Mmcp.market

Vouch MCP server

by vouchdev·io.github.vouchdev/vouch·v1.2.2
No reviews yet

Git-native, review-gated knowledge base for LLM agents. Cited retrieval, audited writes.

B71/100grade B
Write a reviewInstall
What users say
No reviews yet
Be the first
Safety scan
B71/100

full report

Adoption
Growing

82 stars23 downloads/wk

Reviews

Write one

Nobody has reviewed Vouch yet.

If you have run it, two minutes of your experience saves the next person an afternoon.

Vouch tools (58, 2 write)

write = sends, deletes, buys or posts

Read from the package source without running it. The installed server may list more.

  • kb_approve

    Approve a proposal → durable artifact. Use carefully.

  • kb_archive
  • kb_audit

    Return the last N audit events, filtered to the viewer scope.

  • kb_capabilities

    Return the protocol capabilities of this server.

  • kb_cite

    Return resolved citations (sources or evidence records) backing a claim.

  • kb_compile

    Compile live approved claims into topic-page proposals.

  • kb_confirm
  • kb_context

    Build a ContextPack ready to inject into an agent prompt.

  • kb_contradict
  • kb_crystallize

    Approve every pending proposal in `session_id` (host must trust the agent).

  • kb_dedup_scan

    Find near-duplicate artifacts via embedding cosine.

  • kb_detect_themes

    Detect recurring entity clusters across completed sessions.

  • kb_digest

    Read-only reviewer briefing: pending proposals oldest-first, recent decisions, stale claims, and followups due.

  • kb_doctor
  • kb_embeddings_stats

    Model identity, per-kind counts, query cache stats.

  • kb_eval_embeddingswrite action

    Run retrieval eval over a JSONL queries file.

  • kb_expire

    Expire stale pending proposals (dry-run unless apply=True).

  • kb_export
  • kb_export_check
  • kb_graph_export

    Render the provenance DAG (or one session's subgraph) as dot/mermaid.

  • kb_impact

    Forward impact: dependents, and breakage if op (archive/contradict/supersede).

  • kb_import_apply
  • kb_import_check
  • kb_index_rebuildwrite action

    Drop and rebuild state.db from the durable files.

  • kb_lint
  • kb_list_claims

    List all claims, optionally filtered by status.

  • kb_list_entities
  • kb_list_pages

    List pages, optionally filtered by kind and frontmatter.

  • kb_list_pending

    List proposals awaiting human review.

  • kb_list_relations

    List all relations; if node_id is given, only edges touching it.

  • kb_list_sources
  • kb_neighbors

    Return graph neighbors of a claim, page, entity, or source.

  • kb_propose_claim

    Propose a new claim. Becomes durable only after `kb_approve`.

  • kb_propose_entity
  • kb_propose_page
  • kb_propose_relation
  • kb_propose_theme

    Propose a theme synthesis page from a detected cluster.

  • kb_provenance_rebuild

    Rebuild the prov_edges cache from durable files; returns edge count.

  • kb_read_claim

    Return a claim with its citation list.

  • kb_read_entity
  • kb_read_page

    Return a page (title, body, claim ids).

  • kb_read_relation
  • kb_register_source

    Register a source. Evidence intake is NOT gated (registering raw evidence is harmless and de-duplicates by content hash).

  • kb_register_source_from_path
  • kb_reindex_embeddings

    Re-encode every artifact under the current embedding adapter.

  • kb_reject
  • kb_reject_extracted

    Mass-reject pending edges the auto-extractor filed (issue #224).

  • kb_search

    Search the KB.

  • kb_session_end
  • kb_session_start
  • kb_source_verify

    Re-hash every source and report any drift.

  • kb_stats

    Observability: pending by agent, review rates, citation coverage.

  • kb_status

    Return KB artifact counts and health summary.

  • kb_supersede
  • kb_synthesize

    Answer a query from approved claims only, with inline `[claim_id]` citations, an explicit gaps block, and a synthesis_confidence grade.

  • kb_trace

    Shortest typed-edge path between two artifacts (or found=false).

  • kb_volunteer_context

    Poll confidence-gated context volunteered for an active session.

  • kb_why

    Backward provenance for a claim: cites, session, supersedes, approval.

Public scan report

scanner v0.1.9 · 2026-09-26 · same rubric, same numbers if you re-run it

2 medium
  • Code scan214 source files scanned15/25
  • –Live reliabilityno gateway calls yet and no remote to proben/a
  • –Tool poisoningtools not inspected (local package is not executed); not countedn/a
  • Auth qualitylocal package, no credentials required12/15
  • Maintenancelast push 34 days ago12/15
  • Maintainer identityregistry namespace matches repository owner7/10

Findings (2)

  • mediumeval / new Function usedexec.eval
    vouch_kb-1.2.2/desktop/src/catalog/methods.json: …p results to retrieve per query for the eval (the k in recall@k)." } ], …
  • mediumsubprocess with shell=Trueexec.shell-true
    vouch_kb-1.2.2/src/vouch/compile.py: …ubprocess.run( llm_cmd, shell=True, cwd=tmp, input=prompt,…
Overall 71/100. Components that don't apply are left out of the denominator. Any critical finding is an F.RubricAppeal a findingJSON

What the publisher says

From the Vouch repository's README, as published. We do not edit it. Read it on GitHub

vouch

Git-native, review-gated knowledge base for LLM agents. MCP server + JSONL tool server + CLI.

<!-- mcp-name: io.github.vouchdev/vouch --> <!-- the mcp registry verifies package ownership by matching this marker against server.json's name; keep it in lockstep with server.json at the repo root. -->

Agents should not start every session with amnesia — but they shouldn't get to write whatever they want either.

vouch gives LLM agents durable memory with an explicit review gate: sessions capture themselves, agents propose writes, and nothing becomes durable knowledge until you approve it. Approved artifacts are plain files under .vouch/ — YAML claims, markdown pages — so the KB lives in your repo, is reviewed like code, diffs cleanly, and travels with git clone.

The destination is the one Andrej Karpathy's llm-wiki idea file sketches: stop using LLMs as search engines that rediscover your documents on every question — use them as tireless knowledge engineers that compile, cross-reference, and maintain a living wiki, while humans curate and think. vouch is that idea with the write path made trustworthy. vouch compile has an LLM draft the topic pages, but every page cites approved claims, every [claim: …] citation is machine-verified before the draft is filed, and the drafts pass through the same review gate as every other write. The LLM compiles; the human approves; the wiki compounds.

"I can do this with one prompt"

Often true — and worth being honest about. If you want your agent to remember things, a paragraph in CLAUDE.md, a memory file, or your host's built-in auto-memory gets you most of the way, costs nothing, and needs no install. Reach for that first. Recall is not a hard problem.

What is hard is trust in the write path, and that's a different problem than memory. Single-writer memory needs no trust model: you're the only author, and a bad line costs you a shrug. The moment writes come from more than one author — several agents, a teammate, a future you who forgot the context — the question stops being "what did we say?" and becomes "who decided this was true, on what evidence, and can I audit it later?" A prompt cannot answer that, no matter how good the prompt is.

That's the whole of vouch:

Shortened. The full README is on GitHub.

Nothing above is checked by us. What we check is on the safety report.

Install directly

claude mcp add vouch -- uvx vouch-kb
Add to Cursor

Vouch: common questions

Is Vouch MCP server safe?
Mostly: it is graded B (71/100). Read the Vouch safety report
How do I install Vouch?
It runs on your machine. Copy the Claude Code, Claude Desktop or Cursor config from the install section.
Does Vouch need an API key?
No secret keys are declared. It reads 1 setting from the environment.
Is Vouch maintained?
The last commit was 36 days ago (2026-08-23). The latest release is v1.2.2.
What can I use instead of Vouch?
Servers from other publishers that do the same job: Git MCP server, Contentrain MCP server and diffctx MCP server. Compare all Vouch alternatives.

Alternatives to Vouch

Same job from other publishers: the closest match first, then the best rated.

All Vouch alternatives →
  • Git
    Comprehensive Git MCP server enabling native git tools including clone, commit, worktree, & more.
    A
  • Contentrain
    Git-native content governance for AI agents — 27 deterministic MCP tools over stdio and HTTP.
    A
  • diffctx
    Selects the minimum code an LLM needs to review a git diff
    A
  • Gnt
    Git-native policy layer for AI agents: check_action verdicts against rules approved via PR.
    A
  • Agent Memory
    Git-native project memory for AI coding agents: Markdown source of truth, reviewable, secret-safe.
    A
Vouch vs GitVouch vs ContentrainVouch vs diffctx

More from vouchdev →