What can you do with this server?

Solumbe's MCP server is a local-first, deterministic trust layer for AI-assisted development: it builds task-aware repo context, ranks change impact, scores intent-vs-diff convergence, and gates merge readiness — without sending code off the machine. * Inspect & map repos — repo_inspect (languages, package managers, scripts, entrypoints, git metadata) and repo_map (code map filtered by domain, file kind, or controller route; external cache, repo untouched). * Index & search — repo_index builds per-user external indexes and a local catalog (with discover/dryRun), and repo_search queries it by path, domain, kind, route, imports, exports, or symbols. * Build context — context_pack returns primary/related/test files plus validation commands for a task, optionally asking a model (online:true) to read intent. * Rank impact & risk — change_impact ranks files most likely to own a plain-English change, with risk flags, suggested tests, and an implementation plan. * Score agent-friendliness — agent_experience gives a 0–100 AX score blending Changeability, Containment, Guardrails, and Clarity. * Recommend models — model_route advises a cheap/mid/premium tier (advisory only; never feeds the gate). * Score convergence — convergence_score measures intent vs. actual diff (Coverage, Scope, Risk alignment) and emits a recomputable receipt. * Gate merges — review_gate returns PASS/WARN/FAIL for a local base-ref gate or an open GitHub PR (state, reviews, CODEOWNERS, conversations, checks), with policy profiles and optional minConvergence. * Full review pipeline — review_verdict combines impact + context + gate into one verdict with a confidence score. * Review context — review_context returns raw diff metadata, optionally enriched with GitHub PR comments (and can write a sticky comment). * Multi-repo reporting — workspace_report produces a product-level report across related repositories. * Generate harnesses — repo_harness emits setup, validation, runtime, and context commands for agent or CI use.

Which integrations are available for this server?

Reads and reasons over local Git repositories: it gates the exact staged tree against a base ref (e.g. origin/main), resolves tree/parent/base hashes, and binds convergence receipts to the staged-tree identity so a change can be verified before merge.

Solumbe

by BASHBOP

Overview Schema Source Related Servers Discussions Admin

JavaScript

Local

Solumbe's MCP server is a local-first, deterministic trust layer for AI-assisted development: it builds task-aware repo context, ranks change impact, scores intent-vs-diff convergence, and gates merge readiness — without sending code off the machine.

Inspect & map repos — repo_inspect (languages, package managers, scripts, entrypoints, git metadata) and repo_map (code map filtered by domain, file kind, or controller route; external cache, repo untouched).
Index & search — repo_index builds per-user external indexes and a local catalog (with discover/dryRun), and repo_search queries it by path, domain, kind, route, imports, exports, or symbols.
Build context — context_pack returns primary/related/test files plus validation commands for a task, optionally asking a model (online:true) to read intent.
Rank impact & risk — change_impact ranks files most likely to own a plain-English change, with risk flags, suggested tests, and an implementation plan.
Score agent-friendliness — agent_experience gives a 0–100 AX score blending Changeability, Containment, Guardrails, and Clarity.
Recommend models — model_route advises a cheap/mid/premium tier (advisory only; never feeds the gate).
Score convergence — convergence_score measures intent vs. actual diff (Coverage, Scope, Risk alignment) and emits a recomputable receipt.
Gate merges — review_gate returns PASS/WARN/FAIL for a local base-ref gate or an open GitHub PR (state, reviews, CODEOWNERS, conversations, checks), with policy profiles and optional minConvergence.
Full review pipeline — review_verdict combines impact + context + gate into one verdict with a confidence score.
Review context — review_context returns raw diff metadata, optionally enriched with GitHub PR comments (and can write a sticky comment).
Multi-repo reporting — workspace_report produces a product-level report across related repositories.
Generate harnesses — repo_harness emits setup, validation, runtime, and context commands for agent or CI use.

Solumbe

Models generate the change. Solumbe proves whether it is safe to merge.

Formerly Òtítọ́ (@bashbop/otito), renamed in 4.0.0; the CHANGELOG maps every old name to its new one.

solumbe demo

Solumbe MCP server – quality and maintenance score on Glama

Solumbe is a local-first, deterministic, model-agnostic trust layer for AI-assisted development. It builds task-aware repository context before an agent edits, scores how much a change actually touches, and gates merge readiness against the exact staged tree, with no server, no account, and no code leaving the machine.

It does not replace Claude Code, Codex, Cursor, Gemini, or any native agent harness. It runs beside them and keeps working as the models change underneath.

A passing local gate is never an automatic merge approval: hosted CI, GitHub review, CODEOWNERS, and the human release decision remain separate authorities.

Install

npm install -g @bashbop/solumbe
solumbe doctor
solumbe install --host all

The last command connects Solumbe's MCP server to every agent host it finds: Claude Code, Claude Desktop, Codex CLI, Cursor, VS Code, Gemini CLI and Kimi Code CLI. Add --dry-run to see the plan first, or name hosts with --host claude-code,cursor. Restart each host afterwards.

Or without installing: npx -y @bashbop/solumbe doctor.

Related MCP server: Graft

What it does

Rank what a change actually touches, from the request alone, with no model, no embeddings, and no network:

$ solumbe impact . "add refund handling to checkout" --top 3

  concepts: money flow

   1.  src/payment/checkout.service.ts    score 155
       |- role: required
       |- path matches: checkout
       |- symbol matches: checkout, refund
       |- concept match: money flow (×1.4 + 6.0)
       |- risk: money flow

   2.  src/payment/stripe.webhook.ts    score 6
       |- role: advisory
       |- concept match: money flow (×1.4 + 6.0)

Gate the exact staged tree, and say precisely why:

$ solumbe gate . --staged --base origin/main --request "add refund handling to checkout" --min-convergence 80

  [OK]    Changed files          1 changed file found.
  [OK]    Staged snapshot        Changed-file scope and convergence evidence are captured from the exact staged Git tree.
     |- Tree: 7d0b513dd256877d6bd309a6406468a9b1eaeedc
     |- Base: b776f04fcfbd02c8ce8d63e1a0c2450be58de039
  [OK]    Secret safety          No secret file paths and no credential values found in the changed content.
  [WARN]  Risk review            Risk-sensitive files changed; maintainer review should be explicit.
     |- src/payment/checkout.service.ts
  [OK]    Convergence            Task and diff satisfy the convergence requirement.
     |- Score: 100/100 (aligned)
     |- Receipt handle: rcpt_a8d31efae06e
     |- Subject tree: 7d0b513dd256877d6bd309a6406468a9b1eaeedc

  VERDICT  WARN

The convergence score is the part a model cannot grade for itself: it compares the stated intent against the files the diff actually changed, and produces a receipt bound to the exact base, parent, and staged-tree identity.

Hold the change to what was declared. Record the files a request is expected to touch before the edit, and the gate names any file that was never declared:

$ solumbe declare . "add refund handling to checkout"
$ # ... the agent edits ...
$ solumbe gate . --staged --base HEAD --intent .solumbe/intent.json

  [FAIL]  Scope contract         Touched `src/auth/session.ts`, which was never declared. Amend the intent with the reason it belongs (`solumbe amend <file> --reason "<why>"`), or take it out of the change.
     |- Request: "add refund handling to checkout"
     |- Contract: fe655c2af2b31b7e3edec6aedd248b379235811d1228c829f87ff73796ace487
     |- Changed files: 3 (1 declared, 0 amended, 1 implied, 1 undeclared)
     |- Implied: tests/checkout.service.test.ts (owner-test of src/payment/checkout.service.ts)
     |- Undeclared on a risk-sensitive path: src/auth/session.ts [auth/security]

An amendment adds a file with its reason, hashed onto the contract, so a reviewer reads why the scope grew. Over MCP it is change_impact with declare: true, and review_gate with intent.

The core

Four commands are the product. They call no model, open no socket, and read nothing outside the repository.

Step	Command	What it proves
Before the edit	`solumbe context "add refund handling" --path .`	Which files, tests and commands the request needs
After the edit	`solumbe converge . "add refund handling" --base origin/main --staged`	How far the diff is from the request, with a receipt
Before the merge	`solumbe gate . --staged --base origin/main --request "add refund handling"`	PASS / WARN / FAIL on the exact staged tree
After the merge	`solumbe attest . --verdict verdict.json --merge <sha>`	A hash-chained record of what shipped, under which verdict

solumbe help shows these and how to set up; solumbe help --all lists every command, including the ones that support them (impact, declare, pr, review, map, calibrate, route, ax, dashboard and the rest). Nothing was removed.

Nothing hosted is required, and nothing hosted can change a verdict. Local Core, Optional Hosted lists exactly what the optional pieces send, and where.

Supporting commands

Goal	Command
Rank change blast radius	`solumbe impact . "add refund handling" --top 12`
Score intent vs. execution	`solumbe converge "add refunds" --path . --base HEAD --staged`
Score a committed change exactly	`solumbe converge "add refunds" --path . --base HEAD~1 --head HEAD`
Gate a product change across repos	`solumbe workspace-gate ../web ../api --request "ship change"`
Grade risk flags against history	`solumbe calibrate . --window 30`
Grade route tiers against history	`solumbe regret . --window 30 --offline`
Regrade a saved run, same answers	`solumbe regret --rescore run.json`
Score Agent Experience	`solumbe ax . "add a new MCP tool"`
Recommend a model tier	`solumbe route . "add a new MCP tool"`
Sharpen context with a model read	`solumbe context "add a new MCP tool" --path . --online`
Inspect one repo	`solumbe repo . --json`
Build a code map	`solumbe map . --json`
Index and search local projects	`solumbe index ~/projects --discover` then `solumbe search "events controller"`
Generate an agent harness	`solumbe harness . --out .solumbe/harness.md`
Run the MCP server	`solumbe mcp`

Every command takes --json, and solumbe help lists the full set with flags.

MCP

Solumbe ships a stdio MCP server exposing 14 tools: repo_inspect, repo_map, repo_index, repo_search, context_pack, change_impact, agent_experience, model_route, convergence_score, review_context, review_gate, review_verdict, workspace_report, and repo_harness.

{
  "mcpServers": {
    "solumbe": {
      "command": "npx",
      "args": ["-y", "@bashbop/solumbe", "mcp"]
    }
  }
}

solumbe install --host all writes this entry for you, using absolute paths so hosts that do not inherit your shell PATH can still start it. Published in the MCP Registry as io.github.BASHBOP/solumbe. Repo-map lookups use an external per-user cache and never write into the inspected repository. Host-specific setup for Claude Code, Claude Desktop, Codex, Cursor, VS Code, Gemini CLI, and Kimi Code is in MCP and Agent Workflows.

How it compares

Approach	Strengths	Where solumbe differs
Sourcegraph / Cody context	Powerful hosted code search and embedding-based context across an org	Local-first and deterministic: no server, no account, no code leaves the machine, and the same query always yields the same packet
Hand-written `CLAUDE.md` / rules files	Curated, intent-rich guidance	Hand-written context goes stale; solumbe regenerates context from the actual code (symbols, imports, routes, tests) on every run and complements a short `CLAUDE.md`
`grep` / `ripgrep`	Fast, universal text matching	solumbe ranks whole files by task intent across paths, symbols, exports, and tests, then adds patterns and validation commands, producing a context packet rather than a list of matching lines

Documentation

Full command reference, agent workflows, release process, evaluation method, and the design theses behind the trust layer:

bashbop.github.io/solumbe

Contributing

git clone https://github.com/BASHBOP/solumbe.git
cd solumbe && npm ci && npm run ci

npm run ci is the full gate: format, lint, typecheck, version check, tests, coverage floors (70% lines / 60% branches / 75% functions), three evaluation corpora, dependency audit, and a packaged-tarball smoke test. Run it before requesting review.

Start with CONTRIBUTING.md and the Code of Conduct. All changes need maintainer review; main requires passing gates and resolved conversations. Solumbe follows Semantic Versioning, so say whether a PR is no-impact, patch, minor, or major.

Part of the toolchain

solumbe is one of four tools that form a deterministic trust layer for AI-assisted development. Each uses static analysis to answer a question people keep handing to an LLM.

solumbe (this tool), for context: what does this change actually touch?
tieline, for contracts: did the front end and back end quietly stop agreeing?
bouncer, for compliance: could you defend this to Ofcom?
aiglare, for governance: where can the model do something you can't undo?

More at segunolumbe.com. static analysis, never the model.

Available Tools

14 tools

agent_experienceAgent Experience (AX)A

Read-only

Score Agent Experience (AX): a single 0–100 number answering "how cheap and safe is it for an agent to make this change here?", blending Changeability (token cost), Containment (blast radius), Guardrails (tests / validation / CODEOWNERS / CI), and Clarity (task groundedness). Returns sub-scores, drivers, and concrete recommendations for raising AX.

Inputs: query (required), top, path, includeMarkdown.

change_impactChange ImpactA

Read-only

Given a plain-English change request, rank the files most likely to own the change, with risk flags, suggested tests, and an implementation plan. Optional diff base surfaces exact changed-file evidence alongside the heuristic. When the repository's .solumberc.json lists companions, companions also carries the top few files each companion repository would own for the same request.

Inputs: query (required), top, path, amend, +5 more.

context_packContext PackA

Read-only

Generate a task-aware local context packet with primary files, related files, tests, and validation commands. Deterministic by default; online:true also asks a read model for the request's intent and which candidate files it needs, reported under modelRead.

Inputs: query (required), path, limit, paths, +3 more.

convergence_scoreConvergence ScoreA

Score convergence: a deterministic 0–100 measure of the distance between a stated task (intent) and the actual git diff (execution), with sub-scores for Coverage (did the intent happen?), Scope (did only the intent happen?), and Risk alignment (did unrequested drift land on risk-sensitive paths?). No model. A task that predicts no owner file is band inconclusive, with no score. Emits a recomputable receipt (a hash anyone can regenerate). Requires a base git ref to diff against; scores the working tree's tracked changes unless head or staged selects an exact subject.

Inputs: base (required), query (required), top, head, +5 more.

model_routeModel RouteA

Read-only

Recommend a model tier (cheap, mid or premium) for a coding request before any work starts. Combines solumbe's deterministic repository half (AX, containment, top-severity risk paths) with a System One model's calibrated read of the request (specificity, blast radius, novelty). The model is called only when TYPESAFE_API_KEY is set in this server's environment and offline is not true; otherwise the read is a labelled offline estimate, and model.source says which. Advisory: it recommends a tier and never feeds review_gate or review_verdict. Pass host to map the tier to that host's model id.

Inputs: query (required), top, host, path, +2 more.

repo_harnessRepository HarnessA

Read-only

Generate setup, validation, runtime, and context commands for an agent or CI harness.

Inputs: path, includeMarkdown.

repo_indexIndex RepositoriesA

Index local repositories: generate per-user external indexes and add them to the local catalog. Pass discover:true to find repository roots under the given paths first. Pass dryRun:true to discover and report without writing any indexes or catalog (read-only). Without dryRun this mutates the persistent local catalog, so it is not a pure read.

Inputs: path, depth, limit, paths, +3 more.

repo_inspectInspect RepositoryA

Read-only

Inspect repository shape, languages, package managers, script names, entrypoints, and git metadata. Returns up to 200 representative file paths; pass includeScripts:true to get full script command bodies instead of just names.

Inputs: path, includeScripts.

repo_mapMap RepositoryA

Read-only

Map a repository into a compact JSON code map, optionally narrowed by domain, file kind, or controller route. Pass domain to find files for a feature, kind to find files of a type (route, controller, service, apiClient, component, test), or route to match Nest controller routes by substring/regex.

Inputs: kind, path, limit, route, +2 more.

repo_searchSearch Repository CatalogA

Read-only

Search indexed local repositories by path, domain, kind, route, imports, exports, and symbols. Omit query to list the local repository catalog instead. Searches the repository at path (the working directory by default) only; pass paths to search those repositories, or all: true to add every repository in the local catalog (call repo_index to add one). Outside a repository it searches the catalog.

Inputs: all, path, limit, paths, +3 more.

review_contextReview ContextA

Produce PR review context from local git diff metadata, optionally enriched with GitHub PR comments. Use review_context when you want the raw diff/comment context for a change, not a verdict. Use review_verdict instead for the full impact + review + gate composite, or review_gate for the gate verdict alone.

Inputs: base, head, path, github, +3 more.

review_gateReview GateA

Gate a change for merge and return a PASS / WARN / FAIL verdict. Omit pr to run the local, no-GitHub gate against a base ref (changed files, secret safety, risk-sensitive paths, release discipline, validation commands, dependency audit, policy profile). Set pr to gate an open GitHub PR via gh (PR state, review decision, CODEOWNERS approvals, unresolved conversations, branch protection, status checks). A check the mode cannot evaluate is SKIPPED and never changes the verdict. Use review_verdict instead when you want the full impact + review + gate composite, or review_context when you want diff/comment context with no verdict.

Inputs: pr, base, head, path, +7 more.

review_verdictReview VerdictA

Read-only

Run the full review pipeline in one shot: change_impact plus review_context plus review_gate, returning a unified verdict with a derived confidence score. When the repository's .solumberc.json lists companions and a request is given, impactSummary.companions names the top few files each companion repository would own for that request. Use review_verdict when you want the complete picture of a change in a single call. Use review_context instead for diff metadata only (no verdict), or review_gate for the gate verdict alone.

Inputs: pr, base, head, path, +7 more.

workspace_reportWorkspace ReportB

Read-only

Generate a product-level report across multiple related repositories.

Inputs: paths (required), includeMarkdown.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

Deploy Server

Claude Code, inside the Godot 4 editorFlockbay puts an AI agent in the real Godot 4 editor: it writes GDScript, runs your game and fixes what breaks. Free, or bring your own Claude Code or Codex.flockbay.comAd

TDQS

A3.7/5.0

Scored 2026-10-31 22:15 across 14 tools

Disambiguation3/5

The descriptions do heavy lifting with explicit cross-references (review_verdict = change_impact + review_context + review_gate; use review_context for diff only, etc.), but real overlaps remain: change_impact and context_pack both rank files for a task, repo_map and repo_search both narrow by domain/kind/route, and repo_inspect versus repo_map both profile repository shape. An agent can usually sort it out from the text, but several boundaries require reading before choosing.

Naming Consistency4/5

Everything is snake_case and grouped with sensible prefixes (repo_*, review_*), which keeps it predictable. However the internal pattern mixes verb_noun (repo_inspect, review_gate) with noun_noun (convergence_score, agent_experience, workspace_report) and a noun_verb outlier (context_pack), so naming is not a clean single convention.

Tool Count4/5

14 tools sits comfortably in the well-scoped range and each tool maps to a recognizable step in the repo-analysis/review lifecycle. It is arguably one or two tools heavier than needed given the composite tools already bundle sub-capabilities, but nothing feels gratuitous.

Completeness4/5

The surface covers the full arc: index, inspect, map, search, context packing, impact, gating, verdict, convergence scoring, model routing, and multi-repo reporting, plus a harness generator. Missing pieces are minor for an analysis-focused server (no direct file-read or doc/comment-fetch tool), and these are workarounds rather than blockers.

Maintenance

ActivityActive

ResponsivenessNo recent community issues

Resources

Related MCP Connectors

ctxA
io.github.vanshulgoyal101
Turn a GitHub repo or docs site into agent-ready context: pack it or search it, over MCP.
RAG Systems Search Documentation Access
611
https://ctx.vanshul.com/mcp
Graphify
io.github.Graphify-Labs
Shared context for your codebase: a hosted code knowledge graph over MCP. Free plan, OAuth sign-up.
Code Analysis Knowledge & Memory
1
https://api.graphify.com/mcp
Spiderbrain
ai.spiderbrain
Deterministic context layer for your codebase: change impact, blast radius, answers with receipts.
Code Analysis Developer Tools RAG Systems
13
https://mcp.spiderbrain.ai/mcp
ContextStream MCP Server
io.github.contextstream
The context platform for AI agents: code search, team decisions, and grounded project context.
Knowledge & Memory Code Analysis RAG Systems
146
https://mcp.contextstream.io/mcp

Related MCP Servers

Sverklo
Code Analysis Developer Tools Knowledge & Memory
sverklo
A
license
A
quality
C
maintenance
Local-first code intelligence MCP server with hybrid BM25 + ONNX vector search, symbol-level impact analysis, diff-aware PR review with risk scoring, and persistent memory tied to git state.
Updated 3 months ago (2026-08-12 15:48 UTC)
36
314 npm
82
MIT
Graft
Code Analysis Developer Tools
amaar-mc
A
license
B
quality
D
maintenance
Local-first codebase context engine that parses code into a ranked dependency graph and serves it to AI tools via MCP for deep structural understanding.
Updated 8 months ago (2026-03-28 20:18 UTC)
5
10 npm
1
MIT
itaca
Developer Tools Code Analysis Coding Agents
Lukapetro
A
license
Not graded
quality
C
maintenance
Local-first MCP server that indexes all projects on your machine and provides coding agents with briefings about each project's stack, services, and run instructions, so they can start sessions with context across all repos. No cloud, no accounts, no telemetry.
Updated a month ago (2026-10-16 16:01 UTC)
6 npm
MIT
Repo Context Forge
Code Analysis Version Control Developer Tools
negativexq
A
license
A
quality
C
maintenance
MCP server that inspects local codebases and generates source-grounded context packs for coding agents, with tools for repository search, symbols, dependencies, Git analysis, and context generation.
Updated 3 months ago (2026-08-04 23:51 UTC)
6
MIT

Solumbe

Install

What it does

The core

Supporting commands

MCP

How it compares

Documentation

Contributing

Part of the toolchain

Available Tools

Tool Schema Changelog

TDQS

Maintenance

Resources

Related MCP Connectors

Related MCP Servers

Sverklo

Graft

itaca

Repo Context Forge