system-brain-mcp
Queries GitHub issues and pull requests to report open backlog items.
Optionally reads Obsidian vaults for backlog and roadmap items.
Detects and cross-checks Railway deployment configurations, reporting canonical target and stray platform configs.
Queries SQLite database schemas and row counts for schema and analytics tools.
Detects Vercel configuration files and flags them as stray or leftover deploy targets.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@system-brain-mcp where does src/api/routes/health.ts deploy?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
system-brain-mcp
Eleven read-only MCP tools for inspecting a software workspace and returning answers tied to observable evidence. Missing or degraded evidence produces a partial result or no recommendation, never invented certainty.
Start
npx -y -p system-brain-mcp system-brain-init /path/to/codeOr from source:
git clone https://github.com/QEbellavita/system-brain-mcp
cd system-brain-mcp
npm install
node bin/init.js /path/to/codeThe initializer scans for repositories and deployment configuration, then writes a draft manifest. It leaves uncertain fields unresolved for human review.
Point an MCP client at the server:
{
"mcpServers": {
"system-brain": {
"command": "npx",
"args": ["-y", "system-brain-mcp"],
"env": {
"SYSTEM_BRAIN_DEPLOY_MANIFEST": "/path/to/deploy-targets.json",
"SYSTEM_BRAIN_FABRICATION_DIRS": "/path/to/source"
}
}
}
}Related MCP server: OPP Observatory MCP
Tools
Tool | Purpose |
| Compare declared and detected deployment evidence |
| Flag narrow placeholder patterns for human verification |
| Summarize accessible open work and local branches |
| Inspect configured local database structure |
| Measure a configured outcome-labeling loop |
| Inventory configured local model artifacts |
| Index configured architecture documents |
| List available reasoning lenses |
| Find open checklist items in configured notes |
| Produce evidence-linked candidate actions |
| Apply a reasoning lens to one candidate action |
All tools are read-only. They do not write to databases, repositories, or deployment targets.
Configuration
Variable | Purpose |
| Path to a reviewed deployment manifest |
| Directories eligible for static inspection |
| Local SQLite database path |
| Directories containing model artifacts |
| Architecture documents to index |
| Optional named vault paths |
| Reviewed analytics configuration |
| Reviewed work taxonomy |
The files under config/ are neutral examples only. Copy them to user-specific files and
replace every example table, label, taxonomy, and threshold before interpreting analytics.
The project intentionally does not ship a production schema or recommended operating
threshold.
brain_analytics measures labels within the configured outcome ledger. Other activity
tables are reported separately so they cannot silently change that denominator. The tool
reports incomplete evidence when a required query fails.
brain_fabrication_audit returns candidate locations, not verdicts. Read the surrounding
implementation before classifying a capability.
Tests
npm testTests use temporary files and injected command behavior. Network access and a database are not required.
Public-data boundary
Public fixtures are synthetic. Do not add customer data, real internal identifiers, production table names, operating thresholds, incident details, or exact production-shaped examples to this repository or its release archives.
Licence
MIT. See LICENSE.
Available Tools
11 toolsbrain_analyticsB
Row counts and recency for known prediction/feedback tables in the local dev SQLite database. Honestly labeled as local/dev, not production; flags feedback famine.
Inputs: no named parameters.
brain_architectureA
Curated architecture docs (gateway skills). No area: list doc names/sections/sizes. With area: return one doc's content (truncated to 8000 chars), matched by name/basename substring.
Inputs: area.
brain_backlogB
Unified open items across GitHub PRs/issues and local git. includeVault (default false) opts into a bounded, most-recent-first Obsidian vault marker-sample ; off by default because even the bounded scan adds iCloud latency.
Inputs: repos, includeVault.
brain_db_schemaA
List tables in the configured local SQLite database, or (with table) inspect one table's columns/indexes/row count. Read-only via the system sqlite3 CLI; table names are validated to prevent injection.
Inputs: table.
brain_fabrication_auditA
Read-only static scan for fabricated-core leaves (Math.random()-fed confidence/accuracy/score/decision values) across configured engine directories. A signal, not proof ; verify each finding against stub-detection-audit before certifying REAL/ABSENT. dirs overrides the configured SYSTEM_BRAIN_FABRICATION_DIRS list for this call.
Inputs: dirs.
brain_lensesA
List available reasoning lenses (mental models / thinker-inspired frames) for framing a decision. kind optionally filters to one lens kind. Output includes an applyGuidance string the calling model is expected to execute: pick the 1-2 lenses that bite hardest on the live problem and apply them, rather than treating the list as reference material.
Inputs: kind.
brain_ml_modelsA
List on-disk ML model artifact files across configured model directories. Presence on disk does NOT mean a model is loaded or active.
Inputs: no named parameters.
brain_recommendA
Rank next steps from the brain's own read-only evidence (analytics, backlog, models; deep: true adds the roadmap scan). A deterministic rule table, NOT a model: every candidate cites the tool and number it came from, and per-source health is reported so degraded evidence cannot read as confirmed support. Output includes a reasoning contract the calling model is expected to execute: rank the candidates against live session context and state what would change its mind.
Inputs: deep.
brain_reframeA
Reframe a real brain_recommend recommendation (looked up by recommendationKey), or a bounded task-only subject when omitted, through a reasoning lens. Priority/evidence/title/why/action/systems/phases are always derived server-side from the real recommendation and its evidence health ; never accepted from the caller. Output includes a reasoning contract (static, in-session instructions) the calling model is expected to execute: actually answer the lens questions against the cited evidence and emit a proceed/reframe/reject verdict.
Inputs: lens, taskContext, recommendationKey.
brain_roadmapA
Scan configured Obsidian vaults for plan/roadmap/next-steps notes (filename-matched, most-recently-modified first) and count their open - [ ] checklist items. Content reads are bounded by limit (default 6, max 12) so an iCloud-hosted vault cannot stall the call.
Inputs: limit.
brain_where_deploysA
Resolve the canonical deploy target for a file/path and cross-check it against repo config.
Inputs: path (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 11 tools
Each tool targets a distinct resource or action: deploys, backlog, DB schema, ML models, analytics, architecture docs, fabrication audit, lenses, roadmap, recommend, and reframe. Descriptions clearly separate related tools like db_schema vs. analytics and backlog vs. roadmap.
All tools share the 'brain_' prefix, creating a clear family, but the suffixes mix nouns (backlog, analytics, architecture), verbs (recommend, reframe), and query-like phrases (where_deploys). This is mostly consistent but not a strict verb_noun pattern.
11 tools is well within the ideal range for a decision-support brain. Each tool covers a discrete capability needed for the apparent domain, and none feel redundant or superfluous.
The tool surface covers the full gather-analyze-recommend-reframe loop using available evidence sources. Minor gaps exist—e.g., no direct GitHub issue/PR detail lookup beyond the backlog aggregate, and no tool for explicit doc searches—but agents can work around these.
Maintenance
Related MCP Connectors
Read-only AI coding tools for change verification, release readiness, capacity, and guidance.
Read-only MCP access to a documented IT fleet: state, changes, posture. 15 tools.
Read-only release evidence for coding agents: findings, deliveries, source and runtime data.
A read-only verified record of agent-operable GTM tools: search, fetch, compare, track changes.
Related MCP Servers
- FlicenseAqualityBmaintenanceProvides coding agents read-only access to Depot's CI failure diagnoses, container build forensics, run history, cache effectiveness, registry contents, and usage data through MCP tools.2866 npm1-
- AlicenseNot gradedqualityBmaintenanceProvides read-only production observability through a small set of tools for health, job status, latest data, approved logs, and deployment metadata, enabling AI planning and review without a production write path.MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI coding agents to analyze repository architecture, changed-file and diff impact, dependency and endpoint exposure, API contract compatibility, architecture boundaries, bounded capacity simulations, and security/quality evidence through MCP tools.3Apache 2.0

PrufGate MCPofficial
AlicenseBqualityBmaintenanceExposes nine read-only tools that let a coding agent audit a repository for fabricated fallbacks, hollow or weakened tests, silently disabled checks and missing authorization scoping. Agents can run the six deterministic engines on changed or whole-repository code and get structured findings back without modifying anything.10177 PyPI1Apache 2.0