slateVault is a local-first markdown vault with Git sync, a global wiki, docs health signals, and a built-in MCP server so AI coding agents can start from the same context as your team.
slateVault workflowKeep documentation in markdown, review it like code, and make it available to coding agents through MCP.
Specs, ADRs, guides, runbooks, todos, and wiki rules stay portable as markdown.
Use Git-backed pull, commit, and push flows to share trusted documentation.
Generate briefs and context bundles so agents begin with current project memory.
Features
01 Vault
Keep specs, decisions, runbooks, guides, todos, and context notes in a local-first markdown vault that can be committed, reviewed, and shared like code.
02 Agents
slateVault exposes your docs through MCP so tools like Codex and Claude Code can search, read, summarize, and propose documentation updates from the same trusted source.
03 Sync
Use Git as the collaboration layer. Review local docs, pull shared updates safely, commit changes, and keep wiki docs moving with the rest of the vault.
04 Health
Docs Health highlights aging files, draft-heavy projects, AI-authored docs, and protected sources of truth so the vault stays reliable over time.
MCP included
Agents can ask slateVault for canonical context, recent changes, stale docs, branch summaries, and focused bundles before they touch your codebase. That keeps implementation work closer to the decisions your team already made.
$ slatevault-mcp
26 tools exposed for project docs,
wiki, context, assets, and Git sync.bootstrap_projectCreate a project, source folder link, and initial context in one step
create_projectCreate a new project folder with project.toml
list_projectsReturn all projects with metadata and doc counts
get_project_contextReturn configured AI context files for a project
write_documentCreate or overwrite a markdown document
append_to_docAppend content without overwriting
read_documentRead a document by project and path
list_documentsList documents with optional tag filters
search_documentsFull-text search across the vault or project
propose_doc_updateCreate a reviewable branch update for protected or canonical docs
convert_to_specTurn a scratchpad note into a structured spec
list_wiki_docsList vault-wide wiki documents
read_wiki_docRead a wiki document
search_wikiSearch vault-wide standards, rules, and guidance
get_agent_rulesRead wiki/ai-agent-rules.md
generate_agent_briefBuild a structured briefing for a project
build_context_bundleSearch, rank, and concatenate relevant docs
get_canonical_contextReturn source-of-truth documents
detect_stale_docsFind docs that have not been updated recently
get_recent_changesSummarize recent vault documentation changes
list_project_assetsList non-markdown project assets
read_assetRead text assets or report binary metadata
get_project_source_folderReturn the linked local source folder
set_project_source_folderSet or clear a project's source folder
summarize_branch_diffSummarize changes between two branches
git_commitCommit staged vault changes
git_pushPush committed vault changes to origin
Downloads
Release links are placeholders for now. Drop in the final GitHub release URLs when the artifacts are uploaded.
FAQ
No. It is local-first. Your docs are markdown files on disk, and team sharing is designed around Git-backed vaults.
No. The vault, editor, wiki, docs health, and sync workflows work without AI. MCP and assistant features become more useful when connected to an agent or model.
slateVault is opinionated for software projects: source-of-truth docs, project folders, wiki rules, Git sync, agent context bundles, and review signals.