{"id":45575724,"url":"https://github.com/axect/magi-researchers","last_synced_at":"2026-04-02T23:12:11.465Z","repository":{"id":339191254,"uuid":"1160776328","full_name":"Axect/magi-researchers","owner":"Axect","description":"Three AI models, one synthesis — Claude, Gemini \u0026 Codex cross-verify each other for rigorous multi-perspective research","archived":false,"fork":false,"pushed_at":"2026-02-23T07:46:01.000Z","size":263,"stargazers_count":9,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2026-02-23T15:56:36.622Z","etag":null,"topics":["ai-research","claude-code","claude-code-plugin","codex","cross-verification","gemini","mcp","multi-model","research-automation","scientific-computing"],"latest_commit_sha":null,"homepage":null,"language":null,"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/Axect.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","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":"2026-02-18T11:05:28.000Z","updated_at":"2026-02-23T07:46:05.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/Axect/magi-researchers","commit_stats":null,"previous_names":["axect/claude_researchers"],"tags_count":7,"template":false,"template_full_name":null,"purl":"pkg:github/Axect/magi-researchers","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Axect%2Fmagi-researchers","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Axect%2Fmagi-researchers/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Axect%2Fmagi-researchers/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Axect%2Fmagi-researchers/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Axect","download_url":"https://codeload.github.com/Axect/magi-researchers/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Axect%2Fmagi-researchers/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29851221,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-25T22:37:40.667Z","status":"online","status_checked_at":"2026-02-26T02:00:06.774Z","response_time":89,"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":["ai-research","claude-code","claude-code-plugin","codex","cross-verification","gemini","mcp","multi-model","research-automation","scientific-computing"],"created_at":"2026-02-23T09:48:37.726Z","updated_at":"2026-04-02T23:12:11.456Z","avatar_url":"https://github.com/Axect.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cp align=\"center\"\u003e\n  \u003cimg src=\"MAGI-Researchers.png\" width=\"600\" alt=\"MAGI Researchers architecture: Claude, Gemini, and Codex collaborate through cross-verification, adversarial debate, MAGI-in-MAGI hierarchical scaling, and resumable 6-phase pipeline\" /\u003e\n\u003c/p\u003e\n\n\u003ch1 align=\"center\"\u003eMAGI Researchers\u003c/h1\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003cstrong\u003eThree AI models. One synthesis. Zero lost progress.\u003c/strong\u003e\u003cbr/\u003e\n  \u003cem\u003eMulti-model research pipeline for Claude Code — Claude, Gemini, and Codex debate, cross-verify, and synthesize publication-ready artifacts.\u003c/em\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://github.com/Axect/magi-researchers/stargazers\"\u003e\u003cimg src=\"https://img.shields.io/github/stars/Axect/magi-researchers?style=social\" alt=\"GitHub Stars\" /\u003e\u003c/a\u003e\u0026nbsp;\n  \u003ca href=\"https://github.com/Axect/magi-researchers/tags\"\u003e\u003cimg src=\"https://img.shields.io/github/v/tag/Axect/magi-researchers?label=version\" alt=\"Version\" /\u003e\u003c/a\u003e\u0026nbsp;\n  \u003cimg src=\"https://img.shields.io/badge/claude--code-plugin-blueviolet\" alt=\"Claude Code Plugin\" /\u003e\u0026nbsp;\n  \u003cimg src=\"https://img.shields.io/badge/python-3.11%2B-blue\" alt=\"Python 3.11+\" /\u003e\u0026nbsp;\n  \u003cimg src=\"https://img.shields.io/badge/license-MIT-green\" alt=\"License: MIT\" /\u003e\u0026nbsp;\n  \u003ca href=\"https://github.com/Axect/magi-researchers/issues\"\u003e\u003cimg src=\"https://img.shields.io/badge/PRs-welcome-brightgreen\" alt=\"PRs Welcome\" /\u003e\u003c/a\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"#why-magi\"\u003eWhy MAGI?\u003c/a\u003e \u0026bull;\n  \u003ca href=\"#get-started\"\u003eGet Started\u003c/a\u003e \u0026bull;\n  \u003ca href=\"#features\"\u003eFeatures\u003c/a\u003e \u0026bull;\n  \u003ca href=\"#usage\"\u003eUsage\u003c/a\u003e \u0026bull;\n  \u003ca href=\"#roadmap\"\u003eRoadmap\u003c/a\u003e \u0026bull;\n  \u003ca href=\"CHANGELOG.md\"\u003eChangelog\u003c/a\u003e\n\u003c/p\u003e\n\n---\n\n\u003e *Like the MAGI system in Evangelion — three supercomputers cross-verifying each other — this plugin orchestrates Claude, Gemini, and Codex for rigorous, multi-perspective research.*\n\n## Why MAGI?\n\nSingle-model research has blind spots. One model hallucinates a citation or misses a critical constraint — and nobody catches it.\n\n| | **Single Model** | **MAGI (3 Models)** |\n|:---|:---|:---|\n| **Brainstorming** | One perspective | Three independent perspectives |\n| **Verification** | Self-review (unreliable) | Cross-model peer review |\n| **Blind spots** | Undetected | Caught by competing models |\n| **Output** | Raw text | Structured report with consensus \u0026 divergence analysis |\n\n- **Claude (MELCHIOR)** — *The Scientist.* Active third MAGI personality — synthesis, planning, implementation, and original analytical contributions.\n- **Gemini** — *The Critic.* Creative brainstorming, cross-verification, broad knowledge.\n- **Codex** — *The Builder.* Feasibility analysis, code review, implementation focus.\n\n### Case Study: Damped Oscillator Equation Discovery\n\nWe gave all three single models and MAGI the same physics problem: *discover an unknown damping function from noisy sensor data*. No single model proposed combining classical diagnostics with modern ML — only MAGI's cross-verification caught that gap.\n\n| Source | Score | Highlight |\n|:-------|------:|:----------|\n| **MAGI** | **90** | Staged pipeline: rapid diagnostics → symbolic discovery → validation → fallback |\n| Claude | 84 | Best code coverage — runnable snippets for every approach |\n| Codex | 80 | Elegant physics-informed neural ODE constraints |\n| Gemini | 67 | Most accessible for general audience |\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eExperiment details\u003c/strong\u003e\u003c/summary\u003e\n\n- **Task:** Discover $f(\\dot{x})$ in $m\\ddot{x} + f(\\dot{x}) + kx = 0$ from noisy displacement data\n- **Setup:** Identical prompt → 4 sources → anonymized blind evaluation via MAGI\n- **Evaluation:** Two MAGI evaluator personas scored, cross-reviewed, and debated before synthesis\n- **Limitations:** N=1 case study, self-evaluation (MAGI evaluated MAGI), ~7:1 compute ratio\n- **Full report:** [`examples/damped_oscillator_comparison/evaluation_report.md`](examples/damped_oscillator_comparison/evaluation_report.md)\n- **Raw outputs:** [`examples/damped_oscillator_comparison/`](examples/damped_oscillator_comparison/)\n\n\u003c/details\u003e\n\n## Get Started\n\n**Prerequisites:** [Claude Code](https://docs.anthropic.com/en/docs/claude-code) + Python 3.11+ with [uv](https://docs.astral.sh/uv/) + [Gemini CLI](https://github.com/google-gemini/gemini-cli) + [Codex CLI](https://github.com/openai/codex)\n\n**1. Install the plugin** (inside Claude Code):\n```\n/plugin marketplace add Axect/magi-researchers\n/plugin install magi-researchers@magi-researchers-marketplace\n```\n\n**2. Set up MCP servers** (one-time):\n```bash\nclaude mcp add -s user gemini-cli -- npx -y gemini-mcp-tool\nclaude mcp add -s user codex-cli -- npx -y @cexll/codex-mcp-server\nclaude mcp add -s user context7 -- npx -y @upstash/context7-mcp@latest\n```\n\n**3. Run your first research:**\n```\n/magi-researchers:research \"your research topic\" --domain physics\n```\n\nMAGI generates cross-verified hypotheses, writes implementation code, renders publication-quality plots, and synthesizes a structured report — all saved to `outputs/{topic}/`.\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eAlternative: Local Development\u003c/strong\u003e\u003c/summary\u003e\n\n```bash\ngit clone https://github.com/Axect/magi-researchers.git\nclaude --plugin-dir /path/to/magi-researchers\nuv add matplotlib SciencePlots numpy\n```\n\u003c/details\u003e\n\n## Features\n\n### Research Pipeline\n\n| Phase | What Happens | Output |\n|:---|:---|:---|\n| **Brainstorm** | Three models generate and cross-review ideas with expert personas | `brainstorm/` |\n| **Plan** | Concrete research plan with execution metadata, stress-tested by a hostile reviewer | `plan/` |\n| **Implement** | Language-agnostic implementation with dry-run verification and frontmatter update | `src/` |\n| **Execute** | Deterministic code execution from plan frontmatter; generates result artifacts | `results/` |\n| **Test \u0026 Visualize** | Workspace-aware two-tier testing + publication-quality plots | `tests/` + `plots/` |\n| **Report** | Structured report with cross-verified claim-evidence integrity | `report.md` |\n\n### Highlights\n\n- **MAGI-in-MAGI** — `--depth max` scales to N domain specialists, each running a full mini-MAGI brainstorm in parallel with adversarial meta-debate\n- **Adversarial review** — Models debate, cross-verify, and attack each other's plans (murder board) before synthesis\n- **Resume anywhere** — `--resume` picks up from existing artifacts. No state files — your outputs *are* the checkpoints.\n- **Publication-quality output** — `matplotlib` + `scienceplots` (Nature theme), LaTeX math, PNG 300 dpi + vector PDF, structured reports with MAGI traceability\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eAll features\u003c/strong\u003e\u003c/summary\u003e\n\n**Quality Assurance**\n\n- **Holistic \u0026 weighted scoring** — Default expert-judgment ranking; optionally supply explicit JSON weights or `adaptive` prompt-analyzed weights\n- **Dynamic persona casting** — Each model gets a topic-specific expert identity, sharpening ideation\n- **Phase gates** — Automated quality checkpoints with conditional MAGI mini-review before each user approval step\n\n**Resilience**\n\n- **Artifact contracts** — Each phase validates upstream files before running. Catches silent failures before they cascade.\n- **Agent substitution** — `--substitute \"Gemini -\u003e Opus\"` replaces a rate-limited model with Claude across all pipeline stages.\n- **Workspace anchor** — `.workspace.json` locks the output directory path, preventing artifact drift after context compression.\n- **Gemini fallback chain** — Resilient 3-tier model fallback: `gemini-3.1-pro-preview` → `gemini-2.5-pro` → Claude\n\n**More**\n\n- **MAGI traceability review** — All three models cross-verify the final report for orphaned claims and figures\n- **Report gap detection** — Auto-generates missing visualizations from existing data\n- **Domain templates** — Built-in context for Physics, AI/ML, Statistics, Mathematics, and Paper Writing\n- **Journal strategy** — Venue recommendations for [Physics](docs/journal-strategies.md#particle-physics-phenomenology), [AI/ML](docs/journal-strategies.md#aiml-conferences--journals), and [Interdisciplinary](docs/journal-strategies.md#interdisciplinary-science-ml--natural-sciences) research\n\n\u003c/details\u003e\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eUnder the hood\u003c/strong\u003e\u003c/summary\u003e\n\n- **Plot manifest** — Structured `plot_manifest.json` with metadata, section hints, and captions for automated report integration\n- **Common Restrictions** — Phase 4 enforces four output-interface contracts: `plot_manifest.json` (fixed schema), PNG + PDF/SVG dual format, execution evidence, dependency spec file. Internal process is autonomous.\n- **Workspace Detection** — Phase 3 and 4 detect languages and ecosystems from actual `src/` files (package managers first, then file extensions). Priority: reality (`src/`) \u003e plan intent \u003e domain defaults.\n- **Two-tier testing** — Tier 1 (unit, mock-based, always runs) and Tier 2 (integration, depends on `results/`, skipped gracefully if absent). Test frameworks match the detected workspace language.\n- **Deterministic execution** — Phase 3.5 reads `execution_cmd` and `dry_run_cmd` directly from `research_plan.md` YAML frontmatter. No heuristics, no entry-point guessing.\n- **`research_plan.md` frontmatter** — Carries `languages`, `ecosystem`, `execution_cmd`, `dry_run_cmd`, `expected_outputs`, and `estimated_runtime` fields as machine-readable metadata for downstream phases.\n- **Cross-phase artifact contracts** — Each phase validates incoming artifacts before running (tool-based Glob/Read, not LLM guesswork)\n- **Depth-controlled token budget** — `--depth low` skips cross-review for fast/cheap runs; `--depth high` enables full adversarial debate\n- **`@filepath` artifact references** — MCP tool calls use `@filepath` syntax instead of inline content, so large artifacts are read directly from disk with zero truncation\n\n\u003c/details\u003e\n\n## Usage\n\n| Command | Description |\n|:---|:---|\n| `/magi-researchers:research \"topic\"` | Full pipeline (Brainstorm → Plan → Implement → Execute → Test → Report) |\n| `/magi-researchers:research-brainstorm \"topic\"` | Brainstorming with cross-verification |\n| `/magi-researchers:research-write --source \u003cdir\u003e` | Collaborative writing from research artifacts |\n| `/magi-researchers:research-explain \"concept\"` | Concept explanation with Teacher/Critic pipeline |\n| `/magi-researchers:research-implement` | Language-agnostic implementation (needs existing plan) |\n| `/magi-researchers:research-execute` | Execute research code; generate `results/` artifacts |\n| `/magi-researchers:research-test` | Workspace-aware testing \u0026 visualization |\n| `/magi-researchers:research-report` | Report generation |\n\n### Depth Control\n\nThe `--depth` flag controls how thoroughly models review each other's work:\n\n| Depth | What Happens | Cost |\n|:---|:---|:---|\n| `low` | Independent brainstorming, no cross-review | Cheapest |\n| `medium` (default) | Cross-model peer review + synthesis | Standard |\n| `high` | Full adversarial debate (defend/concede/revise) | Higher |\n| `max` | MAGI-in-MAGI: N specialist subagents, each running a full mini-MAGI | Highest |\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eAll flags\u003c/strong\u003e\u003c/summary\u003e\n\n| Flag | Values | Default | Description |\n|:---|:---|:---|:---|\n| `--domain` | `physics` `ai_ml` `statistics` `mathematics` `paper` | auto-inferred | Research domain for context |\n| `--weights` | JSON / `adaptive` | holistic | Scoring mode: omit for expert-judgment ranking, JSON for weighted, `adaptive` for prompt-analyzed |\n| `--depth` | `low` `medium` `high` `max` | `medium` | Review thoroughness |\n| `--personas` | `2`–`5` | `auto` | Number of domain-specialist subagents for `--depth max` |\n| `--resume` | `\u003coutput_dir\u003e` | — | Resume an interrupted pipeline from the last completed phase |\n| `--claude-only` | flag | off | Replace Gemini/Codex with Claude subagents for single-model usage |\n| `--substitute` | `\"Gemini -\u003e Opus\"` `\"Codex -\u003e Opus\"` | — | Replace a specific model with Claude when hitting rate limits |\n\n\u003c/details\u003e\n\n```bash\n# Quick brainstorm with default settings\n/magi-researchers:research \"neural ODE solvers for stiff systems\" --domain physics\n\n# Deep analysis with adversarial debate\n/magi-researchers:research \"causal inference in observational studies\" --domain statistics --depth high\n\n# Resume a crashed session — MAGI picks up where you left off\n/magi-researchers:research \"neural ODE solvers\" --resume outputs/neural_ode_solvers_20260225_v1\n\n# Hierarchical multi-persona analysis (MAGI-in-MAGI)\n/magi-researchers:research \"variational inference for Bayesian deep learning\" --domain ai_ml --depth max --personas 4\n\n# Substitute Gemini with Claude when hitting rate limits\n/magi-researchers:research \"neural ODE solvers\" --domain physics --substitute \"Gemini -\u003e Opus\"\n\n# Fast ideation only (no cross-review, lowest cost)\n/magi-researchers:research-brainstorm \"transformer alternatives for long sequences\" --domain ai_ml --depth low\n```\n\n\u003e If MAGI saves you research time, consider leaving a [star](https://github.com/Axect/magi-researchers/stargazers) so other researchers can find it.\n\n### Output Structure\n\n```\noutputs/{topic_YYYYMMDD_vN}/\n├── .workspace.json   # Workspace anchor (absolute path for artifact safety)\n├── brainstorm/       # Personas, ideas, cross-reviews, debate, synthesis\n├── explain/          # Teacher/Critic analysis, strategy, final explanation\n├── write/            # Intake, outline, draft, review, final document\n├── plan/             # Research plan (with YAML frontmatter), murder board, mitigations, phase gate\n├── src/              # Implementation (any language) + phase gate\n├── results/          # Generated artifacts from Phase 3.5 (data, checkpoints, logs)\n├── tests/            # Test suite (Tier 1 unit + Tier 2 integration) + phase gate\n├── plots/            # PNG + PDF + plot_manifest.json\n└── report.md         # Final structured report\n```\n\nEach phase produces artifacts that double as resume checkpoints — just pass `--resume` to continue from where you left off.\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eFull artifact tree\u003c/strong\u003e\u003c/summary\u003e\n\n```\n.workspace.json               # Workspace anchor (absolute output path)\nbrainstorm/\n├── weights.json              # Scoring weights\n├── personas.md               # Expert personas\n├── gemini_ideas.md           # Gemini brainstorm\n├── codex_ideas.md            # Codex brainstorm\n├── gemini_review_of_codex.md # Cross-review (depth ≥ medium)\n├── codex_review_of_gemini.md # Cross-review (depth ≥ medium)\n├── disagreements.md          # Disagreement summary (depth = high)\n├── debate_round2_gemini.md   # Adversarial debate (depth = high)\n├── debate_round2_codex.md    # Adversarial debate (depth = high)\n└── synthesis.md              # Weighted synthesis\n\nplan/\n├── research_plan.md          # Research plan (YAML frontmatter: languages, execution_cmd, etc.)\n├── murder_board.md           # Plan stress-test\n├── mitigations.md            # Flaw mitigations\n└── phase_gate.md             # Plan quality gate\n\nsrc/\n├── *                         # Research implementation (any language)\n└── phase_gate.md             # Implementation quality gate\n\nresults/\n├── run_log.txt               # Full execution log\n├── pre_execution_status.json  # Structured status (state, error_class, severity, retryable, next_action)\n└── *                         # Generated artifacts (csv, npz, pt, etc.)\n\ntests/\n├── test_*                    # Tier 1 unit tests (mock-based)\n├── test_integration_*        # Tier 2 integration tests (guarded by results/)\n└── phase_gate.md             # Test quality gate\n```\n\n\u003c/details\u003e\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eFull artifact tree — \u003ccode\u003e--depth max\u003c/code\u003e\u003c/strong\u003e\u003c/summary\u003e\n\n```\n.workspace.json               # Workspace anchor (absolute output path)\nbrainstorm/\n├── weights.json              # Scoring weights\n├── personas.md               # N domain-specialist personas\n├── persona_1/                # Persona 1 mini-MAGI output\n│   ├── gemini_ideas.md\n│   ├── codex_ideas.md\n│   ├── gemini_review_of_codex.md\n│   ├── codex_review_of_gemini.md\n│   └── conclusion.md\n├── persona_2/\n│   └── ...                   # (same 5 files per persona)\n├── persona_N/\n│   └── ...\n├── meta_review_gemini.md     # Gemini meta-review of all conclusions\n├── meta_review_codex.md      # Codex meta-review of all conclusions\n├── meta_disagreements.md     # Meta-disagreement summary\n├── meta_debate_gemini.md     # Adversarial debate — Gemini\n├── meta_debate_codex.md      # Adversarial debate — Codex\n└── synthesis.md              # Enriched final synthesis\n```\n\n\u003c/details\u003e\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cstrong\u003eRecommended Permissions\u003c/strong\u003e\u003c/summary\u003e\n\nAdd to `.claude/settings.local.json`:\n\n```json\n{\n  \"permissions\": {\n    \"allow\": [\n      \"Bash(uv:*)\",\n      \"Bash(uv run:*)\",\n      \"Bash(uv run python3:*)\",\n      \"Bash(uv add:*)\",\n      \"Bash(uv sync:*)\",\n      \"Bash(mkdir:*)\",\n      \"mcp__gemini-cli__ask-gemini\",\n      \"mcp__gemini-cli__brainstorm\",\n      \"mcp__codex-cli__ask-codex\",\n      \"mcp__codex-cli__brainstorm\",\n      \"mcp__plugin_context7_context7__resolve-library-id\",\n      \"mcp__plugin_context7_context7__query-docs\"\n    ]\n  }\n}\n```\n\n\u003c/details\u003e\n\n## Roadmap\n\n**Latest — v0.16.0:** Trust the top findings — Mechanism Depth Test catches tautological explanations, Type D convergence detects shared methodological blind spots, Decisive Experiment + tiered timeline overhauls action plans, MELCHIOR Comprehensive Self-Review adds holistic quality gate to synthesis. See [CHANGELOG.md](CHANGELOG.md) for full history.\n\n**Up next:**\n- [ ] Terminal demo GIF — one-command walkthrough\n- [ ] More domain \u0026 journal strategy templates\n- [ ] Ubiquitous Context7 — live doc lookups during testing and report writing\n- [ ] Conditional variance enforcement — smart error-bar policy\n- [ ] Cost estimation — token budget preview before execution\n\n## Contributing\n\nContributions welcome — especially new domain templates. See [CONTRIBUTING.md](CONTRIBUTING.md).\n\n## License\n\n[MIT](LICENSE)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faxect%2Fmagi-researchers","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faxect%2Fmagi-researchers","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faxect%2Fmagi-researchers/lists"}