Accelerators MCP server

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

• aci_classify_decision

  Classify an AI agent action as APPROVE, ESCALATE, or BLOCK using rule-based analysis, domain scoring, and optional LLM evaluation. Use before any agent action that could affect production systems, finances, or compliance -- this is the primary governance gate. Records an immutable audit entry as a side effect. Returns classification with confidence score and reasoning.

• aci_compliance_check

  Evaluate an action against compliance frameworks (SOC2, GDPR, HIPAA, PCI-DSS). Use before actions that handle sensitive data or affect auditable systems. Returns framework-specific violations, warnings, and remediation steps. No side effects beyond audit logging.

• aci_estimate_blast_radius

  Estimate the blast radius of a proposed action -- affected systems, users, data records, cascade depth, and recovery time. Use before approving high-risk actions to understand potential impact. Returns quantitative impact metrics and affected dependency graph. No side effects -- read-only analysis.

• aci_financial_gate

  Estimate the financial impact of an action and determine if it passes budget gates. Use before any action with cost implications (provisioning resources, purchasing services, scaling infrastructure). Returns cost breakdown, budget impact analysis, and APPROVE/ESCALATE/BLOCK decision. Records audit entry as side effect.

• aci_generate_rollback

  Generate a step-by-step rollback plan for a proposed action. Includes commands, expected durations, prerequisites, and warnings. Use before approving high-risk operations to ensure reversibility. Returns structured rollback plan. No side effects -- generates plan document only.

• aci_governance_health

  Get aggregate governance health metrics for the organization. Use for governance dashboards, compliance reporting, and calibration monitoring. Returns classification distribution, calibration accuracy, false positive/negative rates, and overall health score. No side effects -- read-only metrics query.

• aci_log_decision

  Explicitly record a governance decision in the append-only audit log. Use when decisions are made outside the classify pipeline (e.g., manual human approvals, out-of-band escalations). Returns decision ID and timestamp. Side effect: writes to immutable audit log.

• aci_parallel_review

  Initiate parallel governance review across multiple domain agents (security, financial, legal, ops, compliance, data). Dispatches review tasks via SQS FIFO queue. Use for high-impact decisions that need multi-domain sign-off. Returns a session ID for tracking. Use aci_synthesize_reviews to collect and unify results. Side effect: enqueues review tasks.

• aci_recall_precedent

  Semantic search over past governance decisions to find similar precedents. Use when making governance decisions to check how similar actions were classified historically. Returns top-K precedents with similarity scores and outcomes. No side effects -- read-only search.

• aci_record_outcome

  Record the actual outcome of a previously classified governance decision. Feeds the calibration loop to improve future classification accuracy. Use after an approved action completes (or fails) to close the feedback loop. Side effect: updates the decision record in the audit log.

• aci_route_domain

  Route a governance task to the appropriate domain expert agent(s) with sensitivity tagging. Extends s2t_route_task_to_agent with governance awareness and parallel review recommendations. Use when a governance decision needs domain-specific expertise. Returns routing recommendations. No side effects.

• aci_synthesize_reviews

  Collect and synthesize results from a parallel governance review session. Produces a unified APPROVE/ESCALATE/BLOCK decision from individual reviewer assessments. Use after aci_parallel_review to aggregate domain expert opinions. Returns unified decision with confidence and dissent analysis.

• s2t_acquire_file_lock

  Acquire, release, or check file-based mutex locks with stale lock detection. Use to prevent concurrent file access conflicts in multi-agent environments. Side effects: acquire creates a lock file; release removes it; check is read-only.

• s2t_analyze_error_patterns

  Analyze application error logs to identify recurring patterns, root causes, and trends. Use when debugging production issues, performing post-mortems, or monitoring error spikes. Returns categorized patterns with severity and remediation suggestions. No side effects -- read-only analysis.

• s2t_attempt_auto_recovery

  Match an error against known patterns and suggest automated recovery steps. Use when handling errors programmatically or building self-healing systems. Returns matched patterns with historical success rates and step-by-step recovery instructions. Side effects only when auto_execute=true (disabled by default).

• s2t_catalog

  List all available S2T accelerators with their capabilities, pricing tiers, and current usage information. Use to discover available tools, check tier access, or review pricing. Returns the full accelerator catalog with metadata. No side effects -- read-only query.

• s2t_check_data_lake_readiness

  Assess data lake architecture readiness for production deployment. Evaluates five dimensions: S3 storage configuration, Glue catalog setup, security controls, query performance, and operational tooling. Use before promoting a data lake to production. Returns readiness score (0-100) and prioritized recommendations. No side effects -- read-only assessment.

• s2t_classify_action_risk

  Classify an action's risk level as LOW, MEDIUM, HIGH, or CRITICAL based on blast radius, reversibility, and target environment. Use for AI agent safety guardrails and graduated autonomy decisions. Returns risk classification with scoring breakdown. No side effects -- read-only classification.

• s2t_create_trace_contextwrite action

  Generate W3C Trace Context identifiers (traceparent + tracestate) for distributed tracing across multi-agent workflows. Use when starting new traces or creating child spans within existing traces. Returns W3C-compliant trace headers. No side effects -- generates identifiers only.

• s2t_embed

  Generate vector embeddings for text using Amazon Bedrock Titan models. Supports automatic chunking for long documents. Use when building RAG pipelines, semantic search, document similarity, or knowledge base indexing. Returns embedding vectors with chunk metadata. No side effects -- read-only computation.

• s2t_execute_with_resiliencewrite action

  Configure resilience patterns (retry with exponential backoff, circuit breaker) for an operation. Use when protecting critical operations against transient failures. Returns execution configuration with retry metrics and circuit breaker state. Records resilience configuration as a side effect.

• s2t_generate_cloudformation

  Generate production-ready SAM or CloudFormation templates from natural language descriptions. Creates Lambda functions, API Gateway, DynamoDB tables, and S3 buckets with security best practices. Use when scaffolding new AWS infrastructure. Returns YAML template with resource definitions. No side effects -- generates template text only.

• s2t_generate_dynamodb_design

  Generate optimal DynamoDB single-table designs from entity and access pattern descriptions. Use when designing NoSQL data models or migrating from relational databases. Returns key schema, GSI recommendations, entity mappings, and deployable CloudFormation template. No side effects -- generates design document only.

• s2t_interview_createwrite action

  Create a stakeholder interview session for S2T process discovery. Returns a session token for conducting the interview via s2t_interview_message. Use when beginning a new stakeholder engagement or process assessment. Local tool -- no API key charges, runs entirely on the client. Side effect: creates an in-memory session (expires after 24 hours).

• s2t_interview_list

  List all interview sessions with optional status filter. Use to find active, completed, or expired sessions. Returns session metadata including customer name, stakeholder, and creation time. Local tool -- no API key charges. No side effects -- read-only query.

• s2t_interview_messagewrite action

  Send a message in an active interview session and receive the next adaptive question. Use to conduct stakeholder interviews step by step. Returns the AI-generated follow-up question and session progress. Local tool -- no API key charges.

• s2t_interview_summary

  Generate an AI summary from an interview session's transcript. Includes key findings, pain points, time/cost estimates, and prioritized recommendations. Use after completing an interview to produce a structured deliverable. Local tool -- no API key charges. No side effects beyond reading the session.

• s2t_manage_agent_memory

  Store, retrieve, search, or delete persistent agent memory with namespace isolation. Use for maintaining agent state across sessions, caching intermediate results, or sharing context between agents. Side effects depend on operation: store/delete modify state; retrieve/search are read-only.

• s2t_predict_system_issues

  Predict upcoming system issues by analyzing budget trends, error rates, dependency health, and certificate expiry. Use for proactive operations monitoring. Returns prioritized predictions with severity, projected impact dates, and recommended preventive actions. No side effects -- read-only analysis.

• s2t_route_task_to_agent

  Route a task to the optimal AI agent using semantic similarity matching against 54 specialized agent domains. Use when delegating work across a multi-agent system. Returns ranked agent candidates with confidence scores and capability summaries. No side effects -- read-only routing decision.

• s2t_submit_agent_taskwrite action

  Submit a task to the SQS FIFO queue for asynchronous multi-agent execution. Use when delegating work to background agents or building agent pipelines. Returns task ID and queue position for tracking. Side effect: enqueues a message to SQS.

• s2t_usage

  Get your current API usage statistics and remaining quota for the billing period. Use to monitor consumption, check rate limits, or verify tier status. Returns requests made, tokens used, remaining quota, and current tier. No side effects -- read-only query.

• s2t_validate_cli_readiness

  Validate CLI tool availability and API key health on the current system. Use before running agent workflows that depend on external CLIs (codex, claude, git). Returns readiness status per tool with degradation mode recommendations if tools are missing. No side effects -- read-only system check.

• s2t_validate_iam_policy

  Validate an AWS IAM policy document for security best practices. Detects overly permissive permissions, dangerous action wildcards, and missing resource scoping. Use when reviewing IAM policies before deployment or during security audits. Returns security score (0-100) with categorized findings (CRITICAL/HIGH/MEDIUM/LOW) and least-privilege alternatives. No side effects -- read-only validation.

• s2t_validate_mfa_compliance

  Validate IAM users and root account for MFA compliance. Checks console access, programmatic access, hardware vs virtual MFA, and policy-level MFA conditions. Use during security audits or compliance reviews (SOC2, CIS Benchmarks). Returns compliance score and user-level remediation recommendations. No side effects -- read-only check.

• s2t_validate_oauth

  Validate an OAuth 2.0 configuration and detect common misconfigurations before deployment. Supports Google, Microsoft, GitHub, QuickBooks, and generic OpenID Connect providers. Use when setting up or troubleshooting OAuth integrations. Returns validation errors, warnings, and provider-specific recommendations. No side effects -- read-only validation.

• Code scan14 source files scanned25/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitystatic API keys via environment variables6/15
• Maintenancelast push 192 days ago4/15
• Maintainer identityregistry namespace matches repository owner; GitHub account older than a year8/10

Safety score

66/100

higher than 23% of 1,119 Developer tools servers

Tools

36

the typical Developer tools server has 7

Tools that change things

14%

typical: 5%. Fewer write tools means less an agent can break

Downloads a week

47

more than 35% of 436 Developer tools servers with a package

Accelerators: common questions

Is Accelerators MCP server safe?

With care: it is graded C, so read the findings first (66/100). Read the Accelerators safety report

How do I install Accelerators?

It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.

Does Accelerators need an API key?

Yes. The registry entry asks for S2T_API_KEY.

Is Accelerators maintained?

The last commit was 229 days ago (2026-03-20). The latest release is v1.4.2.

What can I use instead of Accelerators?

Servers from other publishers that do the same job: cloud-audit MCP server, TerraVision MCP server and Observatory MCP server.

How does Accelerators compare with other Developer tools MCP servers?

safety score 66/100 (higher than 23% of 1,119 Developer tools servers); tools 36 (the typical Developer tools server has 7); tools that change things 14% (typical: 5%. Fewer write tools means less an agent can break). Full comparison.