gitbasher 5.0.1 → 5.1.0

package/CHANGELOG.md

@@ -2,6 +2,29 @@
 
 All notable changes to gitbasher are generated automatically by [semantic-release](https://github.com/semantic-release/semantic-release) from [Conventional Commits](https://www.conventionalcommits.org/) on `main`. The project follows [Semantic Versioning](https://semver.org/).
 
+## [5.1.0](https://github.com/maxbolgarin/gitbasher/compare/v5.0.1...v5.1.0) (2026-07-03)
+
+### 🚀 Features
+
+* **ai:** add local Claude Code CLI (claude -p) provider ([bc3e663](https://github.com/maxbolgarin/gitbasher/commit/bc3e66341903c1ed93f7191dc6ee6840173bf13b))
+* **config:** keep or replace an existing AI key with one keystroke ([95ba4a6](https://github.com/maxbolgarin/gitbasher/commit/95ba4a649d8ab9288cb5e73760076effce2b8ddc))
+* **commit:** offer manual fallback when AI message generation fails ([39ac769](https://github.com/maxbolgarin/gitbasher/commit/39ac7697d6ca5063c09aa8b661c1263581e2adc3))
+* **config:** turn gitb cfg ai into a provider/key/model setup wizard ([22795fe](https://github.com/maxbolgarin/gitbasher/commit/22795fe90980b2a6e848529eff6aa3a2939f89f5))
+
+### 🐛 Bug Fixes
+
+* **ai:** move OpenRouter default models to the current generations ([82a26a0](https://github.com/maxbolgarin/gitbasher/commit/82a26a0797aafe11e71ff4e8e737830fe567b4c2))
+* **ai:** pick default models by measured latency, speed first ([29d48b8](https://github.com/maxbolgarin/gitbasher/commit/29d48b8d21ac04bd639dee723770a2bcdae38328))
+
+### 📚 Documentation
+
+* **site:** mention the claude CLI provider and the cfg ai wizard ([23c9087](https://github.com/maxbolgarin/gitbasher/commit/23c9087783b805f8ed651735304305a1163a47c3))
+* refresh test count, document the AI fallback, mention claude on the site FAQ ([af6527d](https://github.com/maxbolgarin/gitbasher/commit/af6527d08aceb20bf75ab7a3c0fe745fe1cef16f))
+
+### 💎 Styles
+
+* capitalize the Enter default in every y/n prompt ([37349a0](https://github.com/maxbolgarin/gitbasher/commit/37349a0d166b235b9b7c33dc2767e5b2093ec004))
+
 ## [5.0.1](https://github.com/maxbolgarin/gitbasher/compare/v5.0.0...v5.0.1) (2026-07-03)
 
 ### 🐛 Bug Fixes

package/README.md

@@ -77,7 +77,7 @@ npm install -g gitbasher
 cd your-project
 gitb              # see all commands
 gitb cfg user     # set your name/email once
-gitb cfg ai       # (optional) plug in an AI key for smart commits
+gitb cfg ai       # (optional) set up AI for smart commits (provider, key, model)
 
 gitb status       # what's changed?
 gitb commit       # interactive conventional commit
@@ -111,7 +111,7 @@ Every command has a short alias (`gitb c`, `gitb p`, `gitb pu`, `gitb b`, `gitb
 - **AI commit messages** — `gitb c ai` writes the message from your diff. OpenRouter (Gemini/Claude/GPT/…), OpenAI direct, or **fully local via Ollama** — no key, no network, no data leaves your machine.
 - **Safer git** — push/pull detect conflicts up front, `undo` rolls back commit/amend/merge/rebase/stash, `reset` is interactive with a preview.
 - **Whole workflows, not just commands** — `sync`, `wip`, `branch newd`, `merge to-main`, `squash` chain the steps you'd otherwise do by hand.
-- **One file, no deps** — pure bash for every git operation; only the AI features need `jq` and `curl`. Drop the binary anywhere on `PATH` and go. **870+ BATS tests** cover sanitization, git ops, and branch logic.
+- **One file, no deps** — pure bash for every git operation; only the AI features need `jq` and `curl`. Drop the binary anywhere on `PATH` and go. **900+ BATS tests** cover sanitization, git ops, and branch logic.
 
 <p align="center">
   <img src=".github/push.gif" width="640" alt="gitb push demo" />
@@ -216,19 +216,20 @@ Drop in an API key once (or run a local model with no key at all), then let an L
 
 ### Providers
 
-gitbasher supports three providers behind the same OpenAI-style chat-completions API. Default is `openrouter` — existing setups keep working unchanged.
+gitbasher supports four providers. Default is `openrouter` — existing setups keep working unchanged.
 
 | Provider | Best for | Needs key? |
 |----------|----------|-----------|
 | `openrouter` (default) | Trying many models behind one key (Gemini, Claude, GPT, DeepSeek…) | Yes — [openrouter.ai/keys](https://openrouter.ai/keys) |
 | `openai` | Direct access to GPT-5.4 family at OpenAI's own pricing | Yes — [platform.openai.com/api-keys](https://platform.openai.com/api-keys) |
 | `ollama` | **Fully local, fully private** — no key, no network, runs on your machine | No |
+| `claude` | You already use [Claude Code](https://claude.com/claude-code) — reuses the local `claude` CLI and its login, billed to your Claude account | No |
 
 ### Setup
 
 ```bash
 # 1. Pick a provider (skip to use the OpenRouter default)
-gitb cfg provider     # interactive — choose openrouter, openai, or ollama
+gitb cfg provider     # interactive — choose openrouter, openai, ollama, or claude
 
 # 2. For openrouter / openai: paste your key (local repo or global)
 gitb cfg ai
@@ -237,6 +238,9 @@ gitb cfg ai
 ollama serve &
 ollama pull qwen3:8b
 
+#    For claude: just have the Claude Code CLI installed and signed in
+npm install -g @anthropic-ai/claude-code
+
 # 3. Optional: HTTP proxy (for restricted regions, openrouter/openai only)
 gitb cfg proxy
 ```
@@ -255,18 +259,18 @@ Each task uses a model tuned for speed/cost/quality, picked per provider. Defaul
 
 | Task | Default model | Why |
 |------|---------------|-----|
-| `simple` (one-line message) | `google/gemini-3.1-flash-lite-preview` | Cheapest fast tier |
-| `subject` (after manual type/scope) | `google/gemini-3.1-flash-lite-preview` | Short structured output |
-| `full` (header + body) | `google/gemini-3-flash-preview` | Better prose |
-| `grouping` (atomic-split mapping) | `anthropic/claude-haiku-4.5` | Strict instruction following |
+| `simple` (one-line message) | `google/gemini-3.1-flash-lite` | Fastest model measured (~0.6s), cheapest |
+| `subject` (after manual type/scope) | `google/gemini-3.1-flash-lite` | Short structured output, instant |
+| `full` (header + body) | `google/gemini-3.5-flash` | Same speed class, better prose |
+| `grouping` (atomic-split mapping) | `google/gemini-3.5-flash` | Fast + strict structured output |
 
 **OpenAI** — GPT-5.4 family (released March 2026)
 
 | Task | Default model | Why |
 |------|---------------|-----|
-| `simple` / `subject` | `gpt-5.4-nano` | Built for classification/short well-defined output, ~$0.20 / $1.25 per M tokens |
-| `full` | `gpt-5.4-mini` | Stronger multi-condition instruction following for header + body, ~$0.75 / $4.50 per M |
-| `grouping` | `gpt-5.4-mini` | Holds the strict TSV format under validation, far cheaper than the flagship |
+| All tasks | `gpt-5.4-mini` | Measured as fast as nano (~0.8s) with strictly better quality, ~$0.75 / $4.50 per M |
+
+`gpt-5.4-nano` (~$0.20 / $1.25 per M) remains a good budget override for high-volume use.
 
 **Ollama** — fully local
 
@@ -276,6 +280,15 @@ Each task uses a model tuned for speed/cost/quality, picked per provider. Defaul
 
 Other strong local picks: `llama3.3:8b` (general-purpose), `qwen2.5-coder:7b` (code-heavy diffs).
 
+**Claude Code CLI** — local `claude -p`
+
+| Task | Default model | Why |
+|------|---------------|-----|
+| `simple` / `subject` | `haiku` | Fastest and cheapest for one-line output |
+| `full` / `grouping` | `sonnet` | Better prose and strict structured output |
+
+The alias slugs (`haiku`, `sonnet`, `opus`) always resolve to the current model generation; full ids like `claude-haiku-4-5` work too.
+
 Override per task or globally:
 ```bash
 gitb cfg model                                          # interactive
@@ -807,8 +820,9 @@ gitb origin remove                               # delete the remote
 | `editor` | `ed` `e` | Commit-message editor |
 | `ticket` | `ti` `t` `jira` | Ticket prefix for commits/branches |
 | `scopes` | `sc` `s` | Common scopes |
-| `ai` | `llm` `key` | AI API key |
-| `provider` | `prov` | AI provider (openrouter, openai, ollama) |
+| `ai` | `llm` | AI setup wizard: provider, API key, model |
+| `key` | | AI API key only |
+| `provider` | `prov` | AI provider (openrouter, openai, ollama, claude) |
 | `model` | `m` | Default AI model |
 | `proxy` | `prx` `p` | HTTP proxy for AI calls |
 | `history` | `hist` | How many recent commits to include in AI prompts |
@@ -889,12 +903,12 @@ Overview-first diffs built for the gitbasher workflow — no flag memorization.
 | `gitbasher.ticket` | `gitb cfg ticket` | Ticket prefix (`PROJ-`) |
 | `gitbasher.scopes` | `gitb cfg scopes` | Suggested commit scopes |
 | `gitbasher.ai-api-key-<provider>` | `gitb cfg ai` | Per-provider AI API key (or `GITB_AI_API_KEY_<PROVIDER>` env); legacy `gitbasher.ai-api-key` is read as a fallback |
-| `gitbasher.ai-provider` | `gitb cfg provider` | `openrouter` (default), `openai`, or `ollama` |
+| `gitbasher.ai-provider` | `gitb cfg provider` | `openrouter` (default), `openai`, `ollama`, or `claude` |
 | `gitbasher.ai-base-url` | `git config` | Custom OpenAI-compatible endpoint (LiteLLM, vLLM, remote Ollama) |
 | `gitbasher.ai-model[-task]` | `gitb cfg model` | AI model overrides (per provider) |
 | `gitbasher.ai-proxy` | `gitb cfg proxy` | HTTP proxy for AI calls |
 | `gitbasher.ai-ollama-host` | `git config gitbasher.ai-ollama-host <url>` | Ollama server URL (default `http://localhost:11434`) |
-| `gitbasher.ai-timeout` | `git config gitbasher.ai-timeout <seconds>` | AI request timeout in seconds (default `60`, `300` for ollama) |
+| `gitbasher.ai-timeout` | `git config gitbasher.ai-timeout <seconds>` | AI request timeout in seconds (default `60`, `300` for ollama/claude) |
 | `gitbasher.ai-diff-limit` | `gitb cfg diff` | Diff lines sent to AI (default `300`) |
 | `gitbasher.ai-diff-max-chars` | `gitb cfg diff` | Character cap on the diff sent to AI (default `20000`) |
 | `gitbasher.ai-commit-history-limit` | `gitb cfg history` | Recent commits included in AI prompts (default `10`) |
@@ -998,9 +1012,11 @@ gitb cfg proxy        # in restricted regions
 If `gitb commit ai` hangs, returns `connection refused`, or times out:
 
 - **jq missing** — all AI features require `jq` (check with `command -v jq`; install via `brew install jq` / `apt install jq`). Without it every AI command aborts with an install hint.
+- **Generation failed mid-commit?** gitbasher offers to continue with a manual commit message (staging is preserved) — press Enter to accept and write it yourself, or `n` to abort.
 - **Network reachability** — test the provider directly: `curl -fsSL https://api.openai.com/v1/models -H "Authorization: Bearer $OPENAI_API_KEY"` (or the equivalent for OpenRouter/Ollama). If that fails, gitbasher will too.
 - **Corporate proxy / restricted region** — `gitb cfg proxy` accepts `host:port` or `protocol://host:port`. Verify with `curl -x "$proxy" https://api.openai.com`.
 - **Local Ollama** — confirm the daemon is running (`curl http://localhost:11434/api/tags`). The default model must be pulled first (`ollama pull qwen3:8b`).
+- **Claude Code CLI** — confirm it responds (`echo hi | claude -p`); you may need to sign in first. Install with `npm install -g @anthropic-ai/claude-code`.
 - **Stale or rotated key** — `gitb cfg ai` re-prompts; prefer the env-var path (`export GITB_AI_API_KEY_OPENAI=...`) over `git config` so a leaked repo doesn't carry the secret.
 </details>
 
@@ -1079,7 +1095,7 @@ PRs welcome. See [CONTRIBUTING.md](./CONTRIBUTING.md) for the dev setup, BATS te
 
 ```bash
 make build               # rebuild dist/gitb
-make test                # run all 870+ tests
+make test                # run all 900+ tests
 make test-file FILE=test_sanitization.bats
 ```