tfmodsearch 0.4.0__tar.gz

tfmodsearch-0.4.0/.agents/plugins/marketplace.json

@@ -0,0 +1,14 @@
+{
+  "name": "tfmodsearch",
+  "description": "TFModSearch — current Terraform AWS module documentation search for AI agents",
+  "plugins": [
+    {
+      "name": "tfmod-search",
+      "source": {
+        "source": "local",
+        "path": "./plugins/tfmod-search"
+      },
+      "description": "Search current terraform-aws-modules documentation before writing Terraform: MCP server (hybrid keyword + BM25 + semantic search) plus workflow skills"
+    }
+  ]
+}

tfmodsearch-0.4.0/.claude-plugin/marketplace.json

@@ -0,0 +1,17 @@
+{
+  "name": "tfmodsearch",
+  "owner": {
+    "name": "Alexander Makeev",
+    "url": "https://github.com/SantyagoSeaman"
+  },
+  "description": "TFModSearch — current Terraform AWS module documentation search for AI agents",
+  "plugins": [
+    {
+      "name": "tfmod-search",
+      "source": "./plugins/tfmod-search",
+      "description": "Search current terraform-aws-modules documentation before writing Terraform: MCP server (hybrid keyword + BM25 + semantic search) plus workflow skills",
+      "displayName": "TFModSearch",
+      "keywords": ["terraform", "aws", "mcp", "search", "infrastructure"]
+    }
+  ]
+}

tfmodsearch-0.4.0/.github/workflows/publish.yml

@@ -0,0 +1,47 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags:
+      - "v*"
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v5
+
+      - name: Build distributions
+        run: uv build
+
+      - name: Check that tag matches project version
+        run: |
+          VERSION=$(uv run --no-project python -c "import tomllib; print(tomllib.load(open('pyproject.toml','rb'))['project']['version'])")
+          TAG="${GITHUB_REF_NAME#v}"
+          if [ "$VERSION" != "$TAG" ]; then
+            echo "Tag $GITHUB_REF_NAME does not match pyproject version $VERSION" >&2
+            exit 1
+          fi
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+  publish:
+    needs: build
+    runs-on: ubuntu-latest
+    environment: pypi
+    permissions:
+      id-token: write
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1

tfmodsearch-0.4.0/.gitignore

@@ -0,0 +1,107 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Virtual environments
+.venv
+venv/
+ENV/
+env/
+.env
+
+# IDEs and editors
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# NLTK data
+nltk_data/
+
+# Log files
+logs/
+*.log
+startup.log
+mcp_server.log
+
+# Temporary files
+*.tmp
+.temp/
+temp/
+
+# User-specific Claude Code settings
+.claude/
+CLAUDE.md
+/.serena/
+
+# Internal maintenance docs (not part of the published package)
+modules/UPDATE_PROMPTS.md

tfmodsearch-0.4.0/.pre-commit-config.yaml

@@ -0,0 +1,23 @@
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.5.0
+    hooks:
+      - id: trailing-whitespace
+      - id: end-of-file-fixer
+      - id: check-yaml
+      - id: check-merge-conflict
+      - id: debug-statements
+
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.4.4
+    hooks:
+      - id: ruff
+        args: [--fix]
+      - id: ruff-format
+
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: v1.10.0
+    hooks:
+      - id: mypy
+        additional_dependencies: [types-PyYAML]
+        args: [--ignore-missing-imports]

tfmodsearch-0.4.0/CHANGELOG.md

@@ -0,0 +1,299 @@
+# Changelog
+
+## [0.4.0] - 2026-07-11
+
+[0.4.0]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.4.0
+
+The distribution release: the project is renamed to **tfmodsearch**, published to PyPI, and ships plugins for Claude Code and Codex CLI that bundle the MCP server with agent workflow skills.
+
+### Changed
+
+- **Project renamed to `tfmodsearch`** (repository, PyPI package, and primary command). The GitHub repository moved to `SantyagoSeaman/tfmodsearch`; old `aws-tf-modules-mcp-server` URLs redirect automatically.
+- **New console commands**: `tfmodsearch` (MCP server) and `tfmodsearch-cli` (indexing/search CLI). The previous `aws-tf-modules-mcp-server` and `tfmod-search-cli` commands are kept as deprecated aliases.
+- **Trimmed the published wheel** from ~11 MB to ~1.6 MB: it now ships only the production index (`model/tfmod_bge_base_index.pkl`) and the curated `modules/terraform-aws-modules/` docs, excluding the legacy gte-small index, work-in-progress docs, and internal templates.
+- **Rewrote the MCP server `instructions`**: corrected the stale catalog size, fixed grammar, and added explicit workflow steering ("search before writing Terraform; use documented variable names, not memorized ones"). The advertised server version now derives from the installed package metadata instead of a hardcoded string.
+
+### Added
+
+- **PyPI publication** with Trusted Publishing: a `publish.yml` GitHub Actions workflow builds and publishes to PyPI on version tags via OIDC (no API tokens), with a tag-vs-version consistency check.
+- **Claude Code plugin** (`/plugin marketplace add SantyagoSeaman/tfmodsearch` → `/plugin install tfmod-search@tfmodsearch`): auto-configures the MCP server via `uvx tfmodsearch` and ships two skills.
+- **Codex CLI plugin** with the same two-command install, plus an `agents/openai.yaml` declaring the MCP tool dependency for implicit skill invocation.
+- **Skills** (portable SKILL.md format, shared by both plugins):
+  - `aws-terraform-modules` — auto-invoked when writing/reviewing Terraform for AWS: search first, write from current docs, pin module versions, prefer community modules.
+  - `tf-module` — user-invoked lookup (`/tf-module s3 bucket with lifecycle rules`) returning a ready-to-paste module block with current variable names.
+- **`--warmup` flag** for the MCP server: pre-downloads the embedding model (~220 MB), loads the index, runs a test query, and exits — keeps first server startup within MCP client timeouts.
+
+### Fixed
+
+- **`get_module` no longer guesses**: unknown module names previously fell through to hybrid search and silently returned the highest-scoring module's documentation (a hallucinated name could return the lambda docs). Name lookup is now exact-match first, then unique-substring; unknown or ambiguous names return an error listing the closest matches and directing the caller to `search_modules`.
+
+### Tests
+
+- **New end-to-end suite** (`tests/e2e/`, 20 tests): full MCP stdio protocol sessions against a spawned server process (initialize handshake, tool discovery, all three tools, security rejections, `--warmup`), wheel payload/metadata/entry-point verification, a `uvx` packaged-server smoke test from a foreign directory, plugin manifest and skill contract checks for both the Claude Code and Codex layouts, and a live `claude plugin` install into an isolated config. Full suite: 259 tests passing.
+
+### Documentation
+
+- README: new plugin-first installation section, Codex CLI integration guide (with `startup_timeout_sec` note), `claude mcp add` one-liner, AGENTS.md/CLAUDE.md steering snippet, and a PyPI badge. All install configs now use the published `tfmodsearch` package instead of `git+https` URLs.
+
+## [0.3.0] - 2026-07-11
+
+[0.3.0]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.3.0
+
+This release refreshes the entire module catalog against the latest upstream `terraform-aws-modules` versions, rebuilds the pre-built search index, and overhauls the README. It also rolls up the previously unreleased 0.2.1 changes.
+
+### Changed
+
+- **Full documentation refresh for all 54 modules**: Every document in `modules/terraform-aws-modules/` was regenerated from the current Terraform Registry / GitHub source. Notable version and API updates include:
+  - `security-group` → 6.0.0 — major rearchitecture: per-rule `aws_vpc_security_group_ingress_rule`/`egress_rule` resources replace inline rule blocks
+  - `eks` → 21.24.0 — post-v21 variable names (`name`, `kubernetes_version`, `addons`, `compute_config`); no default add-ons are bootstrapped
+  - `lambda` → 8.8.1, `s3-bucket` → 5.14.1 (new `vectors` submodule), `iam` → 6.6.1, `rds` → 7.2.0, `ec2-instance` → 6.4.0, `alb` → 10.5.0, `apigateway-v2` → 6.1.0, `vpc` → 6.6.1, and others
+  - Corrected stale or incorrect variable names, defaults, and examples that could have produced invalid Terraform (e.g. EKS pre-v21 variables, Lambda Function URL configuration, RDS `create_db_subnet_group`/`storage_type` defaults, IAM `iam-role` trust-policy variables)
+  - Re-curated keyword lists to the 10–20 most relevant terms per module
+- **Rebuilt the pre-built search index** (`model/tfmod_bge_base_index.pkl`) to reflect the refreshed documentation.
+
+### Added
+
+- Canonical service-name keywords to improve retrieval for common full-name queries: `application-load-balancer` (alb), `relational-database` (rds), `simple-notification-service` (sns), `identity`/`access-management` (iam), and `vpn` (vpn-gateway).
+
+### Fixed
+
+- **README accuracy**: corrected the pre-built index filename (`tfmod_bge_base_index.pkl`, previously shown as `tfmod_index.pkl` in several places), the catalog module count (added the missing `emr` module — now a consistent 54), the local-install commands (`pip install -e .` / `uv pip install -e .`), stale test counts, and the `get_module` tool terminology.
+
+### Documentation
+
+- Added a "Why TFModSearch?" value-proposition section, an end-to-end `search_modules` → `get_module` workflow example, an explicit note that the pre-built index and module docs ship inside the installable package, and clarification of the default search weights.
+
+### Tests
+
+- Relaxed `test_module_searchable_by_keyword` to assert the target module appears in the **top-3** results — matching `search_modules`'s documented top-3 return contract and the existing natural-language test — rather than strictly ranking #1. A few semantically adjacent module pairs (e.g. security-group vs network-firewall for "firewall rules") legitimately rank a sibling first while keeping the target in the top-3. Full suite: 239 tests passing.
+
+## [0.2.1] - 2026-01-26
+
+[0.2.1]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.2.1
+
+### Added
+
+- **Configurable Log Level**: New `log_level` option in `config.yaml` (default: INFO)
+  - Configurable via config file or CLI `--log_level` argument
+  - New `ConfigLoader.load_log_level()` method with standard precedence (CLI > YAML > default)
+
+- **Quick Install via uvx**: Added simplified installation method using `uvx`
+  - No repository cloning required
+  - Single config line: `"command": "uvx", "args": ["git+https://github.com/..."]`
+  - Added uv installation instructions
+
+### Fixed
+
+- **Silenced Noisy Third-Party Loggers**: FastMCP internal loggers now set to WARNING
+  - `fakeredis`, `docket`, `asyncio` no longer spam DEBUG output
+  - Server startup is now clean and quiet
+
+### Documentation
+
+- Updated README with `uvx` as recommended installation method
+- Added troubleshooting note for full path to `uvx` when not in PATH
+- Updated CLAUDE.md with new `log_level` config option
+
+## [0.2.0] - 2026-01-25
+
+[0.2.0]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.2.0
+
+### Changed
+
+- **Embedding Model Upgrade**: Switched default model from `thenlper/gte-small` to `BAAI/bge-base-en-v1.5`
+  - Better retrieval performance with 768-dimensional embeddings (vs 384)
+  - Improved similarity distribution (addresses narrow clustering issue)
+  - BGE v1.5 specifically optimized for retrieval tasks with hard negatives training
+  - Model size: ~220MB (vs ~67MB for gte-small)
+
+- **Optional Query Instruction Support**: Added configurable query instruction prefix for BGE models
+  - New `query_instruction` parameter in `compute_scores()` function
+  - Configurable via `config.yaml` or CLI `--query-instruction` argument
+  - Uses modern sentence-transformers `prompt` parameter in `model.encode()`
+  - Disabled by default (BGE v1.5 works well without instruction)
+  - Instruction: `"Represent this sentence for searching relevant passages: "`
+
+- **Default Index Path**: Changed from `tfmod_gte_small_index.pkl` to `tfmod_bge_base_index.pkl`
+  - All references updated across codebase, CLI, and tests
+  - Pre-built index now included at `model/tfmod_bge_base_index.pkl`
+
+- **Module List Description**: `modules_list` tool now uses `extract_purpose()` for cleaner descriptions
+  - Extracts Purpose field from Module Information section
+  - More concise and relevant module descriptions in catalog listing
+
+- **Dependency Updates**:
+  - Updated `numpy` from `>=2.3.3` to `>=2.4.1`
+  - Updated `sentence-transformers` from `>=5.1.1` to `>=5.2.0`
+
+### Added
+
+- **Integration Guides**: Comprehensive setup instructions in README
+  - Claude Code CLI integration with `claude mcp add` command
+  - Claude Desktop configuration for macOS
+  - GitHub Copilot integration for VS Code (requires VS Code 1.99+)
+  - Step-by-step setup and troubleshooting commands
+
+- **New Search Library Functions**:
+  - `extract_purpose(text, max_length)`: Extract Purpose field from Module Information section
+  - `DEFAULT_MODEL_NAME` constant: Default embedding model identifier
+  - `BGE_QUERY_INSTRUCTION` constant: Optional query prefix for BGE models
+
+- **Configuration Loader**: New `ConfigLoader.load_query_instruction()` method
+  - Supports precedence: CLI > YAML > default (None)
+  - Proper logging for configuration source
+
+- **Build Configuration**: Added hatch wheel build settings in `pyproject.toml`
+  - Configured `tool.hatch.build.targets.wheel` for proper package distribution
+  - Source mapping from `src/` directory
+
+- **Model Comparison Tests**: New test file `tests/integration/test_model_comparison.py`
+  - Compares embedding model performance (gte-small vs bge-base-en-v1.5)
+  - Tests across different query types (1-word to long natural language)
+  - Timing measurements and success rate analysis
+
+### Fixed
+
+- **ServerState Initialization**: Added `query_instruction` parameter support
+  - Properly passed through `ServerStateManager.initialize()`
+  - Included in state logging output
+
+### Documentation
+
+- **Module Documentation Updates**: Refreshed all 54 Terraform AWS module docs
+  - Updated module versions and feature descriptions
+  - Improved formatting and consistency across modules
+  - Enhanced best practices and use case sections
+
+- **README Improvements**:
+  - Reorganized Quick Start section with integration-specific guides
+  - Updated all code examples to use new model and index paths
+  - Added GitHub Copilot and VS Code MCP documentation links
+
+- **Configuration Comments**: Enhanced `config.yaml` with detailed comments
+  - Documented query instruction usage and BGE model notes
+  - Added model configuration guidance
+
+### Notes
+
+- **Index Rebuild Required**: After upgrading, rebuild the search index with the new model
+- **Index Size**: Increased from ~4.5MB to ~9MB due to larger embedding dimensions (768 vs 384)
+- **Performance**: First query ~2-3s (model loading), subsequent queries ~0.02s
+- **Breaking Change**: Default index path changed - update any hardcoded paths
+
+## [0.1.2] - 2025-10-13
+
+[0.1.2]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.1.2
+
+### Changed
+
+- **Parser Architecture Refactoring**: Complete rewrite of document parsing system
+  - Removed YAML frontmatter dependency in favor of pure regex-based parsing
+  - Created `ModuleDocumentParser` class with two-strategy parsing approach:
+    1. "## Module Information" section parsing (preferred)
+    2. Inline keywords search (fallback)
+  - Fixed module name parsing to strip backticks from Markdown formatting
+  - Removed YAML frontmatter blocks from 46 module documentation files
+  - Simplified codebase by eliminating `yaml` library dependency
+
+- **Search Algorithm Improvements**: Enhanced hybrid search for better ranking accuracy
+  - **Minmax Normalization for Semantic Search**: Added min-max normalization to cosine similarities
+    - Spreads small embedding differences (0.89-0.92) into full 0-1 range for better discrimination
+    - Addresses limitation of `thenlper/gte-small` model's narrow similarity clustering
+    - All scoring components now equally normalized (keywords, BM25, semantic, exact match)
+  - **Adaptive Query Scaling**: Implemented logarithmic scaling for BM25 and semantic scores
+    - Formula: `log(len(matched_keywords) + 1, base=3)` dampens scores for short queries
+    - Reduces BM25 noise on 1-2 word keyword queries where lexical matching is unreliable
+    - Improves ranking for precise keyword searches (e.g., "tgw", "firewall", "ssh-keys")
+
+- **Search Index**: Rebuilt index with refactored parser
+  - Index size optimized from 4.58MB → 4.51MB (1.5% reduction)
+  - All 54 modules now use consistent text-based parsing
+  - Maintains 1,529 unique keywords across corpus
+
+### Added
+
+- **Comprehensive Test Suite**: Created extensive integration tests for search quality
+  - New file: `tests/integration/test_all_modules_searchable.py` (387 lines, 169 tests)
+  - **ModuleTestCase Dataclass**: Explicit document→query triple linkage
+    - Each module linked to: keyword query, exact name query, natural language query
+    - Clear traceability for test failures with document paths and query types
+  - **Parametrized Tests**: 162 tests (54 modules × 3 query types)
+    - `test_module_searchable_by_keyword`: Verifies keyword-based search
+    - `test_module_searchable_by_name`: Validates exact module name matching
+    - `test_module_searchable_by_natural_language`: Tests semantic search quality
+  - **Quality Validation Tests**: 7 additional tests
+    - Module completeness (keywords, names, content)
+    - Data integrity (no duplicates, sufficient keyword vocabulary)
+    - Search quality metrics (average keyword count, vocabulary size)
+  - **NLTK Initialization**: Created `tests/conftest.py` for pytest setup
+    - Automatic NLTK data initialization before test session
+    - Eliminates `punkt_tab` resource errors
+  - **Test Results**: 160/169 passing (94.7% pass rate)
+    - 9 failures due to expected search quality variations (keyword collisions, semantic ambiguity)
+
+### Fixed
+
+- **Module Name Parsing**: Resolved backtick capture in regex patterns
+  - Changed pattern from `(.+?)` to `([^`]+?)` to exclude backticks from capture group
+  - All module names now parse correctly without Markdown formatting artifacts
+- **Parser Fallback Logic**: Improved reliability with dual-strategy approach
+  - Primary strategy: Structured "## Module Information" section
+  - Fallback strategy: Inline keyword search in document body
+  - More resilient to documentation format variations
+
+### Documentation
+
+- **Test Coverage**: Added inline documentation for all test classes and methods
+  - Detailed docstrings explaining test purpose and validation criteria
+  - Examples of query types and expected behavior
+- **Parser Documentation**: Enhanced ModuleDocumentParser class documentation
+  - Strategy descriptions with regex patterns
+  - Normalization and deduplication logic explained
+- **Search Algorithm Notes**: Added comments explaining scoring improvements
+  - Minmax normalization rationale and impact
+  - Logarithmic scaling formula and use cases
+
+## [0.1.1] - 2025-10-12
+
+[0.1.1]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.1.1
+
+### Changed
+- **Expanded Module Catalog**: Search index now includes 54 Terraform AWS modules
+- **Search Index**: Rebuilt index with all 54 modules
+  - 1,535 unique keywords indexed
+  - Enhanced semantic search coverage across AWS services
+
+### Documentation
+- Each new module includes:
+  - Comprehensive descriptions with 2-3 paragraphs
+  - 10-25 key features per module
+  - 8-10 real-world use cases
+  - 20-70+ best practices across multiple categories
+  - Integration examples and code snippets
+  - Links to official AWS documentation
+  - AI agent implementation notes
+
+## [0.1.0] - 2025-10-11
+
+[0.1.0]: https://github.com/SantyagoSeaman/tfmodsearch/releases/tag/v0.1.0
+
+### Added
+- Initial release of TFModSearch MCP Server
+- Hybrid search engine combining semantic embeddings, BM25, and keyword matching
+- FastMCP-based server with stdio transport
+- Three MCP tools: `search_modules`, `get_module`, `modules_list`
+- CLI interface with `index` and `search` subcommands
+- Support for Terraform AWS modules documentation
+- Configurable search weights via YAML config
+- Security features: path validation, access controls
+- Comprehensive test suite (39 integration tests)
+- Documentation with examples and integration guides
+- Pre-commit hooks for code quality (ruff, mypy)
+- Claude Desktop integration support
+
+### Features
+- **Hybrid Search**: Combines 4 scoring signals for accurate module discovery
+- **CPU-Only Inference**: Uses `thenlper/gte-small` model for semantic embeddings
+- **Fast Indexing**: Pre-built search index with BM25 and embeddings
+- **Type Safety**: Full type hints with mypy checking
+- **Comprehensive Documentation**: README, tests documentation, inline comments

tfmodsearch-0.4.0/CONTRIBUTING.md

@@ -0,0 +1,103 @@
+# Contributing to TFModSearch MCP Server
+
+Thank you for your interest in contributing! This document provides guidelines for contributing to this project.
+
+## Development Setup
+
+1. Fork the repository
+2. Clone your fork:
+   ```bash
+   git clone https://github.com/YOUR_USERNAME/tfmodsearch.git
+   cd tfmodsearch
+   ```
+
+3. Set up development environment:
+   ```bash
+   uv venv --python 3.13
+   source .venv/bin/activate
+   uv pip install -e ".[dev]"
+   ```
+
+4. Install pre-commit hooks:
+   ```bash
+   pre-commit install
+   ```
+
+## Development Workflow
+
+1. Create a feature branch:
+   ```bash
+   git checkout -b feature/your-feature-name
+   ```
+
+2. Make your changes following the code style guidelines
+
+3. Run tests and quality checks:
+   ```bash
+   # Run all tests
+   pytest tests/ -v
+
+   # Run type checker
+   mypy src/
+
+   # Run linter and formatter
+   ruff check src/ tests/
+   ruff format src/ tests/
+
+   # Or run all checks at once
+   pre-commit run --all-files
+   ```
+
+4. Commit your changes:
+   ```bash
+   git add .
+   git commit -m "feat: add amazing feature"
+   ```
+
+5. Push to your fork and create a Pull Request
+
+## Code Style Guidelines
+
+- Follow PEP 8 style guidelines (enforced by `ruff`)
+- Add type hints to all functions (checked by `mypy`)
+- Write docstrings for public functions and classes
+- Keep line length to 120 characters
+- Use meaningful variable and function names
+
+## Commit Message Convention
+
+We use conventional commit messages:
+
+- `feat:` - New feature
+- `fix:` - Bug fix
+- `docs:` - Documentation changes
+- `test:` - Adding or updating tests
+- `refactor:` - Code refactoring
+- `chore:` - Maintenance tasks
+
+Example: `feat: add support for custom embeddings models`
+
+## Testing Guidelines
+
+- Write tests for new features
+- Ensure all tests pass before submitting PR
+- Aim for high test coverage
+- Use descriptive test names that explain what is being tested
+
+## Pull Request Process
+
+1. Update documentation if you're adding features
+2. Ensure all tests pass and code quality checks succeed
+3. Update the README.md if needed
+4. Provide a clear description of the changes in your PR
+5. Link any related issues
+
+## Questions or Issues?
+
+- Open an issue for bugs or feature requests
+- Use discussions for questions and ideas
+- Be respectful and constructive in all interactions
+
+## License
+
+By contributing, you agree that your contributions will be licensed under the MIT License.

tfmodsearch-0.4.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Alexander Makeev
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.