feat(mcp): implement 7 test-loop improvements from real usage feedback

[mrdailey99]

Sources: PROVARDX_CLI_FEEDBACK.md from provar-manager-regression session.

Source changes:

• testCaseValidate.ts: DATA-001 warning for (silently ignored by CLI standalone runner); ASSERT-001 warning for AssertValues with namedValues id="values" format (silently passes as null=null for variable comparisons); improved TC_012/TC_031 suggestion strings naming crypto.randomUUID() and the UUID v4 variant byte rule.
• automationTools.ts: filterTestRunOutput() strips com.networknt.schema and logger-lock SEVERE noise before returning testrun response; output_lines_suppressed field added when lines are dropped.
• rcaTools.ts: incrementSiblingDirs() detects Results(N) siblings (Provar Increment mode creates siblings, not numeric children); 5 new Salesforce DML error RCA rules: SALESFORCE_VALIDATION, SALESFORCE_PICKLIST, SALESFORCE_REFERENCE, SALESFORCE_ACCESS, SALESFORCE_TRIGGER.

Tests (22 new, 626 total):

• testCaseValidate.test.ts: DATA-001 (fires/doesn't fire), ASSERT-001 (fires/doesn't fire for correct args/non-AssertValues), TC_012/TC_031 suggestion text.
• automationTools.test.ts: filterTestRunOutput pure-function tests (7), testrun output_lines_suppressed handler tests (2).
• rcaTools.test.ts: Results(N) sibling detection (2), Salesforce category classification (4).

Docs: docs/mcp.md — DATA-001/ASSERT-001 warning rules, Salesforce RCA categories with infrastructure_issues note, output_lines_suppressed field.

[Copilot]

Pull request overview

Implements several MCP “test loop” improvements driven by real-world Provar CLI usage feedback: adds additional validator warnings/suggestions, reduces noisy CLI output returned from testrun, and improves RCA artifact resolution and classification.

Changes:

• Add new testcase validation warnings (DATA-001, ASSERT-001) and improve UUID v4 suggestion text (TC_012/TC_031).
• Filter Provar testrun stdout to suppress known schema-validator/logger noise and surface output_lines_suppressed.
• Enhance RCA tools: detect Increment-mode Results(N) sibling dirs and classify common Salesforce DML failures into new RCA categories; update docs and add unit tests.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
src/mcp/tools/testCaseValidate.ts	Adds DATA-001 / ASSERT-001 warnings and updates UUID suggestion strings.
src/mcp/tools/automationTools.ts	Introduces filterTestRunOutput() and plumbs output_lines_suppressed into testrun responses.
src/mcp/tools/rcaTools.ts	Adds Salesforce RCA rules and Increment-mode Results(N) sibling directory resolution.
test/unit/mcp/testCaseValidate.test.ts	Adds unit tests covering new warnings and updated suggestion text.
test/unit/mcp/automationTools.test.ts	Adds testrun suppression tests and pure-function tests for output filtering.
test/unit/mcp/rcaTools.test.ts	Adds tests for Results(N) sibling resolution and Salesforce category classification.
docs/mcp.md	Documents new warnings, new RCA categories, and the output_lines_suppressed field/behavior.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

filterTestRunOutput() splits on "\n" only, so Windows CRLF output will leave trailing "\r" on each line. That can prevent NOISE_PATTERNS from matching (and can also break blank-line collapsing / downstream string comparisons). Consider splitting with /\r?\n/ (or normalizing \r\n→\n first) and trimming a trailing \r from each line before pattern checks.

Suggested change

	const lines = raw.split('\n');
	const kept: string[] = [];
	let suppressed = 0;
	let lastKeptWasBlank = false;
	for (const line of lines) {
	const lines = raw.split(/\r?\n/);
	const kept: string[] = [];
	let suppressed = 0;
	let lastKeptWasBlank = false;
	for (const rawLine of lines) {
	const line = rawLine.endsWith('\r') ? rawLine.slice(0, -1) : rawLine;

[mrdailey99]

Fixed in b8f3c87 — changed raw.split('\n') to raw.split(/\r?\n/) and added const line = rawLine.endsWith('\r') ? rawLine.slice(0, -1) : rawLine to strip any residual CR before pattern checks. Added a CRLF-specific test (627 passing).

[Copilot]

resolveResultsLocation() uses run_index directly when selecting a Results(N) sibling. If a caller passes run_index <= 0 (or a negative number), this will construct a path like Results(0)/Results(-1) and report it as the resolved directory. Consider validating run_index is a positive integer at the tool schema level (e.g. z.number().int().positive()) or returning an error when run_index <= 0.

[mrdailey99]

Fixed in b8f3c87 — both provar.testrun.report.locate and provar.testrun.rca now use z.number().int().positive().optional() for run_index, so the MCP layer rejects 0 and negative values before they reach resolveResultsLocation.