feat: add Ollama, Enhanced Sandbox, and Sub-agents

[robotlearning123]

Summary

• Ollama Provider: Local LLM support via Ollama's OpenAI-compatible API at localhost:11434
• Enhanced Sandbox: Safe command execution with Docker, Firejail, or ulimit backends
• Sub-agents System: Spawn specialized agents for focused tasks

New Tools

Tool	Description
sandbox	Run commands in isolated environment with resource limits
sandbox_info	Get available sandbox backends info
spawn_agent	Spawn specialized sub-agent for task
list_agents	List available sub-agent types

Sub-agent Types

• research - Explore codebases (glob, grep, read_file, list_dir, web_search)
• code-review - Review code (glob, grep, read_file, diff, git_diff)
• test-writer - Write tests (glob, grep, read_file, write_file, bash)
• refactor - Refactor code (glob, grep, read_file, edit_file, bash)
• documentation - Generate docs (glob, grep, read_file, write_file, edit_file)

Test plan

• Build passes with npm run build
• Test Ollama with nano --provider ollama --model llama3.2
• Test sandbox with sandbox_info tool
• Test sub-agents with list_agents and spawn_agent

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features
  • Added agent-readiness CLI tool for scanning repository maturity across 9 pillars and 5 levels with extensible profiles.
  • Added Ollama support for local language model inference.
  • Added sub-agent system for specialized tasks (research, code review, test writing, refactoring, documentation).
  • Added sandboxed command execution with Docker, Firejail, and direct isolation backends.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[coderabbitai]

Caution

Review failed

The pull request is closed.

📝 Walkthrough

Walkthrough

Introduces a comprehensive Factory-compatible repository readiness scanner as a new module (agent-readiness) with nine pillars and five levels, plus Ollama provider support and sub-agent spawning capabilities with sandbox isolation to the main codebase.

Changes

Cohort / File(s)	Summary
Implementation Plan Updates IMPLEMENTATION_PLAN.md	Updated status markers to reflect completed features across Plan Mode, Web Search, Image Support, Sandbox, Ollama Integration, Sub-agents, and Git Auto-commit.
Agent-Readiness Project Setup agent-readiness/.github/workflows/ci.yml, agent-readiness/package.json, agent-readiness/tsconfig.json, agent-readiness/eslint.config.js, agent-readiness/.prettierrc, agent-readiness/.prettierignore, agent-readiness/.gitignore, agent-readiness/.husky/pre-commit	Project configuration, CI/CD pipeline (lint, typecheck, test, build), and code quality tooling with ESLint, Prettier, Husky, and lint-staged.
Agent-Readiness Documentation agent-readiness/README.md, agent-readiness/LICENSE, agent-readiness/FACTORY_COMPARISON.md	Comprehensive documentation including project overview, Factory-compatible model comparison, and licensing.
Check Implementations agent-readiness/src/checks/file-exists.ts, agent-readiness/src/checks/path-glob.ts, agent-readiness/src/checks/any-of.ts, agent-readiness/src/checks/github-workflow.ts, agent-readiness/src/checks/github-action.ts, agent-readiness/src/checks/build-command.ts, agent-readiness/src/checks/log-framework.ts, agent-readiness/src/checks/dependency-detect.ts, agent-readiness/src/checks/index.ts	Eight distinct check types for validating repository readiness across multiple pillars, with unified dispatcher and composition support.
Engine & Scoring agent-readiness/src/engine/context.ts, agent-readiness/src/engine/level-gate.ts, agent-readiness/src/engine/index.ts	Repository context building, level gating logic with 80% threshold, pillar and level summaries, and overall scoring calculations.
Command Implementations agent-readiness/src/commands/scan.ts, agent-readiness/src/commands/init.ts	Scan and init commands with output handling, validation, and template-based file generation.
Output Formatters agent-readiness/src/output/json.ts, agent-readiness/src/output/markdown.ts	JSON serialization and terminal-friendly Markdown-formatted output with progress visualization and monorepo support.
Profile & Template System agent-readiness/src/profiles/index.ts, agent-readiness/src/profiles/factory_compat.yaml, agent-readiness/src/templates/index.ts, agent-readiness/templates/*	Profile loading from YAML with validation, factory-compatible profile definition, and template system for initialization with eight template files.
Utility Modules agent-readiness/src/utils/fs.ts, agent-readiness/src/utils/git.ts, agent-readiness/src/utils/yaml.ts, agent-readiness/src/utils/exec.ts, agent-readiness/src/utils/regex.ts, agent-readiness/src/utils/lru-cache.ts, agent-readiness/src/utils/index.ts	File system operations with caching, Git utilities, YAML parsing with validation, safe command execution, ReDoS-safe regex utilities, and LRU cache implementation.
Core Type System agent-readiness/src/types.ts, agent-readiness/src/scanner.ts, agent-readiness/src/index.ts	Comprehensive TypeScript types for checks, profiles, results, and context; main scanner orchestrator with monorepo support; CLI entry point with Commander.
Test Suite & Fixtures agent-readiness/test/checks.test.ts, agent-readiness/test/engine.test.ts, agent-readiness/test/security.test.ts, agent-readiness/test/fixtures/*	Comprehensive tests for check execution, level-gating logic, security utilities, and two test fixture repositories (minimal and standard).
Ollama Provider Integration src/providers/ollama.ts, src/providers/index.ts, src/constants.ts, src/types.ts	New OllamaProvider with OpenAI-compatible API, fetch-based streaming, and tool-call support; updated provider factory, constants, and config types.
Sub-Agent & Tool System src/agents/subagent.ts, src/tools/spawn.ts, src/tools/sandbox.ts, src/tools/index.ts	Sub-agent spawning with five predefined types (research, code-review, test-writer, refactor, documentation), sandbox execution tools with Docker/Firejail/direct backends, and agent listing tools.
Sandbox Isolation src/sandbox/index.ts	Multi-backend sandbox runner supporting Docker, Firejail, and direct execution with timeout, memory limits, and network isolation.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~50 minutes

Possibly related PRs

• nano-opencode#10: Adds matching CI/CD configuration (GitHub Actions, ESLint, Prettier, Husky/lint-staged setup) to agent-readiness module, indicating parallel tooling standardization efforts.

Poem

🐰 A rabbit hops through code so grand,
With readiness checks across the land,
From Ollama's wisdom to sandboxes tight,
Sub-agents dancing, pulling all night!
Nine pillars strong, five levels deep—
Quality metrics we're here to keep! ✨

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}