v0.8.0 - Deprecated, Gemini CLI retiring 2026-06-18

Deprecated and unmaintained

Google is transitioning Gemini CLI to Antigravity CLI on 2026-06-18. After that date, Gemini CLI stops serving Google AI Pro, Ultra, and free-tier (Gemini Code Assist for individuals) accounts. Enterprise Gemini Code Assist Standard/Enterprise licenses retain access.

Because this bridge wraps Gemini CLI as a subprocess, it stops working for individual users on the same date. This is the final release. The npm package will be marked deprecated and this repository archived shortly after publish.

Migration

• Antigravity CLI is Google's successor (Go-based, includes Agent Skills, Hooks, Subagents, plugins). Not open source.
• Sibling bridges for other terminal agents: claude-mcp-bridge, codex-mcp-bridge.

Changes in 0.8.0

• README banner and migration pointers.
• package.json description prefixed [DEPRECATED] for npmjs.com listings.
• No code changes; existing installs continue to function until 2026-06-18.

Full Changelog: v0.7.0...v0.8.0

v0.7.0

What's Changed

• fix(server.json): shorten description to meet MCP Registry 100-char limit by @hampsterx in #32
• refactor: adopt AGENTS.md as canonical agent instructions by @hampsterx in #33
• feat!: drop review and assess tools (ADR-001) by @hampsterx in #38

Full Changelog: v0.6.0...v0.7.0

v0.6.0

What's new

Added

• changeMode on query tool - opt-in flag that asks Gemini to emit structured **FILE: <path>:<start>-<end>** / ===OLD=== / ===NEW=== edit blocks. Parsed edits land on _meta.edits as a machine-applicable array, raw response stays in response. Legacy OLD: / NEW: markers still accepted per-block.
• Change-mode write guardrail - pre/post-spawn snapshot of tracked-file mtime+size and git status --porcelain. If Gemini mutates files during the spawn, _meta.appliedWrites is set and edits is omitted so callers can't re-apply half-applied state.
• MCP Registry manifest (server.json) - registers the bridge at registry.modelcontextprotocol.io. Downstream aggregators (PulseMCP, FindMCP, VS Code @mcp gallery) pick it up from there.
• mcpName in package.json (io.github.hampsterx/gemini-mcp-bridge) - required for Registry ownership verification on publish.
• query-change-mode smoke target - node scripts/smoke-test.mjs query-change-mode <repo> exercises the full change-mode path end-to-end.

Notes

• Change mode is text-only for v1 (image files rejected) and requires a git working directory (the guardrail needs git ls-files / git status).
• Change mode spawns without --approval-mode plan and without --yolo: plan mode refuses to emit edit blocks on CLI 0.38.0. The snapshot guardrail is the safety net against the default-mode write capability.

Full Changelog: v0.5.0...v0.6.0

v0.5.0

What's new

Added

• Response chunking for query, review, and search: oversized responses now return chunk 1 plus cache metadata (10-min TTL, max 50 entries, logical-boundary splits) instead of risking MCP client truncation.
• fetch-chunk tool: retrieves cached follow-up chunks via cacheKey and 1-based chunkIndex.
• assess change-kind classification: empty / code / mixed / non-code / generated. Doc-only diffs stay reviewable by default instead of being auto-dismissed as churn.
• assess guidance field: human-readable recommendation string tailored to the detected change kind and complexity.
• Spawn pacing controls: GEMINI_MIN_INVOCATION_GAP_MS enforces a minimum gap between CLI starts; GEMINI_SPAWN_JITTER_MAX_MS adds startup jitter.
• Structured capacity-failure metadata on deep reviews: explicit 429 / 503 / RESOURCE_EXHAUSTED responses surface as structured metadata so callers can decide whether to back off or downgrade depth.

Changed

• Server response formatting: prose-style tool responses flow through a shared formatter that appends metadata and applies chunking consistently.
• structured tool intentionally unchunked: preserves machine-consumable JSON output.
• Deep review no longer falls back internally on capacity errors: returns structured capacity metadata instead of retrying. Other tools and review depths keep existing fallback behavior.

Full Changelog: v0.4.0...v0.5.0

v0.4.0

What's new

Added

• assess tool - Zero-cost diff analysis pre-flight. Runs git diff --numstat locally (no CLI spawn, no model call) and returns diff stats, complexity classification (trivial/moderate/complex), and a recommended review depth with estimated wall-clock time. Use before review to set timeout expectations.
• depth parameter on review - Three tiers replace the binary quick/agentic split:
  • scan: diff-only, single-pass, no repo exploration. Constant 180s timeout.
  • focused: diff + CLI reads changed files in plan mode (no shell). Timeout scales 120s + 15s * files, capped at 300s (240s fallback).
  • deep (default): full agentic exploration with --yolo. Timeout scales 240s + 45s * files, capped at 1800s (600s fallback).
• scaleTimeoutForDepth, defaultTimeoutForDepth, resolveDepth - New exports for per-depth timeout and input resolution.

Changed

• BREAKING: ReviewResult.mode values - "agentic" | "quick" → "scan" | "focused" | "deep". Callers inspecting result.mode need to update. Pre-1.0 mechanically allows a minor bump; called out explicitly.
• Deep (previously agentic) timeout scaling - Per-file budget raised 30s → 45s and base raised 180s → 240s. A 10-file diff goes from 480s to 690s.
• Partial response hint - Timeout annotations now read consider depth: "scan" or narrow the base (was consider quick: true). Scan timeouts no longer annotated.
• Tool description on review - Rewritten to explain the three depths and new timeout formulas.

Deprecated

• quick parameter on review - Superseded by depth. Still works: quick: true → depth: "scan", quick: false → depth: "deep". depth wins when both are set.

Note

Focused mode containment is prompt-driven, not CLI-enforced. Plan mode removes shell access but does not scope read_file / grep_search / list_directory to changed files — Gemini could still read any non-gitignored file. Containment relies on the prompt instruction plus reduced tool surface.

Full Changelog: v0.3.0...v0.4.0

v0.2.5

Added

• MCP tool annotations: Tools declare readOnlyHint, destructiveHint, openWorldHint, and idempotentHint for informed client permission/caching decisions
• Execution metadata: All tool results include executionTime, timedOut, and resolvedModel fields
• Rich tool descriptions: Inline parameter docs and usage examples for clients that display them
• Progress heartbeats: Long-running operations emit MCP progress notifications
• MCP transport wiring tests and CI smoke step (npm run smoke:ci)

See CHANGELOG.md for full details.

v0.2.4

Added

• Stream-JSON output format: Switch from --output-format json to --output-format stream-json (NDJSON) for progressive capture. Timeouts now return partial responses instead of generic error messages.
• Smoke test script: npm run smoke / scripts/smoke-test.mjs for testing tool functions directly without restarting the MCP client. Supports all four tools with configurable workingDirectory.
• Resolved working directory: All tool results now include resolvedCwd showing the actual directory used after git root resolution and path validation.
• Latency budget documentation: Document ~16s CLI cold start, per-layer timing breakdown, and implications for timeout configuration.

Changed

• parseStreamJson() replaces parseGeminiOutput() as the primary parser, with automatic fallback to legacy JSON parsing for older CLI versions.
• tryParsePartial() extracts partial content from NDJSON on timeout, forwarding stderr for fallback parsing.
• Smoke test timeouts aligned with real CLI cold start times (60-120s).

Full Changelog: v0.2.3...v0.2.4

v0.2.3

Added

• Response length awareness: Optional maxResponseLength parameter (in words) on query, search, and review tools. Appends a soft length instruction to the prompt.
• Conciseness guidance in review and search prompt templates to reduce verbose output by default.
• GEMINI_DEFAULT_MODEL env var support for default model selection.
• Auto-retry with fallback model on quota exhaustion (GEMINI_FALLBACK_MODEL env var).

Changed

• maxResponseLength zod schemas enforce .int().positive().

npm: npm i gemini-mcp-bridge@0.2.3

v0.2.2

What's Changed

• feat: add search tool and image support for query by @hampsterx in #5
• feat: add structured output tool with JSON Schema validation by @hampsterx in #6

Full Changelog: v0.2.1...0.2.2

v0.2.1

Changed

• Review prompts extracted from inline template literals to standalone markdown files in prompts/, loaded at runtime via loadPrompt().

Fixed

• Prompt placeholder replacement no longer corrupts diffs containing {{word}} patterns (Handlebars, Go templates, etc.).
• loadPrompt now uses basename() to prevent path traversal.

Added

• 9 tests for prompt template loading and placeholder substitution.