feat(mcp): 6 spec-driven authoring tools (resolve_step, get_scenario/step_definition, step_catalog, list_tags, project_dependencies) - #1
Merged
Conversation
…uthoring resolve_step: match a Gherkin phrase to existing step definition(s) via the real StepMatcher (regex/cucumber, keyword-agnostic) — exact/ambiguous/none, with near-match suggestions when nothing binds. Reuse-first: stops agents re-authoring steps that already exist. unbound_steps: scenario steps with no matching definition, each with its scenario/feature/location — the missing-glue worklist. Read-only, existing tools untouched. +6 tests (241 total, all green).
…e multi-attribute duplicate - suggestions now OR the phrase's salient tokens and rank by shared-term count, so a near-miss that substitutes a word still surfaces the closest steps (previously AND-of-tokens returned nothing on any substitution). - a [Given]+[When]-on-one-method step now reads as one 'exact' step, not a false 'ambiguous'. +1 test (242 total, all green).
get_scenario: full scenario detail (feature, tags, kind, ordered steps) by name. get_step_definition: step-def detail + the scenarios that bind it, by expression. list_tags: tag taxonomy with per-tag scenario counts, most-used first. Read-only; existing tools untouched. +3 tests (245 total, all green).
step_catalog: reusable step vocabulary with placeholder/allowed-value extraction
(cucumber {type}, regex (a|b) enums, free params) — compose scenarios from what exists.
coverage_gaps: untested endpoints (0 scenario reach) + unused step definitions.
project_dependencies: cross-project dependency graph (depends-on/depended-on-by),
derived like the CLI map — answers 'what depends on <project>?' over MCP.
Read-only; existing tools untouched. +3 tests (248 total, all green).
… surface advertised
…tors - durable test for the regex-alternation -> enum values path (was eval-only). - coverage_gaps now returns totalEndpoints/totalStepDefinitions so '0 of 0' (no data indexed) is distinct from '0 of 50' (genuinely covered). +1 test (249).
Karzone
changed the title
…s step hygiene Both detected step-gap/hygiene (unbound scenario steps, unused step definitions), which SpecHygiene already covers authoritatively — avoid a second, map-staleness-prone source of truth. resolve_step retains the per-step authoring check; list_endpoints still exposes per-endpoint scenario reach. Net 6 new tools (11 total). 247 tests green.
Karzone
changed the title
Karzone
marked this pull request as ready for review
July 29, 2026 18:51
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds 6 read-only MCP tools that ground an AI agent in what already exists when authoring/maintaining BDD tests — reuse-first, not regenerate. Takes the server from 5 → 11 tools.
Tools
resolve_step— resolve a Gherkin phrase to the existing definition(s) that would bind it, via the realStepMatcher(regex/cucumber, keyword-agnostic).exact/ambiguous/none; onnone, returns existing steps ranked by shared terms (a near-miss that swaps a word still surfaces the closest steps).get_scenario— full scenario detail (feature, tags, kind, ordered steps) by name.get_step_definition— step-def detail + the scenarios that bind it, by expression.step_catalog— reusable step vocabulary with extracted placeholders & allowed values (cucumber{type}, regex(a|b)enums).list_tags— tag taxonomy with per-tag scenario counts.project_dependencies— the implied project dependency graph (depends-on / depended-on-by), derived like the CLImap. Answers "what depends on the Party project?" over MCP.Scope boundary (why no gap/hygiene tools)
Earlier drafts of this branch also added
unbound_stepsandcoverage_gaps. Removed: SpecHygiene already owns step hygiene in both directions (unbound scenario steps + unused step definitions) authoritatively, so duplicating it here would add a second, map-staleness-prone source of truth.resolve_stepretains the per-step authoring check;list_endpointsstill exposes per-endpoint scenario reach for coverage questions.Safety
resolve_stepreuses the indexer's binder, so "would this bind?" matches runtime resolution.Tests / eval
235 → 247tests, all green; the 235 originals unchanged.step_catalogenum extraction andresolve_stepranked near-miss suggestions).project_dependenciesreturned the 8 projects depending onOneFAT.API.Party;step_catalogpulled real allowed-values[Auto, Allianz, AllianzItaly, AllianzVoe].Known limitations
resolve_stepsuggestions andstep_catalogallowed-values are lexical/structural, not semantic.testatlas index(endpoints + full step-binding); a stale v4 map yields thin results.Draft: prototype for review before any release / version bump.