Show HN: Kodiqa Agent

Kodiqa Solutions released Kodiqa Agent, an open-source AI coding agent that runs locally for free with Ollama or via seven cloud API providers including Claude, OpenAI, and DeepSeek. The agent offers 26 built-in tools, multi-model consensus, cross-provider failover, and lazy MCP tools that use approximately 94 percent fewer tokens than competitors. Kodiqa Agent is available under AGPL-3.0 and can be installed via pip.

The AI coding agent that runs anywhere — free locally with Ollama, or supercharged by 7 cloud APIs. One agent, every model, zero limits. 78 slash commands • 26 tools • lazy MCP tools • RAG search • custom personas • plugins • sub-agents • LSP • 5 themes | Feature | Kodiqa | Claude Code | Aider | Gemini CLI | OpenCode | |---|---|---|---|---|---| Price | Free Ollama or pay-per-token | $20/mo Pro or pay-per-token | Pay-per-token only | Free Gemini Flash | Pay-per-token only | Local/offline | Yes Ollama | No | No | No | Yes Ollama | API providers | 7 Ollama, Claude, OpenAI, DeepSeek, Groq, Mistral, Qwen | 1 Claude | 10+ OpenAI, Claude, etc. | 1 Gemini | 75+ OpenAI, Claude, Gemini, Ollama, etc. | Tools | 26 built-in | ~15 built-in | ~10 built-in | ~12 built-in | ~12 built-in | MCP support | Yes local + remote + OAuth | Yes | No | Yes | Yes | OpenAPI / GraphQL as tools | Yes | No | No | No | No | Lazy MCP tools token-efficient | Yes ~94% fewer tokens | No | No | No | No | Multi-model | Yes consensus mode | No | No | No | No | Cross-provider failover | Yes | No | No | No | No | Plan mode | Yes | Yes | No | No | No | Permission modes | 3 default/relaxed/auto | 2 normal/auto | 1 confirm all | 2 normal/sandbox | 2 normal/auto | Batch edit review | Yes per-file accept/reject | No | No | No | No | Auto model discovery | Yes live from APIs | No | No | No | No | Budget limit | Yes /budget | No | No | No | No | Auto-lint | Yes /lint | No | Yes built-in | No | No | Auto git commit | Yes /autocommit | Yes | Yes default | No | No | Undo | Yes 10 levels per file | No | Yes git-based | No | No | Conversation branching | Yes /branch | No | No | No | No | Context management | Auto-compact at 85% | Auto-compact | Repo map | Auto 1M context | LSP-based | Web search | Yes 3 engines | No | No | Yes Google | No | Persistent memory | Yes SQLite | Yes CLAUDE.md | No | Yes Gemini memory | No | Tab autocomplete | Yes | Yes | No | Yes | Yes | Thinking display | Yes spinner + summary | Yes | No | Yes | Yes | Project indexing | Yes symbol extraction | Yes | Yes repo map | No | Yes LSP | Session recovery | Yes auto-save | Yes | No | No | Yes multi-session | Custom agents | Yes sub-agents | No | No | No | Yes | Desktop app / IDE | Editor bridge --serve | Yes VS Code | No | No | Yes VS Code, desktop | Install | pip install kodiqa | npm install -g | pip install | npm install -g | go install / npm | Language | Python | TypeScript | Python | TypeScript | Go | Tests | 473 | Yes | Yes | Yes | Yes | Open source | Yes AGPL-3.0 | Yes Apache-2.0 | Yes Apache-2.0 | Yes Apache-2.0 | Yes MIT | Kodiqa's unique advantages : free local models, 7 API providers, multi-model consensus, custom plugins, sub-agents, LSP integration, 5 themes, project templates, batch edit review, conversation branching, budget limits, auto-lint, and auto model discovery — features no other agent offers together. pip install kodiqa kodiqa Or from source: git clone https://github.com/Kodiqa-Solutions/Kodiqa-agent.git cd Kodiqa-agent pip install . kodiqa Claude Code-style UI — ❯ prompt with separator line prompt toolkit , arrow-key navigation for all prompts 26 tools — file ops, git, search, web, memory, clipboard, multi-edit, undo, diff apply 7 API providers — Ollama local/free , Claude, OpenAI, DeepSeek, Groq, Mistral, Qwen Editor/IDE bridge — kodiqa --serve exposes a local HTTP API /ask , /diagnostics for VS Code/Zed/Neovim extensions Cross-provider failover — if a provider is down/rate-limited, the turn auto-retries on the next configured provider and continues /failover TOON output — /toon re-encodes JSON tool results into a compact tabular form ~60% fewer tokens on large arrays Custom commands — drop .kodiqa/commands/<name .md and run it as /<name with $ARGUMENTS / $1 substitution ; /commands lists them MCP server support — connect external tool servers via Model Context Protocol Lazy MCP tools — large MCP servers are discovered on demand mcp search / mcp call instead of injecting every tool schema each turn — ~94% fewer tool-schema tokens /mcp lazy Auto model discovery — new Claude/Qwen models appear automatically from APIs Interactive pickers — /model and /key show numbered menus, navigate with arrows Tab autocomplete — slash commands, model names, file paths prompt toolkit Compact streaming — hides code output, shows progress instead toggle with /verbose Stream interrupt — press Esc or Ctrl+C to stop any response instantly Stream stall indicator — animated spinner when response pauses so you know it's still working Thinking display — shows spinner for <think reasoning blocks, line count summary Multi-model consensus — query all models, merge best answers 3 permission modes — default confirm all , relaxed auto file ops , auto no confirms Plan mode — AI explores + plans, you approve, then it implements Batch edit review — queue edits, accept/reject per file with arrow keys Context window management — warns at 70%, auto-compacts at 85%, visual progress bar Conversation branching — save/switch between conversation states Token tracking — cost per response, session totals, tok/s speed Prompt caching — Claude API cache for faster + cheaper responses Auto-retry — exponential backoff on API errors 429, 5xx, timeouts Undo / redo / rewind — per-file undo up to 10 levels with /redo , plus /rewind to revert ALL file changes from the last turn s Checkpoints — save/restore conversation state Session export — export conversation to markdown Git-aware context — auto-detects git repo, includes diff stats Project indexing — symbol extraction def/class/function , cached Shell env detection — auto-detects OS, shell, dev tools Diff preview — colored diff before every file write/edit Parallel tools — read-only operations run concurrently Session summary — auto-saves context summary on quit, loaded on next start Conversation recovery — auto-saved sessions, resume on crash or with kodiqa -c / --resume Workspace boundary — asks permission before accessing files outside working directory Smart Ollama lifecycle — starts on launch, stops when switching to cloud, restarts on local switch Dynamic model library — fetches available Ollama models from ollama.com with pull counts Unlimited iterations — no artificial cap, AI keeps working until the task is done Live API model routing — auto-discovered models from Claude/Qwen APIs routed to correct provider Auto git commit — toggle with /autocommit , auto-commits after AI edits with descriptive message— per-project file exclusion like .kodiqaignore .gitignore for scans/searches Budget limit — /budget 5 sets $5 session limit, warns at 80%, blocks at 100% Auto-lint — /lint ruff check --fix runs linter after edits, feeds errors back to AI Custom personas — /persona switches AI expertise security-expert, code-reviewer, teacher, architect, debugger RAG search — /embed indexes codebase, /rag searches with AI-enhanced context Test generation — /test <file auto-generates unit tests Git diff review — /review-local AI reviews staged changes Interactive debugger — /debug <script runs, catches errors, debugs with AI Diagram generation — /diagram generates Mermaid diagrams via AI File watcher — /watch <path monitors for changes Config profiles — /profile save/load manages settings presets Multi-file refactoring — /refactor rename/extract across project Session history — /history browses and resumes past sessions Clipboard patches — /patch applies diffs from clipboard Changelog — /changelog shows version history Session stats — /stats shows metrics files, tools, time, cost 317 tests — pytest test suite, all passing All interactive prompts use arrow keys — no typing letters: Allow: Write file: ~/project/app.py ❯ Yes Yes, don't ask again — for this action type No Navigate with ↑↓ arrows or j/k , press Enter to select, or 1/2/3 to jump. Prompt uses a separator line like Claude Code : ──────────────────────────────────────── ❯ your prompt here | Command | What it does | |---|---| /model <name | Switch model interactive picker if no arg | /models | List all available models with live API discovery | /multi <models | Multi-model consensus mode | /single | Back to single model | /scan path | Scan project into context with symbol extraction | /clear | Clear conversation history | /compact | Summarize conversation to save context | /memories | Show stored memories | /forget <id | Delete a memory | /context | Show project context file | /key provider | Add/update API key interactive picker if no arg | /tokens | Session token usage, cost, context bar | /config | Show config / /config reload to reload | /export | Export session to markdown file | /checkpoint n | Save conversation checkpoint | /restore n | Restore checkpoint no arg = list all | /env | Show detected shell environment | /verbose | Toggle compact/verbose streaming | /mode mode | Set permission mode default/relaxed/auto | /plan | Toggle plan mode explore → approve → implement | /accept | Toggle batch edit review | /search <engine | Switch search engine duckduckgo/google/api | /cd <path | Change working directory | /branch | Save/switch/list conversation branches | /mcp | Manage MCP tool servers add/remove/list | /autocommit | Toggle auto git commit after AI edits | /budget <amount | Set session budget limit warns 80%, blocks 100% | /undo path | Undo last edit / list undo history | /redo path | Re-apply an undone edit / list redo history | /rewind n | Revert all file changes from the last n turns default 1 | /diff args | Show git diff supports --staged etc. | /lint <cmd | Auto-lint after edits /lint off to disable | /toon on|off | Compact JSON tool results into TOON saves tokens | /pin <path | Pin file to always include in context | /unpin <path | Remove pinned file | /alias <name <cmd | Create command alias | /commands | List custom prompt-template commands .kodiqa/commands/ .md | /unalias <name | Remove command alias | /notify | Toggle desktop notifications for long tasks | /optimizer | Toggle cost optimizer tips | /theme <name | Switch UI theme dark/light/dracula/monokai/nord | /share | Export session as styled HTML | /pr title | Create GitHub PR via gh CLI | /review number | Review PR diff via gh CLI | /issue number | View GitHub issue via gh CLI | /init template | Scaffold project from template | /plugins | List/reload custom tool plugins | /agent <task | Spawn sub-agent for background task | /agents | List running/completed sub-agents | /lsp start|stop | Start/stop Language Server Protocol | /voice | Voice input via sox + Whisper | /changelog | Show version history | /stats | Session metrics files, tools, time, cost | /review-local | AI review of staged git changes | /test <file | Generate unit tests for a file | /persona <name | Switch AI persona security-expert, code-reviewer, etc. | /patch | Apply diff/patch from clipboard | /profile | Save/load config profiles | /refactor | Multi-file refactoring rename, extract | /history | Browse and resume past sessions | /watch <path | Watch files for changes | /embed path | Index files for RAG search | /rag <query | RAG search + AI answer | /debug <script | Run script, catch errors, debug with AI | /diagram <desc | Generate Mermaid diagram | /help | Show help | /quit | Exit | | Mode | Behavior | |---|---| default | Arrow-key confirm for all writes/commands Yes / Don't ask again / No | relaxed | Auto-approve file operations, only confirm commands + deletes | auto | No confirmations — everything auto-approved | Switch with /mode relaxed or /mode auto . Default is default . Activate with /plan . The AI will: Explore — read files, search, analyze no writes allowed Present plan — show what it intends to do You decide — approve, revise, or reject arrow keys Implement — on approval, AI executes the plan When enabled default ON, toggle with /accept , file edits are queued and presented for review: ? 1/3 app.py — write +15 -3 lines ❯ Accept Reject Show diff Accept all — remaining 3 edits Reject all Navigate with arrow keys, view diffs, accept/reject individually or in bulk. Connect external tool servers via the Model Context Protocol — local or remote : Local stdio server — a command Kodiqa runs /mcp add fs npx -y @modelcontextprotocol/server-filesystem ~/projects Remote HTTP server — a hosted URL, with optional auth /mcp add linear https://mcp.linear.app/mcp --bearer env:LINEAR TOKEN /mcp add api https://example.com/mcp --header "X-Api-Key:abc123" Remote with OAuth login opens your browser /mcp add linear https://mcp.linear.app/mcp --oauth …or machine-to-machine no browser /mcp add api https://example.com/mcp --oauth-client-id env:CID --oauth-client-secret env:CSEC Any REST API via its OpenAPI spec — each operation becomes a tool no codegen /mcp add petstore --spec https://petstore3.swagger.io/api/v3/openapi.json Any GraphQL endpoint — each query/mutation becomes a tool /mcp add gql --graphql https://api.example.com/graphql --bearer env:TOKEN /mcp list show servers + kind stdio / http / openapi / graphql + lazy mode /mcp remove mytools disconnect /mcp lazy on|off toggle lazy tool loading default: on Remote servers use the Streamable HTTP transport. Auth values support env:VAR and file:PATH so tokens aren't typed inline. OAuth --oauth handles discovery, dynamic client registration, the PKCE browser login, and automatic token refresh — tokens are cached under ~/.kodiqa/oauth/ and reused across sessions. MCP tools are automatically available to the AI alongside built-in tools, and work with lazy mode discovered on demand . Most agents paste every MCP tool's JSON schema into every request, so a big MCP server quietly taxes every turn. Kodiqa doesn't. When servers are connected, it exposes 3 fixed meta-tools and lets the model discover tools on demand: | Per-turn tool-schema cost 50-tool server | | |---|---| | Inject all schemas typical agents | ~5,300 tokens | Kodiqa lazy mode | ~310 tokens ~94% less | mcp search — find tools by keyword, ranked by how often you've used them mcp tool schema — fetch one tool's full schema only when needed mcp call — run a tool by name It's on by default , fully automatic the model drives it , and the cost stays flat no matter how many MCP tools you connect. Toggle with /mcp lazy off , or set mcp lazy: false in settings to always inject every schema. | Shortcut | Full Model | Best For | |---|---|---| /model fast | qwen3:30b-a3b | Fast answers, 30B brain at 3B speed MoE | /model qwen | qwen3:14b | General purpose, smart, thinking mode | /model coder | qwen3-coder | Coding agent default without API key | /model reason | phi4-reasoning | Deep reasoning, math, logic | /model gpt-local | gpt-oss | OpenAI's open model, reasoning + agentic | | Shortcut | Full Model | Price in/out per MTok | |---|---|---| /model claude / sonnet | claude-sonnet-4-6 | $3/$15 | /model opus | claude-opus-4-6 | $5/$25 | /model haiku | claude-haiku-4-5 | $1/$5 | /model sonnet-4.5 | claude-sonnet-4-5 | $3/$15 | /model opus-4.5 | claude-opus-4-5 | $5/$25 | /model opus-4.1 | claude-opus-4-1 | $15/$75 | /model sonnet-4 / opus-4 | Legacy Claude 4 | varies | | Shortcut | Full Model | Best For | |---|---|---| /model qwen3.5 / qwen-plus | qwen3.5-plus | Newest flagship | /model qwen-max / qwen3-max | qwen3-max | Most powerful | /model qwen-coder / qwen3-coder | qwen3-coder-plus | Coding | /model qwen-coder-next | qwen3-coder-next | Newest coder | /model qwq | qwq-plus | Deep reasoning | /model qwen-flash | qwen3.5-flash | Fast | /model qwen-turbo | qwen-turbo | Cheapest/fastest | /model qwen-math | qwen-math-plus | Math | /model glm-5 | glm-5 | Third-party Coding Plan | /model kimi | kimi-k2.5 | Third-party Coding Plan | Qwen Coding Plan : If you have a Coding Plan subscription sk-sp- key , /key qwen auto-detects it and configures the dedicated endpoint. Supports $3/mo Lite and $15/mo Pro tiers. | Shortcut | Full Model | Best For | |---|---|---| /model gpt | gpt-4o | General purpose flagship | /model gpt-mini | gpt-4o-mini | Fast and cheap | /model o3 | o3 | Deep reasoning | /model o3-mini | o3-mini | Fast reasoning | /model o4-mini | o4-mini | Latest reasoning | | Shortcut | Full Model | Best For | |---|---|---| /model deepseek | deepseek-chat | V3 general purpose | /model deepseek-r1 | deepseek-reasoner | R1 deep reasoning | | Shortcut | Full Model | Best For | |---|---|---| /model llama | llama-3.3-70b-versatile | Best open model | /model llama-small | llama-3.1-8b-instant | Ultra fast | /model gemma | gemma2-9b-it | Google's open model | /model mixtral | mixtral-8x7b-32768 | MoE, 32K context | | Shortcut | Full Model | Best For | |---|---|---| /model mistral | mistral-large-latest | Flagship | /model mistral-small | mistral-small-latest | Fast and cheap | /model codestral | codestral-latest | Code generation | New models are auto-discovered from the APIs — they appear in /model and /models automatically. You can also use full model names: /model qwen3:14b or /model claude-opus-4-6 Run Kodiqa as a small local HTTP server that your editor VS Code, Zed, Neovim, … can call: kodiqa --serve prints the URL + an auth token or, inside a session: /serve It binds to 127.0.0.1 only and requires the printed bearer token. Set a stable token handy for editor config with KODIQA BRIDGE TOKEN=… , and a fixed port with --port . Protocol: | Endpoint | || |---|---|---| GET /health | no auth | {status, model, version} | POST /ask | {prompt, context?} | {response} — one-shot model answer no history, no file edits | GET /diagnostics?file=PATH | {file, diagnostics} from the LSP start one with /lsp | curl -s localhost:PORT/ask -H "Authorization: Bearer TOKEN" \ -d '{"prompt":"explain this","context":"def f : return 1"}' /ask is a safe, non-streaming Q&A call — ideal for "ask Kodiqa about the selection." A minimal reference client is in examples/bridge client.py /Kodiqa-Solutions/Kodiqa-agent/blob/main/examples/bridge client.py ; editor extensions are thin clients over this API. By default, Kodiqa hides code blocks during streaming and shows progress instead: Kodiqa I'll create the project structure... ⠋ Writing code javascript ... 45 lines, 1,890 chars ╰─ code block: javascript 45 lines, 1,890 chars Now the package.json: ⠋ Writing code json ... 12 lines, 340 chars ╰─ code block: json 12 lines, 340 chars 1,204 in / 847 out | 42.3 tok/s | $0.0061 / session: $0.0183 Use /verbose to toggle full output see all code as it streams . Use /key to add API keys interactively shows all 6 providers , or specify directly: | Provider | Command | Get Key | |---|---|---| | Claude | /key claude | | /key openai https://platform.openai.com/api-keys https://platform.openai.com/api-keys /key deepseek https://platform.deepseek.com/api keys https://platform.deepseek.com/api keys /key groq https://console.groq.com/keys https://console.groq.com/keys /key mistral https://console.mistral.ai/api-keys https://console.mistral.ai/api-keys /key qwen https://bailian.console.alibabacloud.com/?apiKey=1 https://bailian.console.alibabacloud.com/?apiKey=1 Then switch: /model claude , /model gpt , /model deepseek , /model llama , /model mistral , /model qwen3.5 read the file ~/.zshrc create a file called hello.py with a hello world program edit main.py and change the function name from foo to bar move config.json to config.backup.json delete the temp file at ~/scratch.txt rename all occurrences of "oldName" to "newName" in utils.py undo the last edit to main.py find all .py files in ~/projects search for "TODO" in my project run npm install show me the git status commit these changes with message "fix login bug" search the web for kotlin coroutines tutorial fetch the content from https://some-docs-page.com remember that I prefer Kotlin for Android development what do you remember about my preferences? look at this screenshot ~/Desktop/screenshot.png read the PDF ~/Documents/report.pdf paste what's on my clipboard copy this code to clipboard /scan ~/myapp now explain what this project does find any bugs in this code Auto-approved : reading files, listing dirs, searching, web, memory, clipboard read, undo Asks permission : writing/editing files, running commands, git commits, delete, move, clipboard write, patches Workspace boundary : asks before accessing files outside current working directory Allow once / Allow directory / Deny Blocked : rm -rf / , sudo rm , mkfs , dd , fork bombs, etc. Permission modes : /mode default confirm all → /mode relaxed auto file ops → /mode auto no confirms | Category | Tools | |---|---| | File ops | read file, write file, edit file, multi edit, search replace all, create directory, move file, delete file, undo edit | | Search | glob, grep, list dir, tree | | Commands | run command | | Git | git status, git diff, git commit | | Web | web search, web fetch | | Media | read image, read pdf | | Memory | memory store, memory search | | Clipboard | clipboard read, clipboard write | | Patch | diff apply | | UX | ask user | ~/LLMS/kodiqa/ kodiqa.py Main agent 5699 lines actions.py 26 action handlers 1022 lines tools.py Tool schemas 461 lines config.py Config, themes, provider registry 585 lines web.py Web search + page fetch 194 lines memory.py SQLite persistent memory 82 lines mcp.py MCP client 176 lines templates.py 5 project templates 61 lines lsp.py LSP client 220 lines embeddings.py RAG vector store 93 lines repomap.py Tree-sitter/regex repo map 157 lines bin/kodiqa Global install script tests/ 317 tests pytest pyproject.toml Package config pip install . requirements.txt Dependencies ~/.kodiqa/ config.json User-editable config overrides defaults settings.json API keys, default model memory.db Persistent memories session.json Auto-saved conversation input history prompt toolkit FileHistory error.log Error log capped 1MB KODIQA.md Global context always in system prompt projects/ Per-project context files checkpoints/ Conversation checkpoints exports/ Exported session markdown files - All prompts use arrow keys — no typing letters, just navigate and press Enter - Default is compact mode — code hidden during streaming, progress shown instead - Use /verbose when you want to see code as it streams - Use /mode relaxed to skip file edit confirmations - Use /plan for complex tasks — review the plan before implementation - Use /accept to toggle batch edit review on/off - Use /branch save before experimenting — switch back if it goes wrong - Use /mcp add to connect external tool servers - Use /checkpoint before risky operations, /restore to roll back - Use /export to save a conversation for later reference - Use /tokens to monitor API costs and context usage - Use /model with no arg for interactive picker - Use /key with no arg to choose provider - Tab complete works for commands, models, and file paths - New API models appear automatically — no code updates needed - Memories persist forever across sessions - Arrow keys work: up/down for history, left/right to edit - Sessions auto-save — restart if anything goes wrong - Session summary auto-saved on quit — next start has full context - Type quit or exit no slash needed to exit - Ollama starts/stops automatically — stops on cloud switch, restarts on local switch pytest -v 317 tests, all passing - Python 3.9+ - Ollama installed /Applications/Ollama.app on macOS — or just use API models - Models pulled automatically on first run, or ollama pull qwen3-coder - Optional Claude API key for Claude models - Optional DashScope API key for Qwen API models Kodiqa is open source under the AGPL-3.0 License /Kodiqa-Solutions/Kodiqa-agent/blob/main/LICENSE . For commercial use without AGPL obligations, see Commercial License /Kodiqa-Solutions/Kodiqa-agent/blob/main/COMMERCIAL LICENSE.md or contact eniz@kodiqa.com mailto:eniz@kodiqa.com .