@yasserkhanorg/e2e-agents 0.5.15 → 0.5.16

package/dist/agent/ai_mapping.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"ai_mapping.d.ts","sourceRoot":"","sources":["../../src/agent/ai_mapping.ts"],"names":[],"mappings":"AAMA,OAAO,KAAK,EAAC,UAAU,EAAC,MAAM,eAAe,CAAC;AAC9C,OAAO,KAAK,EAAC,qBAAqB,EAAC,MAAM,aAAa,CAAC;AACvD,OAAO,KAAK,EAAC,YAAY,EAAE,QAAQ,EAAC,MAAM,YAAY,CAAC;AA4BvD,MAAM,WAAW,eAAe;IAC5B,OAAO,EAAE,OAAO,CAAC;IACjB,IAAI,EAAE,OAAO,CAAC;IACd,QAAQ,EAAE,MAAM,CAAC;IACjB,WAAW,EAAE,MAAM,CAAC;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,QAAQ,EAAE,YAAY,EAAE,CAAC;IACzB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACtB;AA4WD,wBAAsB,iBAAiB,CACnC,OAAO,EAAE,MAAM,EACf,SAAS,EAAE,MAAM,EACjB,MAAM,EAAE,qBAAqB,EAC7B,KAAK,EAAE,UAAU,EAAE,EACnB,KAAK,EAAE,QAAQ,EAAE,GAClB,OAAO,CAAC,eAAe,CAAC,CA+O1B"}
+{"version":3,"file":"ai_mapping.d.ts","sourceRoot":"","sources":["../../src/agent/ai_mapping.ts"],"names":[],"mappings":"AAMA,OAAO,KAAK,EAAC,UAAU,EAAC,MAAM,eAAe,CAAC;AAC9C,OAAO,KAAK,EAAC,qBAAqB,EAAC,MAAM,aAAa,CAAC;AACvD,OAAO,KAAK,EAAC,YAAY,EAAE,QAAQ,EAAC,MAAM,YAAY,CAAC;AA4BvD,MAAM,WAAW,eAAe;IAC5B,OAAO,EAAE,OAAO,CAAC;IACjB,IAAI,EAAE,OAAO,CAAC;IACd,QAAQ,EAAE,MAAM,CAAC;IACjB,WAAW,EAAE,MAAM,CAAC;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,QAAQ,EAAE,YAAY,EAAE,CAAC;IACzB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACtB;AA4WD,wBAAsB,iBAAiB,CACnC,OAAO,EAAE,MAAM,EACf,SAAS,EAAE,MAAM,EACjB,MAAM,EAAE,qBAAqB,EAC7B,KAAK,EAAE,UAAU,EAAE,EACnB,KAAK,EAAE,QAAQ,EAAE,GAClB,OAAO,CAAC,eAAe,CAAC,CAgP1B"}

package/dist/agent/ai_mapping.js

@@ -422,12 +422,13 @@ async function mapAITestsToFlows(appRoot, testsRoot, config, flows, tests) {
         '- Keep at most 5 tests per flow.',
         '- Use exact flowId values from FLOWS.',
         '- A flow may only map to tests listed under FLOW_CANDIDATE_SIGNALS for that flow.',
-        '- Map a test when its file path OR test content titles demonstrate behavioral overlap with the flow scenario. A file named search_user_post_spec.js with titles about searching messages covers search_messages. Generic subsystem similarity alone is not enough.',
-        '- When FLOW_CANDIDATE_SIGNALS shows 3 or more candidates for a flow, map ALL that have any behavioral connection to the flow domain — collectively they represent comprehensive coverage. For example, 4+ Cypress search specs collectively cover search_messages even if each only covers one search scenario.',
-        '- Only return tests: [] when no candidate has ANY behavioral connection to the flow.',
+        '- Map a test when its file path structure OR test content titles specifically indicate it covers the flow scenario. Behavioral specificity is required — "search_user_post_spec.js" in a /search/ directory covers search_messages because it specifically tests searching for messages. A file named "find_channels.spec.ts" does NOT cover search_messages even if it is in a search-related path because it tests channel navigation, not message searching.',
+        '- Map every candidate that has specific behavioral evidence. Multiple files each covering a different aspect of the same flow should all be mapped.',
+        '- For candidates whose content you have not read, judge by path structure alone: map only when the path clearly names the specific behavior (not just a general subsystem keyword).',
+        '- Only return tests: [] when no candidate has specific behavioral connection to the flow.',
         '- missingScenarios decision tree based on tests.length AFTER you have determined your test mappings:',
-        '  * tests.length >= 4: return missingScenarios: [] — comprehensive coverage, no gap to report.',
-        '  * tests.length 1-3: list only scenarios that are genuinely absent from ALL mapped tests combined.',
+        '  * tests.length >= 3: return missingScenarios: [] — three or more specific tests covering different scenarios = comprehensive coverage.',
+        '  * tests.length 1-2: list only scenarios that are genuinely absent from ALL mapped tests combined.',
         '  * tests.length 0: list 3-5 core user-facing scenarios that must be covered.',
         '  Write each scenario as a short imperative starting with a verb.',
         '',

package/dist/agent/runner.js

@@ -45,10 +45,10 @@ function computeGaps(flows, coverageMap, coverage) {
         if (coveredBy.length === 0) {
             return true; // no tests at all
         }
-        // Flows with 4+ mapped tests are considered comprehensively covered — the AI
-        // maps 4 tests only when multiple spec files collectively demonstrate coverage.
+        // Flows with 3+ mapped tests are considered comprehensively covered — the AI
+        // maps 3 tests only when each has specific behavioral evidence for the flow.
         // missingScenarios from the AI are informational for such flows, not blocking.
-        if (coveredBy.length >= 4) {
+        if (coveredBy.length >= 3) {
             return false;
         }
         // For flows with 1-3 tests, flag as a gap if AI identified missing scenarios.

package/dist/esm/agent/ai_mapping.js

@@ -419,12 +419,13 @@ export async function mapAITestsToFlows(appRoot, testsRoot, config, flows, tests
         '- Keep at most 5 tests per flow.',
         '- Use exact flowId values from FLOWS.',
         '- A flow may only map to tests listed under FLOW_CANDIDATE_SIGNALS for that flow.',
-        '- Map a test when its file path OR test content titles demonstrate behavioral overlap with the flow scenario. A file named search_user_post_spec.js with titles about searching messages covers search_messages. Generic subsystem similarity alone is not enough.',
-        '- When FLOW_CANDIDATE_SIGNALS shows 3 or more candidates for a flow, map ALL that have any behavioral connection to the flow domain — collectively they represent comprehensive coverage. For example, 4+ Cypress search specs collectively cover search_messages even if each only covers one search scenario.',
-        '- Only return tests: [] when no candidate has ANY behavioral connection to the flow.',
+        '- Map a test when its file path structure OR test content titles specifically indicate it covers the flow scenario. Behavioral specificity is required — "search_user_post_spec.js" in a /search/ directory covers search_messages because it specifically tests searching for messages. A file named "find_channels.spec.ts" does NOT cover search_messages even if it is in a search-related path because it tests channel navigation, not message searching.',
+        '- Map every candidate that has specific behavioral evidence. Multiple files each covering a different aspect of the same flow should all be mapped.',
+        '- For candidates whose content you have not read, judge by path structure alone: map only when the path clearly names the specific behavior (not just a general subsystem keyword).',
+        '- Only return tests: [] when no candidate has specific behavioral connection to the flow.',
         '- missingScenarios decision tree based on tests.length AFTER you have determined your test mappings:',
-        '  * tests.length >= 4: return missingScenarios: [] — comprehensive coverage, no gap to report.',
-        '  * tests.length 1-3: list only scenarios that are genuinely absent from ALL mapped tests combined.',
+        '  * tests.length >= 3: return missingScenarios: [] — three or more specific tests covering different scenarios = comprehensive coverage.',
+        '  * tests.length 1-2: list only scenarios that are genuinely absent from ALL mapped tests combined.',
         '  * tests.length 0: list 3-5 core user-facing scenarios that must be covered.',
         '  Write each scenario as a short imperative starting with a verb.',
         '',

package/dist/esm/agent/runner.js

@@ -41,10 +41,10 @@ function computeGaps(flows, coverageMap, coverage) {
         if (coveredBy.length === 0) {
             return true; // no tests at all
         }
-        // Flows with 4+ mapped tests are considered comprehensively covered — the AI
-        // maps 4 tests only when multiple spec files collectively demonstrate coverage.
+        // Flows with 3+ mapped tests are considered comprehensively covered — the AI
+        // maps 3 tests only when each has specific behavioral evidence for the flow.
         // missingScenarios from the AI are informational for such flows, not blocking.
-        if (coveredBy.length >= 4) {
+        if (coveredBy.length >= 3) {
             return false;
         }
         // For flows with 1-3 tests, flag as a gap if AI identified missing scenarios.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yasserkhanorg/e2e-agents",
-  "version": "0.5.15",
+  "version": "0.5.16",
   "description": "Pluggable LLM provider library for AI-powered test automation. Use Claude, Ollama, or your own LLM. Integrate with Playwright, Jest, or any test framework. MCP server for test agents, cost tracking, and hybrid provider mode.",
   "main": "dist/index.js",
   "module": "dist/esm/index.js",