# Context Goblin Context Goblin is an AI code reviewer for GitHub pull requests that uses call-graph and cross-repository context to assess changes. It helps development teams review architecture, security, logic, and test concerns with findings verified against the codebase. - Findings24: - Site: - Preços: Freemium - Categorias: Revisão de código com IA ## O que é Context Goblin? Context Goblin is an AI code reviewer for GitHub pull requests. It builds context from the diff, changed symbols, callers elsewhere in the codebase, and relevant dependencies between repositories before its review agents analyze the change. The core purpose is to evaluate how a change affects the wider system, rather than limiting review to the edited lines. The review pipeline covers architecture, security, logic, and tests. A coordinator combines the specialist findings into one review with a summary, inline comments, categories, severities, and a verdict. Reviews can be started from the dashboard, through GitHub, or with the product's review_repository MCP tool. ## O que Context Goblin pode fazer? ### Call-graph analysis Resolves the symbols touched by a pull request and follows their callers in other files, surfacing downstream code that a line-by-line review may not open. ### Cross-repository dependency map Maps package dependencies, HTTP calls, queues, and shared tables between indexed repositories. Relations relevant to a change are added to its review context. ### Specialist review agents Runs separate agents for architecture, security, logic, and tests. A coordinator removes duplicate findings, resolves disagreements, and produces the published review. ### Verified findings Checks findings against the codebase before posting them. Reviews include categories, severities, a summary, inline comments, and a GitHub review verdict when started through GitHub. ### Custom review context Lets administrators disable built-in agents, add custom agents with their own prompts, and connect MCP servers so agents can read information from tools such as Jira, Confluence, Linear, or Notion. ### Multiple review entry points Supports dashboard reviews, @contextgoblin mentions on GitHub pull requests, automatic reviews for enabled repositories, and review requests through an MCP client. ## Casos de uso ### Review changes with downstream callers Use call-graph context when a pull request changes a shared function or symbol and the important effects may appear in other files that were not edited. ### Assess service-to-service changes Use the cross-repository map when a change affects routes, queues, shared tables, or package dependencies spanning multiple repositories. ### Add ticket and documentation context Connect an MCP-compatible source such as Jira, Confluence, Linear, or Notion when reviewers need to compare an implementation with project requirements or documentation. ### Automate pull request checks Enable automatic reviews for selected repositories, or trigger a review by mentioning @contextgoblin on a pull request. GitHub receives the review summary, inline comments, and verdict. ### Run reviews from an internal workflow Start a review from the dashboard for a result kept in Context Goblin, or call the review\_repository MCP tool from another agent or MCP client. ## Perguntas frequentes ### How does Context Goblin start a review? A review can be started by pasting a pull request URL into the dashboard, mentioning @contextgoblin on a pull request, enabling automatic reviews for an enabled repository, or calling the review\_repository MCP tool. ### What does a review contain? The review includes a Review Overview with a verdict and summary, plus inline comments on changed lines. Findings are categorized as Architecture, Security, Logic, or Tests and carry a severity. GitHub-triggered reviews use GitHub review states such as Approve, Approve with comments, or Request changes. ### Can Context Goblin review more than one repository? Yes. Its cross-repository context can map package dependencies, HTTP calls, queues, and shared tables between indexed repositories. Relations that the pull request touches are included in the review context. ### Can the review agents use information outside the codebase? Yes. Administrators can connect MCP servers and select which tools the agents may call. The site lists Jira, Confluence, Notion, and other MCP-compatible servers as examples. ### What happens after the free credits are used? Reviews pause when the free credits run out. The site states that nothing is charged and nothing is deleted, and says users can ask for more credits. A team offering with more repositories and reviews is available by request. ## Informações rápidas - **Category**: AI code review and developer tool - **Primary platform**: GitHub pull requests - **Review areas**: Architecture, security, logic, and tests - **Entry points**: Dashboard, GitHub App, and MCP client - **Context sources**: Call graph, cross-repository relations, and connected MCP servers - **Free access**: One full repository crawl and initial pull request reviews; no credit card required