edit

Guard

Runs your project's own tsc. Errors a changed dependency explains come with the fix, from the notes or the matching release-note or upgrade-guide section. Moved config keys and renamed files are flagged even when they still compile.

On demand

search_changes

An MCP tool the agent asks in its own words, such as "resolve alias in next.config", to get the matching changes with their sources and code.

Measured blind, before launch

66 tasks drawn by seed from every breaking, deprecation and behavior change that 33 popular npm packages shipped between July and October 2026, after the training cutoff of the models tested. The tasks were written by an author who never saw stalemap. The build was frozen and the analysis pre-registered before any result existed. Hidden tests decide pass or fail.

+20.7 ptsvs no help, claude-sonnet-5-5
95% interval [13.6, 28.3]

+16.4 ptsvs a plain tsc hook after each edit
98.3% interval [7.6, 26.0]

27% → 11%first edits that used a stale API
claude-sonnet-5-5

claude-sonnet-5-5: no help 68%, plain tsc hook 73%, stalemap 89%, one-line hint 98%. claude-sonnet-4-5: 46%, 53%, 64%, 78%. — Share of sessions that passed, both modes pooled, 396 sessions per bar. "One-line hint" is a hand-written note naming the exact change, the ceiling for any note.

Pre-registered comparison	Model	Difference	Interval	Verdict
stalemap vs no help (primary)	claude-sonnet-5-5	+20.7 pts	[13.6, 28.3]	win
stalemap vs plain `tsc` hook	claude-sonnet-5-5	+16.4 pts	[7.6, 26.0]	win
stalemap vs no help	claude-sonnet-4-5	+17.9 pts	[7.6, 28.8]	win
stalemap vs plain `tsc` hook	claude-sonnet-4-5	+10.6 pts	[0.3, 21.7]	win

Two-level bootstrap over tasks and runs; secondary comparisons at 98.3% (Bonferroni over three). Median cost per session with stalemap was $0.050 vs $0.041 with no help on claude-sonnet-5-5. Everything is in the repository: protocol, tasks and hidden tests, blind review, full results.

Install

JavaScript and TypeScript projects. stalemap reads your package.json, lockfile and imports locally; your source code never leaves your machine.

Install stalemap in this project: follow https://stalemap.com/install.md

Paste this into Claude Code, Cursor, Codex or OpenCode. The agent detects itself and its model, runs stalemap setup, verifies it, and tells you what it changed.

/plugin marketplace add Umair/stalemap /plugin install stalemap@stalemap

Adds the notes at session start, the guard after each edit, and the search_changes MCP tool.

npx stalemap setup --agent cursor # or codex, opencode

Writes the notes, registers the MCP server and installs the guard as the agent's post-edit hook, as project files you can commit. --dry-run shows the plan.

npx stalemap # preview what changed npx stalemap sync # write the notes into AGENTS.md or CLAUDE.md npx stalemap check # for CI: fail when the notes are missing or stale

{ "mcpServers": { "stalemap": { "command": "npx", "args": ["-y", "stalemap@latest", "mcp"] } } }

- uses: Umair/stalemap@main with: model: claude-sonnet-5-5 # mode: comment # or comment on PRs that change dependencies

Questions

Which model's cutoff?

The one your agent uses: detected from Claude Code, Codex, OpenCode or Gemini CLI settings, or passed with --model. Cutoffs come from models.dev.

What leaves my machine?

Package names and versions go to the npm registry and GitHub to fetch type declarations and release notes. Your code is scanned locally and never uploaded.

Do current models still need this?

Yes, for anything released after their cutoff. In v4, claude-sonnet-5-5 failed a third of the tasks on its own; with stalemap it passed 89%.

Isn't a type-check hook enough?

It catches what fails to compile, not what to write instead, and not changed defaults. In v4 a plain tsc hook got 73%; stalemap's guard runs the same check and adds the fix.

What does it cost?

stalemap is free and MIT-licensed. With stalemap, agent sessions cost about 20% more on claude-sonnet-5-5 (median $0.050 vs $0.041 per task).

What doesn't it do?

Other languages, monorepo workspaces, and behavior changes that no release note or guide describes. The limits are listed in the README.