@qmediat.io/gemini-code-context-mcp 1.5.1 → 1.5.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -5,6 +5,26 @@ All notable changes to `@qmediat.io/gemini-code-context-mcp` will be documented
5
5
  The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
6
6
  and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
7
7
 
8
+ ## [1.5.2] - 2026-04-22
9
+
10
+ ### Added
11
+
12
+ - **`mcpName` field in `package.json`** — set to `io.github.qmediat/gemini-code-context-mcp`. Required by the [Official MCP Registry](https://github.com/modelcontextprotocol/registry) for verified package publishing. The Registry's `mcp-publisher` tool reads this value to match the server's metadata against its npm package. GitHub-auth-scoped names must start with `io.github.<org>/`; this value locks the server to the `qmediat` GitHub org.
13
+
14
+ ### Changed
15
+
16
+ - **README comparison table and cost-model section now cite measured benchmark data** (2026-04-22, `vitejs/vite@main`'s `packages/vite/`, ~670 k tokens across 451 files, `gemini-pro-latest` with `thinkingLevel: LOW`):
17
+ - First query (cold, cache build): **~125 s** at **$0.60**
18
+ - Repeat query (cache hit): **~14 s** (mean of 15.6 and 13.5) at **$0.60**
19
+ - Inline baseline (`noCache: true`, files embedded in prompt): **~20 s** at **$2.35**
20
+ - **Speedup cache-vs-cold: ~8×.** Cost cache-vs-inline: **~4× cheaper (~75 % reduction)**. Daily saving at 20 queries/day on this workspace: **~$35/day per developer**.
21
+ - Previous README claims of "~30–45 s first call" and "~2–3 s follow-up" were aspirational / scoped to `latest-flash`; the new copy scopes latency to the thinking level and workspace size and links to reproducible ledger output via the `status` tool.
22
+ - **README wording on the `jamubc/gemini-mcp-tool` comparison row** softened from "Abandoned" to "Unmaintained on npm since 2025-07 (v1.1.4); last commit on `main` 2025-07-23; no maintainer reply on 2026 issues (#49/#62/#64)". Factual and specific rather than pejorative. Matches the project's "absorb orphan users, don't attack the original" positioning.
23
+
24
+ ### Rationale
25
+
26
+ This is a **docs + registry-metadata patch**. No runtime behaviour, no API surface, no new env vars, no breaking changes. Safe for any consumer to pick up on next `npx -y` cache refresh or `npm update -g`. The `mcpName` field is read by the Registry publisher only; runtime ignores it.
27
+
8
28
  ## [Unreleased]
9
29
 
10
30
  ### Fixed (v1.5.0 PR #24 round-2 review — 11 findings, applied before release)
package/README.md CHANGED
@@ -16,7 +16,7 @@ An MCP (Model Context Protocol) server that wraps Google's Gemini API with **per
16
16
 
17
17
  | | [jamubc/gemini-mcp-tool](https://github.com/jamubc/gemini-mcp-tool) | **`@qmediat.io/gemini-code-context-mcp`** |
18
18
  |---|---|---|
19
- | Maintenance | Abandoned (last release 2025-07) | Actively maintained |
19
+ | Maintenance | Unmaintained on npm since 2025-07 (v1.1.4); last commit on `main` 2025-07-23; no maintainer reply on 2026 issues (#49/#62/#64 at time of writing) | Actively maintained |
20
20
  | Default model | Hardcoded `gemini-2.5-pro` (main) — no runtime override | Dynamic `latest-pro` alias — resolves against your API key tier at startup |
21
21
  | Backend | Shells out to `gemini` CLI (subprocess per call) | Direct `@google/genai` SDK |
22
22
  | Repeat queries | No caching layer — each call re-tokenises referenced files | **Files API + Context Cache** — repeat queries reuse the indexed codebase; cached input tokens billed at ~25 % of the uncached rate |
@@ -48,7 +48,7 @@ npx @qmediat.io/gemini-code-context-mcp init
48
48
  # > Use gemini-code-context.ask to summarize this codebase
49
49
  ```
50
50
 
51
- First query: ~30–45 s (scan + upload + cache build). Every follow-up: ~2–3 s.
51
+ First query: ~45 s – 2 min depending on workspace size (scan + Files API upload + cache build). Every follow-up (cache hit): ~13–16 s on `latest-pro-thinking` with `thinkingLevel: LOW`, faster on `latest-flash`. Measured on `vitejs/vite@main`'s `packages/vite/` (~670 k tokens, 451 files): cold 125 s, warm ~14 s, $0.60 cached vs $2.35 inline per query (~8× faster, ~4× cheaper on cache hit). Thinking budget dominates warm latency — `HIGH` thinking adds 15–45 s per call on top of the cache-hit floor. Raw ledger reproducible via the `status` tool.
52
52
 
53
53
  See [`docs/getting-started.md`](./docs/getting-started.md) for a 3-minute walkthrough.
54
54
 
@@ -70,8 +70,8 @@ All tools accept an optional `workspace` path (defaults to `cwd`), `model` alias
70
70
  | | `ask` (eager) | `ask_agentic` |
71
71
  |---|---|---|
72
72
  | Workspace size | ≤ ~900 k tokens | any — model reads what it needs |
73
- | First query | 30–45 s (upload + cache build) | 5–15 s (no upload) |
74
- | Repeat queries | ~2–3 s (cache hit) | 10–30 s (new tool-use iterations per question) |
73
+ | First query | ~45 s – 2 min (upload + cache build; 125 s measured on 670 k-token workspace) | 5–15 s (no upload) |
74
+ | Repeat queries | ~13–16 s on pro-thinking LOW, faster on flash-tier (cache hit) | 10–30 s (new tool-use iterations per question) |
75
75
  | Per-call tokens | Full repo in cached input | Only files the model opens |
76
76
  | Best for | Many questions on same repo | One-off questions on huge repos, or repos with large generated files |
77
77
 
@@ -176,7 +176,7 @@ Full threat model + incident response: [`docs/security.md`](./docs/security.md).
176
176
 
177
177
  ## Cost model
178
178
 
179
- Projected savings vs uncached usage on a 500 k-token repo with 20 queries/day: up to **~60 % lower spend** with cache enabled, with repeat-query latency typically an order of magnitude lower than the first (uncached) call. Actual numbers depend on repo size, TTL, and query volume — we'll publish measured benchmarks after the first real-world deployments.
179
+ Measured 2026-04-22 on `vitejs/vite@main`'s `packages/vite/` (~670 k tokens, 451 files, Gemini 3.1 Pro, `thinkingLevel: LOW`): **$0.60 per cached query vs $2.35 per inline query — ~75 % cheaper on cache hit**, with cold-call latency ~125 s and warm-call latency ~14 s (~8× speedup). At 20 queries/day on this workspace that's **$35/day saved per developer** ($12 cached vs $47 inline). Actual numbers scale with workspace tokens × queries/day × thinking budget.
180
180
 
181
181
  Per-tool cost breakdown, free-tier guidance, and all the knobs: [`docs/cost-model.md`](./docs/cost-model.md).
182
182
 
package/package.json CHANGED
@@ -1,6 +1,7 @@
1
1
  {
2
2
  "name": "@qmediat.io/gemini-code-context-mcp",
3
- "version": "1.5.1",
3
+ "version": "1.5.2",
4
+ "mcpName": "io.github.qmediat/gemini-code-context-mcp",
4
5
  "description": "MCP server that gives Claude Code persistent memory of your codebase, backed by Gemini's 2M-token context. Includes eager `ask`/`code` for cached repo Q&A and `ask_agentic` for agentic file access on repos that exceed the model's input-token limit.",
5
6
  "keywords": [
6
7
  "mcp",