MCP server for Polarion ALM — read and write documents, work items, test runs, traceability links, and comments from any MCP-compatible AI assistant
Agent Claude Code
Codebase locator for dev-pipeline Stage 2 — finds the reusable helpers, sibling-tool patterns, test fixtures, and eval-coverage mechanics a plan should build on, reported as file:line references. Invoke before writing any implementation plan. Never proposes fixes or edits files; Bash is for search only.
Agent Claude Code
Executes one bounded implementation task from an approved dev-pipeline plan, strictly TDD (failing test seen before production code). Invoke during dev-pipeline Stage 3 with a plan excerpt, and Stage 7 with review findings to fix. Refuses scope outside the given task. Tool list is code-work only — no live Polarion MCP…
Agent Claude Code
Fresh-context branch-diff reviewer for dev-pipeline Stage 6 — severity-tagged findings judged against the approved spec and plan, ending in an explicit PASS/FAIL verdict on the loop's stop criterion. Invoke after gates pass, and again after every fix round. Never edits files; Bash is for the diff and focused test runs…
Agent Claude Code
Behavior-preserving cleanup pass for dev-pipeline Stage 5 — sweeps the branch diff for reuse, simplification, efficiency, altitude, and comment-density issues, applies the fixes itself, and reports each change with a one-line rationale. Invoke once per pipeline run, after gates first pass and before the first review…
Agent Claude Code
Gathers verified external-contract facts (endpoints, params, schemas, id formats, quirks) for a new feature spec. Invoke during dev-pipeline Stage 1, or any time exact vendor-API facts are needed before designing a tool. Never designs the tool itself or touches repo files; Write is for scratchpad extraction only.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: