SecureLLM MCP Server

Create new Architecture Decision Record

Generate ADR from research_agent validation data with credibility scoring

List all ADRs with optional filters by status, project, or classification

Show detailed ADR content including frontmatter and body

Full-text search across all ADRs (title and content)

Show relation graph for an ADR: supersedes, superseded_by, related_to, enables

Validate ADR(s) against governance contracts (required fields, schema compliance)

ADR Runtime Assurance Gate: probe runtime level before any write (ADR-0065)

Show governance configuration: approval matrix, compliance tags, chain config

Get blockchain status: height, tip hash, last ADR, signed count

Full chain verification report: hash chain integrity, signatures, per-block results

Generate Merkle inclusion proof for an ADR (cryptographic proof of existence in chain)

Show provenance trail for an ADR: stages from research through acceptance

Get the latest cryptographic snapshot: hash, merkle root, stats

ADR economics: quality scores, velocity, distribution by project/classification

Show current SBOM (Software Bill of Materials): dependency count, drift status

Accept a proposed ADR: governance check → move file → chain append → merkle rebuild → provenance → snapshot

Mark an ADR as superseded by another, with bidirectional linking

Pre-sign an ADR before acceptance (required for critical classification)

Sign an existing block in the chain (post-acceptance cryptographic signature)

Create a cryptographic snapshot of the current chain state

Regenerate SBOM from current flake.nix dependencies

Launch advanced browser session with Puppeteer

Extract data from web page using CSS selectors

Interact with web forms (fill, click, select, etc.)

Monitor element for changes over time

Aggregate search results from multiple sources

Analyze directories for waste files based on complex criteria

Execute smart cleanup based on previous analysis

Find and resolve duplicate files using hash comparison

Rotate and compress log files based on size and age

Get comprehensive emergency system status (CPU, memory, SWAP, thermal, load)

Emergency abort: Kill all NIX builds and heavy processes

Force CPU cooldown (powersave governor, disable turbo)

Nuclear option: Kill ALL heavy processes (compilers, indexers, etc)

Emergency SWAP cleanup: Free critical SWAP memory

Comprehensive system health check (thermal, CPU, memory, disk, SWAP)

Check if it's safe to run nixos-rebuild (temperature, load, memory)

Analyze directory structure with detailed statistics

Automatically organize files by type, date, size, or custom rules

Create searchable file catalog with SQLite

Search file catalog with filters

Manage file tags for organization and search

Quick thermal check before operation (temperature verification)

Run complete thermal forensics analysis (3-phase: baseline, stress, rebuild)

Launch real-time thermal monitoring war room (continuous display)

Pre-rebuild safety check (thermal + resources + load)

Generate laptop replacement verdict from forensic evidence

Run complete laptop investigation suite (forensics + verdict)

Force CPU to powersave mode (emergency thermal response)

Reset CPU to performance mode (after cooldown)

Scan files for sensitive data patterns (email, phone, SSN, etc.)

Pseudonymize sensitive data using various methods

Encrypt/decrypt sensitive files using SOPS

Audit access to sensitive resources

Execute command on remote server via SSH

Transfer files via SFTP (upload/download)

Run maintenance checks on remote server

Create SSH tunnel (Local, Remote, or Dynamic)

Connect via Jump Host(s)

Manage persistent SSH sessions

Get current MCP server status, feature flags, and runtime health

Get semantic cache and rate limiter statistics

Advanced TS/JS code analysis for debugging: overview, symbol types/refs, callers/callees, impact, entrypoints (low-friction).

List all available UX design specifications for SecureLLM Bridge components

Retrieve a complete UX specification for generating UI components with precise design rules. Use format='prompt' to get an optimized agent prompt.

Generate an optimized AI agent prompt for creating a specific UI component following Bridge design specs precisely

Validate a generated component against Bridge UX specifications — checks colors, fonts, layout, and anti-patterns

Get the complete Bridge Design System reference — colors, typography, spacing, and component patterns

Create a new UX specification from a template for a Bridge component or view

Query Cerebro RAG for grounded answers with source citations. Uses semantic vector search (Jina Code v2 embeddings, 8192-token context) + cross-encoder reranking + LLM generation. Best for codebase questions, architecture exploration, and knowledge retrieval across indexed projects.

Index documents or code snippets into Cerebro's vector store for future RAG queries. Supports batch ingestion with metadata and idempotent upsert by document ID. Content is embedded via Jina Code v2 and stored with HNSW indexing.

Get Cerebro RAG engine status: loaded embedding models, vector store backend (chroma/pgvector/azure), document count, GPU/CPU device info, and service health. Use before ingestion or to diagnose connectivity issues.

Run an ad-hoc latency benchmark against the live Cerebro RAG engine. Reports p50/p95/p99 query latency, throughput (queries/s), min/max/avg. Optionally benchmarks ingest throughput. Use to size GPU/CPU requirements before deployment.

Search for information from external sources. Use 'web' for general queries, 'github' for code/repos/issues, 'research' for deep multi-source investigation.

Analyze code quality and structure. Use 'complexity' for size/complexity metrics, 'dead_code' for unused exports, 'full' for comprehensive multi-pass analysis.

Validate workspace quality. Use 'all' for a full quality report, or target a specific check: lint, test, docs.

Inspect the local system. Use 'health' for overall status, 'disk' for storage analysis, 'network' for connectivity and open ports, 'security' for vulnerability scanning.

Lint code using ESLint (JS/TS) or Ruff (Python)

Format code using Prettier (JS/TS/Web) or Black (Python)

Run tests using Vitest/Jest or Pytest

Manage GitHub Actions CI/CD workflows

Generate API documentation from TypeScript/JavaScript source code. Extracts JSDoc/TSDoc comments and produces structured Markdown or JSON output.

Measure JSDoc/TSDoc documentation coverage across a project. Reports percentage of exported symbols with documentation and identifies undocumented symbols.

Validate Markdown documentation: checks for broken links, missing frontmatter, structural issues, and spelling consistency.

Discover and map the entire ~/master/ project ecosystem. Shows all projects, their roles, languages, dependencies, and cross-project relationships. Use this FIRST when working across projects.

Trace a symbol, API, file, or concept across the entire ecosystem. Shows where it's used, what depends on it, and what it depends on. Essential for understanding change impact across 18 projects.

Cross-project search across all ~/master/ projects. Searches code, docs, and configs simultaneously. Returns context-aware results showing which projects reference a concept and how.

Git forensics: heatmaps, churn analysis, change summaries, uncommitted review with commit suggestions, author stats, and file history (ADR-0006).

Convert between API schema formats: JSON Schema ↔ TypeScript interfaces ↔ OpenAPI fragments. Supports both inline text and file inputs.

Cross-project interoperability analysis: shared dependencies, version alignment, configuration drift detection across multiple projects.

Convert data between formats: JSON ↔ YAML ↔ TOML ↔ CSV. Handles files and inline data with optional output file writing.

Comprehensive knowledge database compaction with tiered storage, summarization, and archival. Reduces database size while preserving critical information.

Generate LLM-powered summary of a knowledge session, preserving key information in compact form.

Detect and merge duplicate knowledge entries using exact matching, fuzzy matching, or semantic embeddings.

Archive old sessions to compressed JSON files, preserving data while freeing database space.

Restore an archived session from compressed file back into the database.

Get the distribution of sessions across storage tiers (hot, warm, cold, frozen).

Pin a session to prevent it from being compacted or archived.

Get history of compaction operations with statistics.