{"id":50454120,"url":"https://github.com/mizcausevic-dev/mcp-reliability-toolkit","last_synced_at":"2026-06-01T01:05:40.035Z","repository":{"id":357947708,"uuid":"1239264159","full_name":"mizcausevic-dev/mcp-reliability-toolkit","owner":"mizcausevic-dev","description":"MCP server exposing SLO math + reliability config recipes. Compute burn rate, size rate limiters, pick breaker thresholds, get drop-in Python and Rust configs back. Part of the Platform Reliability Stack.","archived":false,"fork":false,"pushed_at":"2026-05-14T23:38:56.000Z","size":47,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-15T01:35:18.164Z","etag":null,"topics":["circuit-breaker","claude","kinetic-gain","mcp","model-context-protocol","rate-limiter","reliability","slo","sre","typescript"],"latest_commit_sha":null,"homepage":"https://kineticgain.com/","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/mizcausevic-dev.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-05-14T23:33:09.000Z","updated_at":"2026-05-14T23:38:57.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/mizcausevic-dev/mcp-reliability-toolkit","commit_stats":null,"previous_names":["mizcausevic-dev/mcp-reliability-toolkit"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/mizcausevic-dev/mcp-reliability-toolkit","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mizcausevic-dev%2Fmcp-reliability-toolkit","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mizcausevic-dev%2Fmcp-reliability-toolkit/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mizcausevic-dev%2Fmcp-reliability-toolkit/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mizcausevic-dev%2Fmcp-reliability-toolkit/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mizcausevic-dev","download_url":"https://codeload.github.com/mizcausevic-dev/mcp-reliability-toolkit/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mizcausevic-dev%2Fmcp-reliability-toolkit/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33755379,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-05-31T02:00:06.040Z","response_time":95,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["circuit-breaker","claude","kinetic-gain","mcp","model-context-protocol","rate-limiter","reliability","slo","sre","typescript"],"created_at":"2026-06-01T01:05:39.913Z","updated_at":"2026-06-01T01:05:40.030Z","avatar_url":"https://github.com/mizcausevic-dev.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# mcp-reliability-toolkit\n\n[![CI](https://github.com/mizcausevic-dev/mcp-reliability-toolkit/actions/workflows/ci.yml/badge.svg)](https://github.com/mizcausevic-dev/mcp-reliability-toolkit/actions/workflows/ci.yml)\n[![Node](https://img.shields.io/badge/node-%3E%3D20-339933)](https://nodejs.org/)\n[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)\n\n**MCP server that exposes the Platform Reliability Stack's math as Claude-callable tools.** Compute SLO burn rate, size a token-bucket rate limiter, pick circuit-breaker thresholds, or generate a whole layered reliability stack — and get drop-in Python + Rust configs back.\n\nPairs with:\n\n- **[slo-budget-tracker](https://github.com/mizcausevic-dev/slo-budget-tracker)** — Python SLO + error-budget library.\n- **[reliability-toolkit-rs](https://github.com/mizcausevic-dev/reliability-toolkit-rs)** — Rust async reliability primitives.\n- **[rate-limit-shield](https://github.com/mizcausevic-dev/rate-limit-shield)** — Python rate-limit + circuit-breaker + retry.\n\n---\n\n## Why\n\nThe Platform Reliability Stack already gives you the math (`slo-budget-tracker`) and the primitives (`reliability-toolkit-rs`). What it didn't give you was the moment in a design conversation where you say \"Claude, given 500 rps and a 99.9 SLO, what should my breaker + bulkhead look like?\" — and have Claude actually compute it instead of vibing.\n\nThis server fills that gap. Every tool is a thin, validated wrapper around pure math from `src/sre_math.ts`. The numbers Claude shows you are the same numbers `slo-budget-tracker` would compute server-side. There's no LLM-in-the-loop math.\n\n---\n\n## Tools\n\n| Tool | What it does |\n| --- | --- |\n| `compute_slo_burn` | Burn rate, error budget remaining, time-to-exhaustion, SRE-workbook alert level — from raw `(target, failures, total, window_seconds)`. Mirrors `slo-budget-tracker.SLOTracker.snapshot()`. |\n| `design_rate_limiter` | Token-bucket sizing from `rps` + `burst_factor`, plus Python (`rate-limit-shield`) and Rust (`reliability-toolkit`) snippets and a JSON config. |\n| `design_circuit_breaker` | Threshold + cool-down + half-open sizing with sanity-check notes. When `protected_slo_target` is supplied it flags settings that look inconsistent with the SLO's error budget. |\n| `compose_reliability_pattern` | Given service name + rps + protected SLO target, returns the canonical layered stack — rate-limit → bulkhead → breaker → retry → SLO tracker — plus a paste-ready Python and Rust config. |\n\nEach tool advertises a JSON Schema. Bad input is rejected with a typed error — no silent coercion.\n\n---\n\n## Install\n\nThe server speaks stdio MCP, the same shape as every Claude Desktop tool.\n\n### Quick start\n\n```bash\nnpm install -g mcp-reliability-toolkit\n```\n\n```jsonc\n// ~/.config/Claude/claude_desktop_config.json\n{\n  \"mcpServers\": {\n    \"reliability-toolkit\": {\n      \"command\": \"mcp-reliability-toolkit\"\n    }\n  }\n}\n```\n\nRestart Claude Desktop. The four tools above will appear under the tools panel.\n\n### Run locally from source\n\n```bash\ngit clone https://github.com/mizcausevic-dev/mcp-reliability-toolkit.git\ncd mcp-reliability-toolkit\nnpm install\nnpm run build\nnode dist/index.js   # speaks MCP stdio\n```\n\nThen wire `dist/index.js` into the same `claude_desktop_config.json` block, using an absolute path.\n\n---\n\n## Example interaction\n\n\u003e *\"My checkout service does 200 rps and we want three nines. Design the stack.\"*\n\nClaude calls `compose_reliability_pattern({ service_name: \"checkout\", rps: 200, protected_slo_target: 0.999 })` and gets back:\n\n```json\n{\n  \"service_name\": \"checkout\",\n  \"layers\": [\n    \"1. RateLimiter — protect the downstream rps budget\",\n    \"2. Bulkhead — cap in-flight concurrency\",\n    \"3. CircuitBreaker — short-circuit when the downstream is unhealthy\",\n    \"4. Retry with full jitter — recover from transient failures only\",\n    \"5. SLO tracker + multi-window burn-rate alerts (1h + 6h) for paging\"\n  ],\n  \"rate_limiter\": {\n    \"rps\": 200, \"burst\": 400, \"refill_interval_ms\": 5,\n    \"bulkhead_capacity\": 400,\n    \"config\": {\n      \"python\": \"from rate_limit_shield import RateLimiter, Bulkhead\\nlimiter = RateLimiter(rps=200, burst=400)\\nbulkhead = Bulkhead(capacity=400)\",\n      \"rust\":   \"use reliability_toolkit::{RateLimiter, Bulkhead};\\nlet limiter = RateLimiter::new(200.0, 400);\\nlet pool    = Bulkhead::new(400);\"\n    }\n  },\n  \"circuit_breaker\": { \"failure_threshold\": 5, \"cool_down_seconds\": 30, ... },\n  \"slo\": { \"target\": 0.999, \"window_seconds\": 2592000 }\n}\n```\n\n…with a paste-ready Python and Rust config block at the end. Claude reads the tool output and explains the recommendation; the math came from a deterministic function, not the model.\n\n---\n\n## Tests\n\n```bash\nnpm install\nnpm run typecheck\nnpm run build\nnpm test\n```\n\nCI matrix runs Node 20 and 22.\n\n---\n\n## Layout\n\n```\nsrc/\n  index.ts        # MCP stdio server entry point\n  tools.ts        # tool registry: zod schemas, JSON-Schema export, handlers\n  sre_math.ts     # pure functions; identical math to slo-budget-tracker\ntests/\n  sre_math.test.ts\n  tools.test.ts\n```\n\nAdding a new tool is one push to the `tools` array in `tools.ts` — zod schema in, handler out, done.\n\n---\n\n## License\n\nMIT. See [LICENSE](LICENSE).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmizcausevic-dev%2Fmcp-reliability-toolkit","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmizcausevic-dev%2Fmcp-reliability-toolkit","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmizcausevic-dev%2Fmcp-reliability-toolkit/lists"}