krodo 0.1.1

Krodo — a local-first, multi-provider coding agent CLI

Krodo

English | 中文

A local-first, multi-provider coding agent CLI, built with Python 3.12+.

Status: 🚧 Pre-alpha (v0.1.1) — Phase 1 feature-complete & on PyPI (REPL + headless + pipe, 11 tools, JSONL sessions, three approval modes). Phase 2 (TUI, MCP client, tree-sitter symbol index) in planning.

Krodo is an open-source coding agent inspired by Claude Code, Codex CLI, and Aider. It runs locally, talks to your codebase through tools (read / edit / shell / git / grep), and supports any LLM provider via LiteLLM — Anthropic, OpenAI, Gemini, DeepSeek, Qwen, plus local models via Ollama / vLLM.

Why another coding agent

• Local-first: your code never leaves your machine except for LLM API calls.
• Multi-provider from day 1: switch between Claude, GPT, Gemini, DeepSeek, Qwen, or local models with a single config flag.
• Three CLI shapes, one core: krodo REPL, krodo "<prompt>" headless, krodo tui (Phase 2) — all share the same agent loop.
• Safety as a default: three approval modes (read_only / auto_edit / full_auto), path firewall, dangerous-command blocklist, automatic git checkpoint before every write.
• Modular monolith: clean Protocol-based interfaces between core / llm / tools / sandbox / memory / obs. Easy to read, easy to contribute to.

For full design rationale, see docs/architecture.md.

Roadmap

Phase	Scope	Status
0	Single-file prototype validating the ReAct loop	✅ done
1 M1	Walking skeleton (3 tools, CLI, agent loop)	✅ done
1 M2	Full tools (11 tools) + three approval modes + pattern trust	✅ done
1 M3	Context management (token budget + dual compression) + 7 recovery scenarios	✅ done
1 M4	.krodoignore + git checkpoint + krodo undo + diff preview	✅ done
1 M5	Persistence + memory: JSONL sessions, krodo resume, AGENTS.md, config files	✅ done
1 M6	Streaming + cost tracking + pipe stdin + REPL slash commands + approval persistence	✅ done
1 M7	Brand rename (Coda → krodo) + mypy strict clean + docs quartet + GitHub release + dogfood PR	✅ done
2	tree-sitter symbol index, repo-map, Textual TUI, MCP client	—
3	OS-level sandbox, evaluation harness, OpenTelemetry / Langfuse	—
4	Production-grade: Rust hot paths, single-binary distribution, LiteLLM Proxy	—

Quick start

Try the v0.1.1 release (Phase 1 feature-complete)

uv tool install krodo                 # or: pipx install krodo
export ANTHROPIC_API_KEY=sk-ant-...   # or OPENAI_API_KEY / ZAI_API_KEY / ...

mkdir -p /tmp/krodo-sandbox

# Headless: run one task and exit
krodo --root /tmp/krodo-sandbox "create hello.py that prints Hello Krodo, then run it"

# REPL: omit the prompt to enter interactive multi-turn mode
krodo --root /tmp/krodo-sandbox
# you> create a simple mario game
# (assistant works, then…)
# you> now add a sound effect when collecting coins
# you> exit          # or Ctrl-D / Ctrl-C twice

# Pipe: stdin becomes the prompt…
echo "create hello.py that prints Hello Krodo" | krodo --root /tmp/krodo-sandbox

# …or extra context when a prompt is given
git diff | krodo "review this change for bugs"

Assistant text streams token-by-token, and every session summary ends with a cost line like tokens : 12.3k in / 4.1k out | cost $0.0231.

In REPL mode the conversation history (including everything the agent did in the previous turn) is carried over, so follow-ups like "now add X" or "fix the bug from before" work naturally. Exit with exit / quit / :q, Ctrl-D, or two consecutive Ctrl-C presses at the prompt.

REPL slash commands

Slash commands are handled locally — they are never sent to the LLM:

Command	Action
:help	List available commands
:sessions	Show the 10 most recent sessions in this workspace
:undo	Restore files to the previous checkpoint
:cost	Show session token / cost totals
:resume <id>	Switch to another session (history is replayed)
:q	Quit (same as exit)

Resuming a previous session

Sessions are persisted automatically. You can pick up right where you left off:

# List recent sessions
krodo resume --list

# Resume by session ID (or unique prefix)
krodo resume a3f2b1

# Resume in a specific workspace
krodo resume --root /tmp/krodo-sandbox a3f2b1

krodo resume replays the stored conversation history into a fresh REPL, so the model remembers everything from the prior session — files edited, tools called, and dialogue.

Install (PyPI)

Krodo is on PyPI. Pick your favourite installer:

uv tool install krodo                # recommended (fast, isolated)
# or
pipx install krodo                   # equally good
# or
pip install krodo                    # works, but pip's global installs can clash

Verify:

krodo --version                      # krodo <version>
krodo --help

To run from source instead (for development):

git clone https://github.com/wadekun/krodo
cd krodo
uv sync                              # creates .venv with all deps + dev group
uv run krodo --help                  # `uv run` needed because not pip-installed

Available tools (11 total)

Tool	Category	Requires approval	Description
read_file	Read	No	Read a file (with optional offset/limit)
list_dir	Read	No	List directory contents (depth-limited, noise-dirs skipped)
glob	Read	No	Find files matching a pattern (**/*.py)
grep	Read	No	Regex search; ripgrep when available, Python fallback
git_status	Git	No	Show working tree status (git status --porcelain)
git_diff	Git	No	Show unified diff (staged/unstaged, optional path filter)
write_file	Write	Yes	Write or overwrite a file
edit_file	Write	Yes	Targeted string replacement with uniqueness enforcement
apply_patch	Write	Yes	Apply a unified diff atomically with rollback on failure
run_shell	Shell	Yes	Execute a shell command inside the workspace sandbox
git_commit	Git	Yes	Commit staged files (API keys auto-redacted from message)

Context & Recovery

Krodo enforces a token budget and offers dual compression strategies so that long sessions never overflow the model's context window.

Token budget (§3.4.1)

The budget is 80% of the model's context window. At 80% usage, compression is triggered. At 95%, hard truncation kicks in as a safety net. If the available budget hits zero, the next turn is refused with a clear message.

Total budget   = model_context_window × 0.80
Output reserve = total_budget × 0.15
Compress at    = 80% of budget (default)
Truncate at    = 95% of budget

Compression strategies

Select via KRODO_COMPRESS environment variable:

Strategy	Env value	Description
LLM summary (default)	llm	Calls the same LLM provider to summarise the oldest N dialogue rounds into a <SUMMARY>…</SUMMARY> block.
Algorithmic	algorithmic	Drops tool_result content, keeps tool-call metadata and file paths. Zero extra LLM cost — great for offline dev or large codebases.

# Use algorithmic compression (no extra LLM calls):
KRODO_COMPRESS=algorithmic krodo "..."

# Override the token ratio for Claude (default 1.1x, tiktoken undercounts):
KRODO_TOKEN_RATIO=1.15 krodo --model anthropic/claude-3-5-sonnet "..."

Error recovery (7 scenarios, §7.5)

#	Scenario	Recovery
1	LLM returns invalid tool-call JSON	Re-inject schema + error; retry ×2, then abort
2	Tool execution timeout	Kill subprocess; skip tool call with truncated partial result
3	Agent stall (3× same write-tool call)	Abort turn; show last 3 calls to user
4	Compression-induced context loss	Re-inject pinned file paths + last user message
5	File externally modified (SHA-256 conflict)	Block write; ask agent to re-read the file first
6	Provider rate limit / 5xx	Exponential back-off ×3 (1 s / 2 s / 4 s)
7	File permission denied (EACCES)	Skip write; report path + permission bits

CLI flags

# Limit tool calls per turn (default 25):
krodo --max-tool-calls 5 "..."

# Set compression window (how many dialogue rounds to compress at once):
krodo --summary-window 3 "..."

Persistence & Memory

Session storage

Every session is automatically saved to .krodo/sessions/<session_id>.jsonl in your workspace. Each line is a JSON event (USER_MESSAGE, ASSISTANT_MESSAGE, TOOL_CALL, TOOL_RESULT, COMPRESSION, etc.) with a monotonic seq number so multi-process appends are safe.

Application logs go to .krodo/logs/<session_id>.log (pure structlog JSONL — separate from session events).

AGENTS.md — project memory

Place an AGENTS.md file anywhere in your project and Krodo will inject it automatically into every session as <project_memory>:

Tier	Location	Purpose
System	~/.config/krodo/AGENTS.md	Personal conventions (applies to all workspaces)
Project	<workspace>/AGENTS.md	Project-specific rules (always included, never dropped)
Subdir	<cwd>/AGENTS.md … up to workspace root	Contextual docs for the directory you're working in

Each file is limited to 8K tokens; total budget is 12K tokens (subdirectory files are dropped first if the limit is hit).

Configuration files

Defaults can be set in .krodo/config.yaml (workspace) or ~/.config/krodo/config.toml (user-global). Precedence: CLI flag > env var > workspace > user > built-in default.

Quick example:

# .krodo/config.yaml — workspace-level default
model: deepseek/deepseek-v4-flash
approval: auto_edit
max_tool_calls: 15

Full field reference + 10 providers + per-provider API keys + troubleshooting (field-name gotchas, proxy caveats, error-pattern diagnosis): see Models & Providers.

Run krodo doctor after every config change to verify what's actually loaded.

.krodoignore & Git checkpoint

Two safety nets: a 4-tier ignore system and automatic git checkpointing before every write.

.krodoignore — 4-tier path filtering (§5.3)

Every read_file, list_dir, glob, and grep call passes through KrodoIgnore before touching the disk. Rules are merged from four sources in increasing specificity order:

Tier	Source	Overridable?
1	Hard-coded defaults (.env, *.pem, id_rsa, node_modules/, etc.)	❌ always active
2	Project .gitignore	—
3	Project .krodoignore (workspace root)	adds custom patterns
4	User-level ~/.config/krodo/krodoignore	personal overrides

When a path matches any rule, the tool returns:

PathIgnoredError: '<path>' is ignored (rule: '<pattern>' from <source>)

Example .krodoignore

# Exclude internal data directories from agent reads
data/raw/
reports/*.csv

# Exclude generated mock files
tests/fixtures/generated/

Git checkpoint (§5.4)

Before every write (write_file, edit_file, apply_patch) and write-heuristic shell command, Krodo creates a lightweight git stash create checkpoint:

1. Collect affected paths.
2. checkpoint_sha = git stash create — does not push to the stash stack; working tree is untouched.
3. Emit a CHECKPOINT SessionEvent to .krodo/logs/<session>.jsonl.
4. Execute the write.

On non-git workspaces, checkpointing degrades to a no-op (warning logged; writes proceed normally).

krodo undo

# Undo the last checkpoint in the most recent session:
krodo undo [--root <workspace>]

# Undo a specific session:
krodo undo --session <session_id> [--root <workspace>]

krodo undo reads the session JSONL, finds the most recent CHECKPOINT event, and runs git checkout <sha> -- <affected_paths> to restore only those paths. Other files are untouched.

Condition	Behaviour
Non-git workspace	Exit 1 with friendly error
No CHECKPOINT found	Exit 1 with log path hint
affected_paths = workspace root (shell command scope)	Prompts for confirmation before restoring

CLI subcommand semantics

Krodo has three named subcommands — resume, undo, and doctor — alongside a free-form headless prompt. The parser resolves the two as follows:

Invocation	Behaviour
krodo "create a mario game"	Headless — prompt is "create a mario game"
krodo	Interactive REPL
krodo resume	Resume subcommand (most recent session)
krodo resume abc123	Resume subcommand with session ID abc123
krodo resume --root /path	Resume subcommand; --root goes to resume
krodo --root /path resume	Resume subcommand; global --root inherited as default
krodo undo	Undo subcommand
krodo doctor	Doctor subcommand

Key rules:

• The first non-option token is checked against registered subcommand names. If it matches, the token triggers subcommand dispatch — it is never treated as the headless prompt.
• Global flags (--root, --model, --approval, etc.) can go before or after the subcommand token. When placed before, they are propagated to the subcommand as defaults; an explicit flag in the subcommand itself always wins.
• Natural-language prompts should be quoted so they arrive as a single token. Without quotes, the first word could match a subcommand name:

  krodo "resume the work from yesterday"   # ✓ headless with full prompt
  krodo resume the work from yesterday     # ✗ routes to resume subcommand; "the" is unexpected arg
  

Local development

Krodo uses uv for dependency and venv management.

git clone https://github.com/wadekun/krodo
cd krodo
uv sync                       # install deps + create .venv
uv run pytest                 # run tests
uv run ruff check             # lint
uv run mypy src               # type-check

Set your LLM credentials in the environment (any subset, depending on the provider you use):

export ANTHROPIC_API_KEY=sk-ant-...
export OPENAI_API_KEY=sk-...
# or use the system keyring (recommended for shared machines, see docs/architecture.md §7.2)

Project layout

krodo/
  src/krodo/
    cli/        # Typer entry, REPL, headless exec
    core/       # Agent loop, Context, Budget, Compression, Recovery, Events
    llm/        # LLMProvider Protocol + LiteLLM adapter
    tools/      # File / shell / patch / search / git tools
    sandbox/    # Path firewall, command policy, approval modes
    memory/     # JSONL session store, krodo resume, AGENTS.md loader, config
    obs/        # structlog + OpenTelemetry + cost tracker
  tests/{unit,integration,e2e}/
  docs/
    architecture.md         # design baseline (read this first)
    reviews/                # past architecture review notes
  scripts/
    prototype.py            # Phase 0 single-file prototype (DEPRECATED — use krodo CLI)

Contributing

This is a learning + production project. Contributions welcome — Phase 1 is feature-complete and the CI gate is stable.

Ground rules:

• All code goes through ruff + mypy --strict + pytest --cov (see CONTRIBUTING.md for the full CI gate).
• All new tools come with 100% unit test coverage and an integration test against a recorded LLM response (vcrpy).
• All changes that touch the agent loop must pass the regression matrix (Phase 2+).
• See docs/architecture.md §11 for the seven engineering principles.

Documentation

Document	What's in it
docs/QUICKSTART.md	5-minute install + first task
docs/MODELS.md	Model & provider config, switching, troubleshooting
CONTRIBUTING.md	Dev setup, CI gate, PR flow, commit conventions
SECURITY.md	Threat model, sandbox boundaries, vulnerability reporting
CHANGELOG.md	Milestone-by-milestone changes
docs/architecture.md	Full design baseline (the source of truth)
AGENTS.md	Auto-loaded project memory (loaded into every session)

License

Apache-2.0 © The Krodo Contributors