Sprint 4 closed: GitHub Integration
Sprint 4 is closed.
The goal was to connect Meronq to GitHub as the first external engineering system — reading repository metadata, issues, and pull requests, then translating that data into CEM and local memory. GitHub is no longer a separate tab the AI ignores; it becomes part of project understanding.
What Sprint 4 delivered
Every item on the roadmap is checked off:
- Repository metadata — stars, default branch, description, last push
- Issues and pull requests — fetched via GitHub REST API
- CEM translation —
issue:owner/repo#Nandpull_request:owner/repo#Nentities - Issue summaries — title, state, labels, author, excerpt
- PR summaries — draft/merged state, base/head refs, excerpt
- Doc draft — markdown from git log + GitHub snapshot
Local MCP is now v1.13.0 with github_sync, github_status, github_summarize, and github_doc_draft.
New package: @meronq/github
GitHub is the first external translator into CEM:
| Module | Role |
|---|---|
fetchGitHubSnapshot() |
Repo + open issues + open PRs |
appendGithubToCem() |
Merge GitHub entities into local CEM |
buildProjectCem() |
Scan + optional GitHub fetch → full snapshot |
summarizeIssue() / summarizePullRequest() |
Human-readable summaries |
draftProjectDocumentation() |
Project update draft from git + GitHub |
Auth uses GITHUB_TOKEN (Personal Access Token). SSH and GPG are not required for the API — only for git push.
Cache path: .meronq/local/github-snapshot.json
Pipeline: local scan + GitHub → memory
scanProject(root)
└── indexToCemSnapshot()
└── fetchGitHubSnapshot() (github_sync)
└── appendGithubToCem()
└── syncFromSnapshot() → memory.db
Handshake reads cached GitHub data when available and returns a github block (full_name, open_issues, fetched_at).
Example project properties after sync:
{
"github": {
"fullName": "Meronq-dev/meronq",
"openIssues": 0,
"defaultBranch": "main"
}
}
MCP and CLI
meronq github sync
meronq github status
meronq github summarize issue 3
meronq github doc
MCP tools mirror the CLI. See GitHub integration guide for token setup.
Sprint cleanup
Three early GitHub issues (#1–#3) from Sprint 0 planning were closed as completed — foundation, ADR-0002 runtime decision, and scanner/indexing shipped in later sprints. The issue board now reflects actual project state.
Sprint 5 — MCP Server
Sprint 5 goal: expose Meronq capabilities through MCP as a product surface — not only local dogfooding.
Planned work:
- Define MCP tool surface for
apps/mcp-server - Expose handshake, memory search, and runtime tools
- Add safety and permission model
GitHub data now lands in memory; Sprint 5 makes that intelligence available beyond the dev repo.
Follow along
- Roadmap — sprint plan
- GitHub setup — token and tools
- Development blog
- GitHub: Meronq-dev/meronq
Sprint 0 gave us a durable repository. Sprint 1 gave us a shared language. Sprint 2 connected that language to real projects. Sprint 3 made it stick. Sprint 4 connects the outside world.