v3.0.0 add word find_many tool

[3.0.0] - 2026-08-02

Added

• word_find_many tool — exact-match search for multiple terms in a single
  call. Accepts a list of 1-25 terms and returns one WordFindEntry
  (term + results) per input term, in the same order as the input, so
  a term matching zero, one, or several vocabulary entries is always
  represented positionally rather than as a flattened, ambiguous list.
  New models/word_find_entry.py (WordFindEntry model) and
  tools/words/word_find_many.py, registered in tools/words/__init__.py.
• asdste100://version MCP resource — installed version numbers of the MCP
  server itself (biz-dfch-asdste100mcp) and its three data-backing
  libraries: biz-dfch-asdste100vocab, biz-dfch-asdste100rules, and
  biz-dfch-asdste100nlp. Backed by importlib.metadata.version(); no
  network calls. New models/version_info.py (VersionInfo model) and
  resources/version.py, registered in resources/__init__.py.
• MCP resource support, in addition to tools:
  • asdste100://rules/toc — table-of-contents outline of the ruleset (mirrors
    the rules_toc tool with no section filter).
  • asdste100://rules/rule/{id_} — a single rule/recommendation/information
    item by exact id, e.g. asdste100://rules/rule/R1.1 (mirrors the
    rules_find tool).
  • New resources/ package (mirroring the existing tools/ package
    structure), with a rules/ sub-package for the rules resources;
    registered in server.py alongside the existing tools import.
  • tests/resources/rules/ — unit tests for both resource functions.

Changed

• Breaking: Renamed the Settings fields files, use_ste100,
  use_ste100_technical_words, and use_ste100_rules to vocab_files,
  use_asdste100_vocab, use_asdste100_technical_words, and
  use_asdste100_rules respectively (settings.py). Because
  Settings.model_config derives environment variable names from the
  field name under the ASDSTE100_MCP_ prefix, this also renames the
  corresponding environment variables: ASDSTE100_MCP_FILES →
  ASDSTE100_MCP_VOCAB_FILES, ASDSTE100_MCP_USE_STE100 →
  ASDSTE100_MCP_USE_ASDSTE100_VOCAB,
  ASDSTE100_MCP_USE_STE100_TECHNICAL_WORDS →
  ASDSTE100_MCP_USE_ASDSTE100_TECHNICAL_WORDS, and
  ASDSTE100_MCP_USE_STE100_RULES → ASDSTE100_MCP_USE_ASDSTE100_RULES.
  Update any .env file, shell environment, or deployment config that
  sets these variables under their previous names. server.py's lifespan
  function and tests/settings/test_factory.py were updated accordingly.
• Bumped the biz-dfch-asdste100vocab runtime dependency from >=0.8.0
  to >=0.8.1.
• Updated README.md wording in the Authentication section to mention
  resources alongside tools, and added a note about the MCP Inspector
  sometimes failing to connect over stdio (use sse instead).

Full Changelog: v2.0.0...v3.0.0

v2.0.0 WordNet Synonyms, MCP Registry Publishing & Breaking Rename

[2.0.0] - 2026-08-02

Added

• New word_synonym read-only tool backed by the biz-dfch-asdste100nlp
  library's Nlp class: looks up WordNet synsets for a word and cross-
  references their lemma names against the ASD-STE100 Issue 9 vocabulary,
  returning approved/rejected entries that are synonyms of the input word.
  Registered in tools/words/word_synonym.py, following the same
  structure as word_find/word_fuzzy; the shared Nlp instance is
  created in server.py's lifespan (_nlp/_get_nlp()), wrapping the
  same Vocab instance used by the other vocabulary tools.
• Runtime dependency on biz-dfch-asdste100nlp>=0.1.0 in pyproject.toml.
• tests/tools/words/test_word_synonym.py — unit tests for the
  Nlp.synonym lookup and the word_synonym tool function.
• MCP Registry publishing support via new publish-to-mcp-registry GitHub Actions job:
  • Automatically publishes server metadata to the official MCP Registry
  • Uses GitHub OIDC authentication (zero secrets, consistent with PyPI Trusted Publishing)
  • Auto-patches server.json version fields from git tag at publish time
  • Runs after PyPI publication succeeds in the CI workflow
• server.json at repository root defining MCP Registry metadata:
  • Server identity: io.github.dfch/biz-dfch-asdste100mcp (GitHub OIDC namespace)
  • PyPI package configuration with biz-dfch-asdste100mcp identifier
  • Documentation of all 8 ASDSTE100_MCP_* environment variables with format hints
  • Conforms to MCP Registry schema v2025-12-11
• PyPI ownership verification marker in README.md long-description for registry discovery

Changed

• Updated release workflow documentation in README.md to include MCP Registry publishing step
• Added "MCP Registry Publishing" section in README.md with setup and verification instructions
• Breaking: Upgraded MCP dependency to version 2.0.0, requiring MCP protocol 2.0 or newer
• Breaking: Renamed the word_similar vocabulary tool to word_fuzzy for clarity
  (tools/words/word_similar.py → tools/words/word_fuzzy.py). Any caller invoking the
  tool by its previous name must switch to word_fuzzy.
• Breaking: Renamed the console script entry point from ste100-mcp to
  asdste100-mcp ([project.scripts] in pyproject.toml), and the MCP
  server/Typer app name to match. Update any launch command, OpenCode
  mcp config, or shell alias that invokes ste100-mcp to use
  asdste100-mcp instead.
• Breaking: Renamed all STE100_MCP_* environment variables to
  ASDSTE100_MCP_* (STE100_MCP_TRANSPORT, STE100_MCP_HOST,
  STE100_MCP_PORT, STE100_MCP_FILES, STE100_MCP_USE_STE100,
  STE100_MCP_USE_STE100_TECHNICAL_WORDS, STE100_MCP_RULES_FILES,
  STE100_MCP_USE_STE100_RULES), matching the asdste100-mcp entry point
  name (settings.py's env_prefix, cli.py's envvar= bindings, and
  server.json). Update any .env file, shell environment, or deployment
  config that sets these variables under their previous names.

Full Changelog: v1.0.0...v2.0.0

v1.0.0 add pagination BREAKING CHANGE

[1.0.0] - 2026-08-01

Added

• max_results (default 25) and offset (default 0) pagination parameters
  on rules_search, rules_examples, word_list, and word_match --
  the four tools whose unpaginated result can grow large enough (over a
  thousand items, for rules_examples; the full multi-thousand-entry
  vocabulary, for word_list/word_match) that a full page is otherwise
  ambiguous.
• New PagedResult base model (src/biz/dfch/asdste100mcp/models/paged_result.py)
  carrying the shared total, offset, max_results, and truncated
  pagination fields, plus three tool-specific subclasses built on it:
  SearchResult (rules_search, results: list[Rule]),
  RulesExamplesResult (rules_examples, results: list[ContentItem]),
  and WordResult (word_list/word_match, results: list[Word]).
  truncated tells callers whether a full page means "that's all the
  matches" or "there are more -- call again with a higher offset".
• src/biz/dfch/asdste100mcp/tools/_pagination.py -- shared MaxResults/
  Offset parameter types and a paginate() slicing helper used by all
  four paginated tools.

Changed

• Breaking: rules_search now returns a SearchResult object instead
  of a bare list[Rule]; the page of matching rules is available at
  result.results.
• Breaking: rules_examples now returns a RulesExamplesResult
  object instead of a bare list[ContentItem]; the page of matching
  content items is available at result.results.
• Breaking: word_list and word_match now return a WordResult
  object instead of a bare list[Word]; the page of matching vocabulary
  entries is available at result.results.

Full Changelog: v0.2.0...v1.0.0

v0.2.0 add rules

[0.2.0] - 2026-08-01

Added

• Eight rules_* read-only tools backed by the biz-dfch-asdste100rules library:
  rules_find, rules_match, rules_search, rules_by_section,
  rules_by_category, rules_examples, rules_overview, rules_toc.
  Return types reuse the library's own pydantic models (Rule,
  ContentItem, RuleOverview, ContentType, EntryType) directly; a
  new TocEntry model wraps the plain tuples returned by Rules.toc().
• --rules-file / -r CLI option on ste100-mcp (repeatable): pass one or
  more rules JSON files at startup, merged with STE100_MCP_RULES_FILES.
• Settings.rules_files and Settings.use_ste100_rules fields;
  Factory.create_instance(extra_files, extra_rules_files) now also
  merges rules file paths.
• tests/tools/rules/ — unit tests for all eight new tools (via Rules
  directly, following the existing tool-test convention).
• tests/models/test_toc_entry.py — unit tests for the new TocEntry model.
• Python 3.14 to the list of supported and CI-tested versions.

Changed

• Renamed the five vocabulary tools with a word_ prefix and moved them
  into a new tools/words/ sub-package: word_find, word_match,
  word_similar, word_list, word_count (was find, match,
  similar, list, count). Breaking: any caller invoking the
  tools by their previous names must switch to the new word_* names.

Full Changelog: v0.1.3...v0.2.0

v0.1.3 Maintenance

[0.1.3] - 2026-07-30

Changed

• Use new vocab package name: biz-dfch-asdste100vocab.

Doc

• Fix README table formatting (pipe alignment in tool tables).
• Rename CI workflow from "Pylint and unittest" to "Lint and Test".

Full Changelog: v0.1.2...v0.1.3

v0.1.2 add file support for custom vocabulary

[0.1.2] - 2026-07-30

Added

• --file / -f CLI option on ste100-mcp (repeatable): pass one or more
  *.jsonl vocabulary files at startup. Paths are merged with those from
  STE100_MCP_FILES; duplicates are removed while preserving order.
• Factory class in settings.py — thread-safe singleton for
  Settings using a double-checked lock. Class-level state removes the
  need for a module-level global variable.
  • Factory.create_instance(extra_files) — creates the singleton, merges
    CLI-supplied paths with env-supplied paths, asserts it is called only once.
  • Factory.get_instance() — returns the singleton, asserts
    create_instance was called first.
• tests/settings/test_factory.py — 8 unit tests covering singleton
  creation, identity, deduplication, and both AssertionError guard paths.

Changed

• cli.py calls Factory.create_instance(files or []) before mcp.run()
  so CLI-supplied paths are available to the server lifespan.
• server.py _lifespan calls Factory.get_instance() instead of
  constructing a new Settings() on every startup.

Fixed

• Removed bogus --file validation block in cli.py that called
  load_dotenv instead of exiting on error (Typer's exists=True already
  rejects non-existent paths).
• Changed --file default from mutable [] to None to avoid the
  mutable-default-argument footgun.

Full Changelog: v0.1.1...v0.1.2

v0.1.1 maintenance

[0.1.1] - 2026-07-30

Changed

• Extracted each MCP tool into its own module under the new
  src/biz/dfch/asdste100mcp/tools/ sub-package (find.py, match.py,
  similar.py, list.py, count.py). server.py now contains only the
  shared infrastructure (mcp, _lifespan, _READ_ONLY, _Term,
  _get_vocab) and imports the tools package at the bottom to trigger
  registration. All nine unit tests continue to pass unchanged.
• Moved tool unit tests (test_find.py, test_match.py, test_similar.py,
  test_count.py) from tests/ into the new tests/tools/ sub-package to
  mirror the source layout.
• Moved test_word.py from tests/ into the new tests/models/ sub-package
  to mirror the source layout.

Full Changelog: v0.1.0...v0.1.1

v0.1.0 initial release

0.1.0 - 2026-07-28

Added

• Initial project scaffold for biz.dfch.AsdSte100Mcp.
• AGPL-3.0-or-later license and SPDX headers in all Python source files.
• Runtime dependencies: python-dotenv>=1.2.2, mcp, typer>=0.12,
  pydantic-settings, biz-dfch-ste100vocab>=0.7.1.
• Console script entrypoint ste100-mcp pointing to biz.dfch.asdste100mcp.cli:app.
• MCPServer with five read-only tools: find, match, similar, list, count;
  backed by the biz-dfch-ste100vocab library.
• Typer CLI entry point with --transport, --host, and --port options for
  stdio and SSE transport modes.
• Settings class (settings.py) reading STE100_MCP_* environment variables
  via pydantic-settings.
• Pydantic models Word, WordMeaning, WordNote in models/ package,
  mirroring the vocab library dataclasses.
• Unit tests for all five tools in tests/ (one file per test class).
• GitHub Actions CI workflow (ci.yml) running ruff, pylint, and unit tests
  across Python 3.11, 3.12, and 3.13.
• GitHub Actions publish workflow (publish.yml) building and publishing to
  TestPyPI and PyPI on version tags, then creating a GitHub release.
• CONTRIBUTING.md, CODE_OF_CONDUCT.md, SECURITY.md, and NOTICE.
• CHANGELOG.md following the Keep a Changelog 1.0.0 format.

Fixed

• Replaced incorrect dotenv dependency with python-dotenv.
• Corrected stale biz.dfch.AsdSte100Vocab project name references in
  README.md, pyproject.toml, cli.py, __main__.py, and __init__.py.
• Fixed TOML syntax error (missing comma in classifiers array in pyproject.toml).
• Removed over-broad pull-requests: write permission from CI workflow.
• Pinned pypa/gh-action-pypi-publish from floating release/v1 tag to full
  commit SHA.
• Removed unnecessary single-entry strategy/matrix from publish workflow jobs.
• Fixed pylint warnings: added missing docstrings, moved import os to top level
  in cli.py, and added targeted pylint: disable comments in server.py for
  justified suppressions (invalid-name, global-statement, protected-access).
• Replaced local editable path dependency on biz-dfch-ste100vocab with PyPI
  reference to fix CI failure on GitHub Actions runners.
• Let uv manage the Python version in CI by passing python-version to
  astral-sh/setup-uv instead of using actions/setup-python.

Removed

• Removed main.py scaffold placeholder; the real entry point is cli.py.

Full Changelog: https://github.com/dfch/biz.dfch.AsdSte100Mcp/commits/v0.1.0