@jarkkojs/readseek 0.7.7 → 0.7.9

package/README.md

@@ -1,25 +1,25 @@
 # readseek
 
-`readseek` is a structural source and PDF reader for scripts, editors, and coding agents.
-It emits compact JSON with stable line/hash anchors, structural symbol maps, parse
-diagnostics, AST search matches, references, and rename plans.
+`readseek` reads source files and PDFs for scripts, editors, and coding agents.
+It returns compact JSON with stable line/hash anchors, symbol maps, parse diagnostics,
+AST matches, references, and rename plans.
 
 ## Installation
 
-Install the npm wrapper and a prebuilt binary:
+Install the npm wrapper with a prebuilt binary:
 
 ```sh
 npm install -g @jarkkojs/readseek
 ```
 
-Install the native program from crates.io:
+Or install the native program from crates.io:
 
 ```sh
 cargo install readseek
 ```
 
 Prebuilt binaries are available for macOS ARM64; Linux ARM64 and x64; and Windows
-x64. The Linux binaries are static glibc PIE executables.
+x64. The Linux binaries are statically linked with musl.
 
 To build and install from source:
 
@@ -27,18 +27,12 @@ To build and install from source:
 make install
 ```
 
-Source builds require CMake, Clang/libclang, and a C++ compiler because image
-inference uses `llama-cpp-2`.
-
-GPU acceleration can be enabled at build time with the `metal`, `opencl`, `rocm`,
-or `vulkan` Cargo feature. CPU builds can enable `openmp` for OpenMP parallelism.
-
 ## Plugins
 
 ### Pi extension
 
-The bundled [pi-readseek extension](packages/pi-readseek/README.md) exposes
-ReadSeek's anchored file and structural-code tools in Pi:
+The bundled [pi-readseek extension](packages/pi-readseek/README.md) adds ReadSeek's
+anchored file and structural-code tools to Pi:
 
 ```sh
 pi install npm:pi-readseek
@@ -46,10 +40,8 @@ pi install npm:pi-readseek
 
 ### OpenCode plugin
 
-The [opencode-readseek plugin](packages/opencode-readseek/README.md) provides the
-same anchored and structural tools in OpenCode.
-
-Add the plugin to `opencode.json`:
+The [opencode-readseek plugin](packages/opencode-readseek/README.md) adds the same
+tools to OpenCode. Add it to `opencode.json`:
 
 ```json
 {
@@ -59,13 +51,13 @@ Add the plugin to `opencode.json`:
 
 ### Vim plugin
 
-The bundled `readseek.vim` plugin is a Vim9-only frontend with anchored range and
-symbol reads, file detection, structural navigation, parser diagnostics, search,
-references, and rename operations. It requires Vim9 with `+job`, `+channel`,
-`+popupwin`, and `+textprop`; Neovim is not supported. Install it directly with a
-Vim plugin manager, for example `Plug 'jarkkojs/readseek'`. Run
-`:ReadSeekInstall` to install the matching prebuilt ReadSeek release binary;
-installation is explicit by default.
+The bundled `readseek.vim` plugin provides anchored reads, structural navigation,
+parse diagnostics, search, references, and renames. It requires Vim9 with `+job`,
+`+channel`, `+popupwin`, and `+textprop`; Neovim is not supported.
+
+Install it with a Vim plugin manager, such as `Plug 'jarkkojs/readseek'`, then run
+`:ReadSeekInstall` to install the matching prebuilt binary. Downloads are explicit
+by default.
 
 ## Common commands
 
@@ -84,16 +76,15 @@ readseek search src 'fn $NAME() { $$$BODY }' --language rust
 readseek rename src/main.rs --line 42 --column 8 --to renamed
 ```
 
-To write JSON output to a file instead of stdout, place the global option before
-the command:
+Global options must precede the command. For example, write JSON to a file with:
 
 ```sh
 readseek --output result.json detect src/main.rs
 ```
 
-Use a `stdin:` target prefix with `detect`, `read`, `map`, `check`, `symbol`,
-and `identify` to analyze unsaved editor buffers while still providing a path
-for language detection and a cursor address:
+Prefix a target with `stdin:` to analyze an unsaved buffer while retaining a path
+for language detection and cursor addressing. This works with `detect`, `read`,
+`map`, `check`, `symbol`, and `identify`:
 
 ```sh
 printf '%s\n' 'fn main() {}' | readseek identify stdin:scratch.rs:1 --column 4
@@ -101,8 +92,8 @@ printf '%s\n' 'fn main() {}' | readseek identify stdin:scratch.rs:1 --column 4
 
 ## Images and PDFs
 
-`detect` reports image metadata and PDF page counts. `read` returns bounded
-base64 images by default; use `--image` for a local analysis mode:
+`detect` reports image metadata and PDF page counts. `read` returns bounded base64
+images by default; select local analysis with `--image`:
 
 ```sh
 readseek read photo.jpg                   # default: bounded base64 image
@@ -112,55 +103,46 @@ readseek read photo.jpg --image ocr       # extracted text
 readseek read photo.jpg --image all       # caption, objects, and OCR in one pass
 ```
 
-Inference defaults to the `balanced` profile. `fast` uses fewer visual tokens for
-lower latency, while `accurate` retains more visual detail for dense OCR and small
-objects:
+Vision analysis uses the `balanced` profile by default. Choose `fast` for lower
+latency or `accurate` for dense OCR and small objects:
 
 ```sh
 readseek read photo.jpg --image caption --vision-profile fast
 readseek read scan.png --image ocr --vision-profile accurate
 ```
 
-Use `--vision-diagnostics` to emit cache status, backend devices, token counts,
-per-stage timings, throughput, and peak RSS as JSON on stderr. For repeatable warm
-measurements, `--vision-benchmark N` runs one warmup followed by `N` measured
-iterations (currently image files only):
+`--vision-diagnostics` writes cache status, devices, token counts, timings,
+throughput, and peak RSS as JSON to stderr. `--vision-benchmark N` runs one warmup
+and `N` measured iterations for an image file:
 
 ```sh
 readseek read fixture.png --image all --vision-benchmark 5 \
   >result.json 2>benchmark.json
 ```
 
-CPU tuning overrides are available through `READSEEK_VISION_THREADS`,
-`READSEEK_VISION_BATCH`, and `READSEEK_VISION_UBATCH`. Values must be positive,
-and the micro-batch cannot exceed the logical batch. Compare overrides on a fixed
-representative image corpus; larger batches can improve prefill while increasing
-peak memory.
+Set `READSEEK_VISION_THREADS` to a positive integer to override Rayon's worker
+count. More workers may improve throughput at the cost of CPU and memory.
 
-PDF reads return page-tagged Markdown and page-associated embedded images. Use
-`--page` to select one page; the same image mode applies to each embedded image.
-After initializing `.readseek/`, `view` creates or reuses a persistent structural
-PDF index and returns an overview that can be narrowed by page, node, kind, or
-depth. Line/hash suffixes, `--end`, `--limit`, and `--language` do not apply to
-visual files.
+PDF reads return page-tagged Markdown and embedded images. Use `--page` to select a
+page; `--image` applies to each embedded image. After `readseek init`, `view` creates
+or reuses a structural PDF index that can be narrowed by page, node, kind, or depth.
+Line/hash suffixes, `--end`, `--limit`, and `--language` do not apply to visual files.
 
-The Qwen3-VL GGUF language model and Q8_0 multimodal projector download lazily
-into the user cache, are checksum-verified once, and then use a metadata-bound
-verification marker on later starts. Inference uses the backends enabled in the
-build; CPU-only captioning can still take substantial time.
+The Qwen3-VL-2B Q8_0 model and multimodal projector download lazily to the user
+cache and are checksum-verified. Inference uses ReadSeek's built-in CPU engine and
+can be slow.
 
 ## Cache
 
-`readseek init [path]` creates and populates `.readseek/maps/` and
-`.readseek/def-index/`. Map-dependent commands update entries on demand and
-discover `.readseek/` by walking up from the target path, or use the directory
-passed by `--readseek-dir`. `view` creates PDF structure indexes and extracted
-assets under `.readseek/documents/` on demand. Image analysis caches results
-under the `.readseek/` found from the current working directory. Cache entries are profile-specific.
+`readseek init [path]` creates `.readseek/maps/` and `.readseek/def-index/`.
+Map-dependent commands update them on demand and find `.readseek/` by walking up
+from the target; `--readseek-dir` selects one explicitly. PDF indexes and extracted
+assets live in `.readseek/documents/`, while image analysis results are cached under
+`.readseek/vision/`. Image cache entries are profile-specific.
 
 ## Documentation
 
-The manual page provides the full CLI reference:
+For the complete CLI reference:
 
 ```sh
 man ./man/man1/readseek.1
@@ -170,10 +152,7 @@ Pass `--help` to any command for command-specific usage.
 
 ## Licensing
 
-The native `readseek` crate is licensed under `LGPL-2.1-or-later`. The
-`@jarkkojs/readseek` wrapper and platform packages declare
-`Apache-2.0 AND LGPL-2.1-or-later`; the bundled plugins use Apache-2.0 or MIT as
-noted in their package directories.
-
-The downloaded `Qwen/Qwen3-VL-2B-Instruct-GGUF` model is licensed under
-`Apache-2.0`.
+The native `readseek` crate is licensed under `LGPL-2.1-or-later`. The npm wrapper
+and platform packages declare `Apache-2.0 AND LGPL-2.1-or-later`; plugin licenses
+are listed in their package directories. The downloaded Qwen model is licensed
+under `Apache-2.0`.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jarkkojs/readseek",
-  "version": "0.7.7",
+  "version": "0.7.9",
   "description": "A structural read command",
   "license": "Apache-2.0 AND LGPL-2.1-or-later",
   "homepage": "https://github.com/jarkkojs/readseek#readme",
@@ -21,10 +21,10 @@
     "LICENSE-LGPL-2.1"
   ],
   "optionalDependencies": {
-    "@jarkkojs/readseek-darwin-arm64": "0.7.7",
-    "@jarkkojs/readseek-linux-arm64": "0.7.7",
-    "@jarkkojs/readseek-linux-x64": "0.7.7",
-    "@jarkkojs/readseek-win32-x64": "0.7.7"
+    "@jarkkojs/readseek-darwin-arm64": "0.7.9",
+    "@jarkkojs/readseek-linux-arm64": "0.7.9",
+    "@jarkkojs/readseek-linux-x64": "0.7.9",
+    "@jarkkojs/readseek-win32-x64": "0.7.9"
   },
   "publishConfig": {
     "access": "public"