{"id":41729945,"url":"https://github.com/zahardev/aicontext","last_synced_at":"2026-04-23T01:00:40.379Z","repository":{"id":334454091,"uuid":"1141252551","full_name":"zahardev/aicontext","owner":"zahardev","description":"Teach AI to build features, not just write code. A development methodology framework for AI coding assistants — structured planning, spec-driven development, automated reviews and tests. Works with Claude Code, Codex, Cursor, and GitHub Copilot.","archived":false,"fork":false,"pushed_at":"2026-04-13T18:50:20.000Z","size":717,"stargazers_count":2,"open_issues_count":0,"forks_count":1,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-04-13T20:26:45.740Z","etag":null,"topics":["ai-assistant","ai-assisted-development","ai-coding","ai-coding-assistant","ai-development","aicontext","claude-code","cursor-ai","spec-driven-development"],"latest_commit_sha":null,"homepage":"","language":"JavaScript","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/zahardev.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","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":"2026-01-24T14:33:18.000Z","updated_at":"2026-04-13T18:49:20.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/zahardev/aicontext","commit_stats":null,"previous_names":["zahardev/aicontext"],"tags_count":9,"template":false,"template_full_name":null,"purl":"pkg:github/zahardev/aicontext","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zahardev%2Faicontext","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zahardev%2Faicontext/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zahardev%2Faicontext/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zahardev%2Faicontext/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/zahardev","download_url":"https://codeload.github.com/zahardev/aicontext/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/zahardev%2Faicontext/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32158349,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-22T17:06:48.269Z","status":"ssl_error","status_checked_at":"2026-04-22T17:06:19.037Z","response_time":58,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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-assistant","ai-assisted-development","ai-coding","ai-coding-assistant","ai-development","aicontext","claude-code","cursor-ai","spec-driven-development"],"created_at":"2026-01-24T23:27:02.790Z","updated_at":"2026-04-23T01:00:40.328Z","avatar_url":"https://github.com/zahardev.png","language":"JavaScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cp align=\"center\"\u003e\n  \u003cimg src=\"assets/logo4.svg\" alt=\"AIContext\" width=\"500\"\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://www.npmjs.com/package/@zahardev/aicontext\"\u003e\u003cimg src=\"https://img.shields.io/npm/v/@zahardev/aicontext\" alt=\"npm version\"\u003e\u003c/a\u003e\n  \u003ca href=\"https://www.npmjs.com/package/@zahardev/aicontext\"\u003e\u003cimg src=\"https://img.shields.io/npm/dm/@zahardev/aicontext\" alt=\"npm downloads\"\u003e\u003c/a\u003e\n  \u003ca href=\"LICENSE\"\u003e\u003cimg src=\"https://img.shields.io/npm/l/@zahardev/aicontext\" alt=\"license\"\u003e\u003c/a\u003e\n\u003c/p\u003e\n\n**Your AI writes code. AIContext teaches it to build features.**\n\nMost AI coding sessions lose context, skip planning, and need constant hand-holding. AIContext gives your AI assistant a complete development methodology — it interviews you before coding, creates a plan, executes it step by step with automated reviews and tests, and picks up exactly where it left off when you start a new session.\n\n**Works with any language or framework** — PHP, Python, JavaScript, TypeScript, Rust, Go, and more.\n\n**Supports multiple AI tools** — Claude Code, Codex, Cursor, and GitHub Copilot.\n\n## Quick Start\n\n```bash\nnpm install -g @zahardev/aicontext\ncd /path/to/your-project\naicontext init\n```\n\n**Every session must begin with `/start`** (Claude Code) or `use start` (Codex, Cursor, Copilot). This loads your project context, rules, and workflow — without it, none of the other commands will work correctly. On first run, the AI will analyze your codebase and generate project context automatically.\n\nRun `/aic-help` (or `use aic-help`) for a guided tour of available workflows and best practices.\n\n## What Makes AIContext Different\n\n### Not just context files — a Spec Driven Development workflow\n\nWriting a `CLAUDE.md` or `.cursorrules` file gives your AI memory. AIContext gives it a **way of working** — built on [Spec Driven Development](https://martinfowler.com/articles/exploring-gen-ai/sdd-3-tools.html), where the spec is the source of truth and code is derived from it:\n\n```text\n/start-feature  →  Interview  →  Spec + Task(s)\n                                      ↓\n                                /run-task  →  Implement + Review + Test (automated per step)\n                                      ↓\n                                /finish-task  →  Sync docs, update worklog, handle git\n```\n\n**The AI interviews you** before writing code — exploring your codebase to avoid asking what it can determine itself. It recommends answers based on what it found, walks every dimension breadth-first so nothing is missed, and captures decisions as it goes. You confirm or correct — not explain from scratch.\n\n**The AI executes the plan** — each step is implemented, reviewed, and tested automatically. You supervise rather than drive.\n\n**The AI reviews its own code** — built-in review catches bugs, security issues, and architectural problems before you even look at the diff.\n\n**The AI tests in the browser** — `/web-inspect` opens real pages, checks console errors, interacts with elements, and captures screenshots. No more copy-pasting console errors.\n\n**The AI drives the process** — after every action, the AI tells you what to do next. Finished a step? \"Run `/next-step` to continue.\" Closed a task? \"Spec has more pending tasks — start the next one?\" You never have to guess the next command.\n\n**The AI ships the code** — after finishing a task, the AI can commit, push, create a PR, and run the review-fix loop automatically. Configure once, and the full pipeline runs hands-free on every task.\n\n**The AI adapts to your workflow** — on first run, it asks how you like to work: reviews after every step or only at the end? Commit per step or per task? Push automatically? It remembers your answers and never asks again.\n\n**The AI remembers across sessions** — specs, tasks, and task-contexts capture everything. Start a new session, run `/resume-task`, and the AI picks up where it left off. No knowledge is lost.\n\n### Three layers of persistent context\n\n| Layer | What it captures | Example |\n|-------|-----------------|---------|\n| **Spec** | What to build and why — requirements, decisions, non-goals | \"Users can reset passwords via email. Not supporting SMS.\" |\n| **Task** | How to build it — step-by-step plan with checkboxes | \"Step 1: Add reset endpoint. Step 2: Email template. Step 3: Token expiry.\" |\n| **Task-Context** | What the AI learned while building — patterns, gotchas, file references | \"Auth middleware checks token in header, not cookie. See `src/auth.js:42`.\" |\n\nSpecs and tasks are committed to git. Task-context files are gitignored — each developer accumulates their own working knowledge.\n\nLearn more in the [development model guide](docs/development-model.md).\n\n## Key Features\n\n### Structured planning\n- `/start-feature` — thorough discovery interview before any code is written\n- `/create-task` — quick task creation from conversation when a full interview isn't needed\n- `/plan-tasks` — break an existing spec into multiple tasks\n- `/add-idea` — capture a deferred idea to the worklog mid-session so it's not lost\n\n### Automated execution\n- `/run-task` — execute all steps with built-in review and test loops\n- `/run-step` — execute a single step with full control\n- `/do-it` — turn a conversation into a task step and implement it immediately\n\n### Code review\n- `/review` — quick correctness scan (bugs, security, edge cases)\n- `/deep-review` — comprehensive architecture + correctness + codebase health review\n- Specialized reviewer agent runs in parallel without consuming your main conversation (Claude Code)\n\n### Session continuity\n- `/resume-task` — read spec, task-context, and task to resume exactly where you left off\n- `/finish-task` — close out a task: sync spec, write completion notes, handle git\n- `/align-context` — sync all context files with current state\n\n### Issue \u0026 PR workflow\n- `/draft-issue` — draft a GitHub issue from conversation, create it on GitHub, and auto-fill the issue ID in subsequent task filenames\n- `/draft-pr` — generate PR description from task context and git history\n- `/finish-task` can auto-create PRs and run the review-fix loop — configure `after_task.pr` and `after_task.review_loop` in `config.yml` for a fully automated code → commit → push → PR → review → fix pipeline\n- `/gh-review-fix-loop` — automate the review-fix-push cycle (works with CodeRabbit, human reviewers, etc.)\n- `/gh-fix-tests` — fix failing CI checks automatically: diagnose, fix, push, retry until green\n\n### Thinking tools\n- `/interview` — structured discovery on any topic — the AI walks dimensions, recommends answers, and captures decisions\n- `/brainstorm` — generate missing angles, better implementations, and new combinations\n- `/thoughts` — quick \"what do you think?\" check-in for feedback mid-conversation\n\n### Documentation generation\n- `/generate-docs` — generate project documentation from code and existing artifacts (reference, guide, or both)\n\n### Project maintenance\n- `/prepare-release` — generate changelog, update version numbers, and prepare a release commit\n- `/tidy-aic` — archive completed tasks and specs, clean up session artifacts, keep the project directory lean\n\n### Browser inspection\n- `/web-inspect` — open pages, check console errors, interact with elements, capture screenshots\n\n### Safety guardrails\n- Blocks destructive commands, enforces TDD, requires explicit permission before implementation\n- Configurable quality checks: what runs after each step vs after the whole task — adapts to your preferences\n\nSee the [full skills reference](docs/skills.md) for detailed descriptions of all skills.\n\n## How It Works\n\nAIContext creates a `.aicontext/` directory with shared rules, prompts, and templates. Each AI tool gets a thin entry point that loads this shared context:\n\n| Tool | How to invoke skills |\n|------|---------------------|\n| Claude Code | `/skill-name` (e.g., `/start-feature`) |\n| Codex, Cursor, Copilot | `use skill-name` (e.g., `use start-feature`) |\n\nClaude Code gets the richest experience with `/command` skills and parallel subagents. All tools share the same underlying prompts and instructions.\n\n## Installation\n\n### npm (Recommended)\n\n```bash\nnpm install -g @zahardev/aicontext\ncd /path/to/your-project\naicontext init\n```\n\nOr use npx for one-time setup: `npx @zahardev/aicontext init`\n\n### Manual Copy\n\nClone the [GitHub repository](https://github.com/zahardev/aicontext) and copy the files you need:\n\n```bash\ngit clone https://github.com/zahardev/aicontext.git /tmp/aicontext\ncd /path/to/your-project\ncp -r /tmp/aicontext/.aicontext .\ncp -r /tmp/aicontext/.claude .    # Claude Code\ncp -r /tmp/aicontext/.codex .     # Codex\ncp -r /tmp/aicontext/.cursor .    # Cursor\ncp -r /tmp/aicontext/.github .    # GitHub Copilot\nrm -rf /tmp/aicontext\n```\n\n### Updating\n\n```bash\naicontext update    # Update framework files (preserves your project-specific files)\naicontext upgrade   # Upgrade the CLI tool itself\n```\n\n## For Teams\n\n| Committed to git | Gitignored |\n|------------------|------------|\n| Rules, prompts, templates, specs, tasks | Task-contexts, reviews, PR drafts, personal settings |\n\nTeam members share the same rules and task history. Each person's task-contexts and preferences stay local. See [project structure](docs/development-model.md#project-structure) for details.\n\n## Customization\n\nOne config file controls how the AI works — no prompt engineering needed:\n\n- **Project settings**: Edit `.aicontext/config.yml` (review/test/commit behavior, task naming, update checks)\n- **Personal overrides**: Create `.aicontext/config.local.yml` (gitignored, overrides shared settings)\n- **Team rules**: Edit `.aicontext/project.md`\n- **Personal rules**: Edit `.aicontext/local.md` (gitignored)\n- **Remove unused tools**: Delete `.cursor/`, `.codex/`, `.github/`, or `.claude/` as needed\n\n## Learn More\n\n- [Development Model](docs/development-model.md) — how specs, tasks, task-contexts, and quality checks work together\n- [Workflow Guide](docs/workflow.md) — step-by-step guides for common workflows\n- [Skills Reference](docs/skills.md) — detailed descriptions of every skill\n\n## Version History\n\n| Version | Highlights |\n|---------|------------|\n| **1.9.0** | Documentation generation (`/generate-docs`). Type-aware test config with per-type scoping. Self-healing updates and `--force` flag. Choose which AI tools to install. Clearer skill names (`/resume-task`, `/review-task`). |\n| **1.8.0** | PR workflow automation — config guards, review-fix loop, resumable close. Project tidying (`/tidy-aic`). TDD-aware planning. \"Brief\" → \"task-context\" rename. Local version cache. ESM compatibility fix. |\n| **1.7.0** | Adaptive workflow — the AI learns your preferences and stops asking. GitHub issue creation. Thinking tools (`/interview`, `/brainstorm`, `/thoughts`). Automated CI fix (`/gh-fix-tests`). Ideas backlog. Smarter interviews that recommend answers. |\n| **1.6.0** | The big workflow release — three-layer context (spec/task/task-context), structured planning, automated execution with review and test loops, PR automation, browser inspection with `/web-inspect`. |\n| **1.5.0** | Codex support, `/draft-issue`, tool-agnostic PR scripts. |\n| **1.4.0** | Slash command skills, PR workflow scripts, agent model upgrades to sonnet/opus. |\n| **1.3.0** | Claude Code subagents — reviewer, researcher, test-runner working in parallel. |\n| **1.2.0** | Auto-update checking, `aicontext upgrade`, `.ai/` → `.aicontext/` rename. |\n| **1.0.0** | Initial release — rules, prompts, templates, multi-tool support. |\n\nSee [CHANGELOG.md](CHANGELOG.md) for full details.\n\n## License\n\nMIT\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fzahardev%2Faicontext","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fzahardev%2Faicontext","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fzahardev%2Faicontext/lists"}