mcp-agent-bridge

Send a plan, diff, or implementation to a Claude instance for independent review. Returns structured JSON with verdict, issues, and suggestions.

Ask a Claude instance a freeform question about the codebase. Returns text.

Specialized code review that analyzes a git diff. Returns structured JSON with verdict, issues, and suggestions.

Analyze a Playwright release-failure evidence bundle and return a structured diagnostic report. The caller supplies the full prompt (run summary, failed/flaky results, and repo context) as `content`; the model's output is constrained to the findings[] schema (runSummary + per-finding rootCause, conf…

Lightweight, non-agentic triage for a stale or low-priority item — e.g. deciding what to do with a PR review that's gone quiet. Returns a verdict from a small fixed set plus a one-sentence reason. Not a full code review.

One turn of an interactive chat with resource context and optional tool proposals. The caller assembles the full prompt (system instructions, resource context, available tool descriptions, and conversation history) — this tool has no memory of prior turns. Returns a text reply plus zero or more prop…

Generate a personalized, compelling cover letter from a resume and job description. Returns an analysis, the complete letter, alternative opening hooks, and interview talking points.

Generate both an ATS-compatible and a designed resume for creative professionals (graphic designers, UX, marketing, writers, photographers, etc.). Returns both versions plus field-specific tips and portfolio link strategy.

Craft a C-suite, VP, or Director-level resume that tells a transformation story. Returns executive profile, core competencies, career highlights, full experience section, and coaching notes.

Generate STAR stories, predicted questions, self-introduction pitch, and interview strategy tailored to a specific role.

Analyze a job posting to extract requirements, calculate match score, detect red flags, and generate tailored application strategy.

Optimize a LinkedIn profile with keyword-rich headline options, rewritten About section, experience bullet rewrites, skills recommendations, and a 30-day action plan.

Transform a project into a compelling portfolio case study with problem, process, solution, results, and learnings. Returns full case study, executive summary, and interview prep notes.

Build a formatted professional reference list with briefing emails, permission scripts, talking points, and timing strategy for each reference.

Optimize a resume to pass ATS screening. Returns keyword gap analysis, match score, formatting fixes, and optimized sections with projected score improvement.

Transform weak, duty-focused resume bullets into achievement-focused statements using the X-Y-Z formula. Returns diagnosed issues, metric-extraction questions, and 2-3 rewritten versions per bullet.

Audit and fix resume formatting for ATS compatibility and human readability. Returns formatting audit, specific fixes, reformatted sections, and ATS compatibility score.

Add metrics and data-driven impact to resume bullets. Returns discovery questions, estimated metrics, and 2 quantified versions (conservative and optimistic) per bullet.

Build all resume sections optimized for career stage and target role. Returns section order, professional summary, skills section, experience guidance, and complete section checklist.

Tailor a resume to a specific job posting with keyword extraction, match audit, rewritten summary, prioritized skills, and bullet rewrites. Maintains authenticity — all changes reflect genuine experience.

Create and organize a master resume, manage tailored versions with naming conventions, track applications, and establish an update workflow.

Optimize resumes for software engineering, PM, data, and DevOps roles with skill section restructuring, technical bullet rewrites, projects section guidance, and GitHub recommendations.

Extract a structured, fact-ID-tagged database of career facts from a resume, LinkedIn profile, brag document, or any source material. Preserves truthfulness for downstream resume tailoring.

Simulate a skeptical hiring manager's 45-second resume screen. Returns a Yes/Maybe/No decision, top reasons and concerns, scores across 6 dimensions, and exact rewrites to make before applying.

Push the active branch and open a GitHub pull request with a structured description, linked ticket, reviewers, and labels. Returns step-by-step workflow instructions.

Systematic file-by-file GitHub PR code review. Posts inline comments and a verdict. Returns step-by-step workflow instructions.

Push the active branch and open an Azure DevOps pull request with a structured description, linked work items, and optional auto-complete. Returns step-by-step workflow instructions.

Systematic file-by-file ADO PR code review. Posts inline thread comments and a vote. Returns step-by-step workflow instructions.

Send a prompt to OpenAI Codex CLI for code generation or analysis.

Continue a conversation with Codex by sending a follow-up reply. Note: Does not maintain actual session state; context is passed inline.

Extract frames from a video (fixed-fps or scene-change sampling) and narrate the sequence with Claude. Returns both the per-frame descriptions and a single synthesized narration string.

Extract frames from a video (fixed-fps or scene-change sampling) without narrating them. Returns a sessionId plus a thumbnail per frame; pass the sessionId to narrate_frames to describe them, or to close_session to free the extracted files early. Sessions auto-expire after 30 minutes.

Narrate frames previously extracted via extract_frames, identified by sessionId. Can be called more than once against the same sessionId (e.g. with a different prompt or batchSize) without re-extracting. Pass close: true to free the session's frame files immediately after this call instead of waitin…

Free the frame files held by a session started with extract_frames. Safe to call on an already-closed or expired sessionId (no-op).