pod
Other formats agents might prefer:
markdownjsonllms.txt

Agent? You probably want markdown or json, or Pod over MCP.

Reported issues for mcp-oss-onramp

Pod holds 7 of 7 problems reported by people outside the maintainer team. Issues filed by the project's own owners, members and collaborators are excluded entirely — a maintainer's release checklist is not a warning to a prospective user.

Back to mcp-oss-onramp.

Most discussed

Future Consideration: Support SSE and Remote Hosting for Smithery

Background

Currently, the Smithery registration process for stdio-based servers via GitHub URL often fails due to environment mismatches (e.g., missing 'gh' CLI in Smithery's scan environment). Transitioning to a Remote/Hosted model via SSE (Server-Sent Events) would simplify registration.

Requirements

  • Remove 'gh' dependency: Replace execSync('gh ...') with a library like octokit to allow execution in environments without GitHub CLI.
  • Implement SSE Transport: Support `SSESe

Read the thread · 2026-05-18 · closed · 2 comments

[Blocker] Smithery Registration for stdio-based server (MCPB/Local Bundle)

Summary

We attempted to register mcp-oss-onramp with Smithery.ai. Currently, registration is blocked due to inconsistencies in Smithery's documentation, CLI, and WebUI support for stdio-based MCP servers.

Attempts and Results

1. URL-based Registration (SSE)

  • Attempt: smithery mcp publish <GitHub URL>
  • Result: 422 Error
  • Cause: Smithery's scanner attempts to treat the URL as a live HTTP/SSE server. Since this project is stdio-based, it fails the automatic scan/h

Read the thread · 2026-05-20 · open · 1 comment

Technical: Remove gh CLI dependency and use native library

Task: Remove gh CLI dependency and use native library

Currently, the server relies on the GitHub CLI (gh) being installed and authenticated in the user's environment to make API calls via execSync. This creates a friction point for users and limits portability.

Goal

  • Replace gh api graphql calls with a native library (e.g., octokit).
  • Support authentication via a GITHUB_TOKEN environment variable or direct configuration.

Benefits

  • Improved portability (works in environm

Read the thread · 2026-05-15 · closed · 1 comment

Support Additional MCP Commands for Non–Beginner-Friendly Issue Discovery

Currently, this tool focuses on discovering beginner‑friendly issues in the selected repository that are not already covered by open PRs.
However, contributors often have different backgrounds, strengths, and goals. To support a wider range of workflows, we should consider adding additional MCP commands that allow more flexible and skill‑aware issue discovery.

Potential enhancements include:

  • Accepting the contributor’s skill set as input and recommending issues with a high relevance match

Read the thread · 2026-05-27 · open · 0 comments

refactor: improve error handling with structured content in tool responses

Currently, errors in tool execution are thrown as standard exceptions. In the MCP ecosystem, returning errors within the 'content' of the tool response can help AI agents understand and self-correct the issue (e.g., if a repository name is misspelled).

Refactor 'server.setRequestHandler(CallToolRequestSchema, ...)' to return a structured error message in the 'content' field when appropriate, instead of just throwing.

Read the thread · 2026-05-16 · open · 0 comments

feat: implement MCP Prompts for contribution workflow

To make the 'onramp' experience even smoother, we can implement MCP Prompts. Prompts provide reusable templates that guide the AI agent and the user through a specific task.

Example Prompt: 'find-oss-task'

  • Arguments: 'repository'
  • Content: Instructions for the agent to scout issues, summarize them, and ask the user which one they'd like to help with.

Read the thread · 2026-05-16 · open · 0 comments

feat: implement MCP Resources for repository metadata

Currently, repository information is only accessible through the 'scout_issues' tool. Following the MCP standard, we should implement Resources to allow AI agents to browse repository metadata using URIs.

Proposed URI pattern: 'mcp://github/[owner]/[repo]/metadata'

This will allow agents to see high-level stats or metadata without a full tool execution.

Read the thread · 2026-05-16 · open · 0 comments

Most recent

The remaining reports are on the project's issue tracker.