@rely-ai/caliber 1.5.3 → 1.5.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +173 -48
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -1,5 +1,5 @@
1
1
  <p align="center">
2
- <img src="assets/social-preview.svg" alt="Caliber" width="640">
2
+ <img src="assets/social-preview.png" alt="Caliber" width="640">
3
3
  </p>
4
4
 
5
5
  <p align="center">
@@ -10,11 +10,13 @@
10
10
 
11
11
  <p align="center"><strong>Analyze your codebase. Generate optimized AI agent configs. One command.</strong></p>
12
12
 
13
- Caliber scans your project — languages, frameworks, dependencies, file structure — and generates tailored config files for Claude Code, Cursor, and Codex. If configs already exist, it audits them and suggests improvements.
13
+ ---
14
14
 
15
- **No API key required** — use your existing Claude Code or Cursor subscription. Or bring your own key (Anthropic, OpenAI, Vertex AI, any OpenAI-compatible endpoint).
15
+ Caliber scans your project — languages, frameworks, dependencies, file structure — and generates tailored config files for **Claude Code**, **Cursor**, and **OpenAI Codex**. If configs already exist, it audits them against your actual codebase and suggests targeted improvements.
16
16
 
17
- ## Quick Start
17
+ 🔑 **No API key required** — use your existing Claude Code or Cursor subscription. Or bring your own key (Anthropic, OpenAI, Vertex AI, any OpenAI-compatible endpoint).
18
+
19
+ ## 🚀 Quick Start
18
20
 
19
21
  ```bash
20
22
  npx @rely-ai/caliber onboard
@@ -35,67 +37,172 @@ caliber onboard
35
37
  > npx @rely-ai/caliber onboard
36
38
  > ```
37
39
 
38
- ## How It Works
40
+ ## ⚙️ How It Works
39
41
 
40
42
  ```
41
43
  caliber onboard
42
44
  │
43
- ├─ 1. Scan Analyze languages, frameworks, dependencies, file structure,
44
- │ and existing agent configs in your project
45
+ ├─ 1. 🔌 Connect Choose your LLM provider — Claude Code seat, Cursor seat,
46
+ │ or an API key (Anthropic, OpenAI, Vertex AI)
47
+ │
48
+ ├─ 2. 🔍 Discover Analyze languages, frameworks, dependencies, file structure,
49
+ │ and existing agent configs via LLM
50
+ │
51
+ ├─ 3. 🛠️ Generate Create tailored config files based on your codebase
52
+ │ (or audit existing ones and suggest improvements)
45
53
  │
46
- ├─ 2. Generate LLM creates tailored config files based on your codebase
47
- │ (or audits existing ones and suggests improvements)
54
+ ├─ 4. 👀 Review See a diff of proposed changes — accept, refine via
55
+ │ chat, or decline. All changes are backed up automatically
48
56
  │
49
- ├─ 3. Review You see a diff of proposed changes — accept, refine via
50
- │ chat, or decline
57
+ ├─ 5. 🔗 MCP Servers Discover and install MCP servers for tools your project
58
+ │ uses (databases, APIs, platforms)
51
59
  │
52
- └─ 4. Apply Files are written with automatic backups, before/after
53
- score is displayed
60
+ └─ 6. 🧩 Skills Search community skill registries and install relevant
61
+ skills for your tech stack
54
62
  ```
55
63
 
56
- ### What It Generates
64
+ Caliber works on **any codebase** — TypeScript, Python, Go, Rust, Terraform, Java, Ruby, and more. Language and framework detection is fully LLM-driven, not hardcoded.
65
+
66
+ ### 📦 What It Generates
67
+
68
+ | File | Platform | Purpose |
69
+ |------|----------|---------|
70
+ | `CLAUDE.md` | Claude Code | Project context — build/test commands, architecture, conventions |
71
+ | `.cursor/rules/*.mdc` | Cursor | Modern rules with frontmatter (description, globs, alwaysApply) |
72
+ | `.cursorrules` | Cursor | Legacy rules file (if no `.cursor/rules/` exists) |
73
+ | `AGENTS.md` | Codex | Project context for OpenAI Codex |
74
+ | `.claude/skills/*/SKILL.md` | Claude Code | Reusable skill files following [OpenSkills](https://agentskills.io) |
75
+ | `.cursor/skills/*/SKILL.md` | Cursor | Skills for Cursor |
76
+ | `.agents/skills/*/SKILL.md` | Codex | Skills for Codex |
77
+ | `.mcp.json` | Claude Code | MCP server configurations |
78
+ | `.cursor/mcp.json` | Cursor | MCP server configurations |
79
+ | `.claude/settings.json` | Claude Code | Permissions and hooks |
57
80
 
58
- | File | Purpose |
59
- |------|---------|
60
- | `CLAUDE.md` | Project context for Claude Code — commands, architecture, conventions |
61
- | `.cursorrules` / `.cursor/rules/` | Rules for Cursor |
62
- | `AGENTS.md` | Project context for OpenAI Codex |
63
- | Skills (`.claude/skills/`, `.cursor/skills/`, `.agents/skills/`) | Reusable skill files following the [OpenSkills](https://agentskills.io) standard |
81
+ If these files already exist, Caliber audits them and suggests improvements — keeping what works, fixing what's stale, adding what's missing.
64
82
 
65
- If these files already exist, Caliber audits them against your actual codebase and suggests targeted improvements — keeping what works, fixing what's stale, adding what's missing.
83
+ ### 🛡️ Safety
66
84
 
67
- ## Commands
85
+ Every change Caliber makes is reversible:
86
+
87
+ - 💾 **Automatic backups** — previous versions saved to `.caliber/backups/{timestamp}/` before every write
88
+ - 📊 **Score regression guard** — if a regeneration produces a lower score, changes are auto-reverted
89
+ - ↩️ **Full undo** — `caliber undo` reverts all changes made by Caliber
90
+ - 🔍 **Dry run** — preview any command's changes with `--dry-run`
91
+
92
+ ## 📋 Commands
68
93
 
69
94
  | Command | Description |
70
95
  |---------|-------------|
71
- | `caliber onboard` | Onboard your project for AI-assisted development |
72
- | `caliber score` | Score your config quality (deterministic, no LLM needed) |
73
- | `caliber skills` | Discover and install community skills for your project |
74
- | `caliber config` | Configure LLM provider, API key, and model |
96
+ | `caliber onboard` | 🏁 Onboard your project — full 6-step wizard |
97
+ | `caliber score` | 📊 Score your config quality (deterministic, no LLM) |
98
+ | `caliber skills` | 🧩 Discover and install community skills |
99
+ | `caliber regenerate` | 🔄 Re-analyze and regenerate your setup |
100
+ | `caliber refresh` | 🔃 Update docs based on recent code changes |
101
+ | `caliber hooks` | 🪝 Manage auto-refresh hooks |
102
+ | `caliber config` | ⚙️ Configure LLM provider, API key, and model |
103
+ | `caliber status` | 📌 Show current setup status |
104
+ | `caliber undo` | ↩️ Revert all changes made by Caliber |
105
+
106
+ ### Examples
75
107
 
76
108
  ```bash
77
- caliber onboard --agent claude # Target Claude Code only
78
- caliber onboard --agent cursor # Target Cursor only
79
- caliber onboard --agent codex # Target OpenAI Codex only
80
- caliber onboard --agent all # Target all three
81
- caliber onboard --agent claude,cursor # Comma-separated
82
- caliber onboard --dry-run # Preview without writing files
83
- caliber score --json # Machine-readable output
109
+ # Onboarding
110
+ caliber onboard # Interactive — picks agent, walks through setup
111
+ caliber onboard --agent claude # Target Claude Code only
112
+ caliber onboard --agent cursor # Target Cursor only
113
+ caliber onboard --agent codex # Target OpenAI Codex only
114
+ caliber onboard --agent all # Target all three
115
+ caliber onboard --agent claude,cursor # Comma-separated
116
+ caliber onboard --dry-run # Preview without writing files
117
+ caliber onboard --force # Overwrite existing setup without prompting
118
+
119
+ # Scoring
120
+ caliber score # Full breakdown with grade (A-F)
121
+ caliber score --json # Machine-readable output
122
+ caliber score --agent claude # Score for a specific agent
123
+
124
+ # Day-to-day
125
+ caliber regenerate # Re-analyze and regenerate (alias: regen, re)
126
+ caliber refresh # Update docs from recent git changes
127
+ caliber refresh --dry-run # Preview what would change
128
+ caliber skills # Browse and install community skills
129
+ caliber hooks # Toggle auto-refresh hooks
130
+ caliber hooks --install # Enable all hooks non-interactively
131
+ caliber status # Show what Caliber has set up
132
+ caliber undo # Revert everything
84
133
  ```
85
134
 
86
- ## LLM Providers
135
+ ## 📊 Scoring
87
136
 
88
- | Provider | Setup | Notes |
89
- |----------|-------|-------|
90
- | **Claude Code** (your seat) | `caliber config` → Claude Code | No API key. Uses your Pro/Max/Team login via `claude -p`. |
91
- | **Cursor** (your seat) | `caliber config` → Cursor | No API key. Uses your subscription via Cursor Agent (ACP). |
92
- | **Anthropic** | `export ANTHROPIC_API_KEY=sk-ant-...` | Claude Sonnet 4.6 default. [Get key](https://console.anthropic.com). |
93
- | **OpenAI** | `export OPENAI_API_KEY=sk-...` | GPT-4.1 default. |
94
- | **Vertex AI** | `export VERTEX_PROJECT_ID=my-project` | Uses ADC. Region `us-east5`. |
95
- | **Custom endpoint** | `OPENAI_API_KEY` + `OPENAI_BASE_URL` | Any OpenAI-compatible API (Ollama, vLLM, Together, etc.) |
137
+ `caliber score` gives you a deterministic quality score — no LLM calls, no network, instant results.
138
+
139
+ ```
140
+ Config Score: 87/100 (A) ✨
141
+
142
+ FILES & SETUP 22/25 ████████████████████░░
143
+ QUALITY 21/25 ████████████████████░░
144
+ COVERAGE 18/20 ██████████████████░░░░
145
+ ACCURACY 13/15 █████████████████░░░░░
146
+ FRESHNESS & SAFETY 8/10 ████████████████░░░░░░
147
+ BONUS 5/5 ████████████████████████
148
+ ```
149
+
150
+ | Category | Points | What it checks |
151
+ |----------|--------|----------------|
152
+ | **Files & Setup** | 25 | Config files exist, skills present, cross-platform parity |
153
+ | **Quality** | 25 | Has build/test commands, not bloated, no vague text, no duplicates |
154
+ | **Coverage** | 20 | Mentions actual dependencies and services |
155
+ | **Accuracy** | 15 | Documented commands and file paths are valid |
156
+ | **Freshness & Safety** | 10 | Recently updated, no leaked secrets, permissions set |
157
+ | **Bonus** | 5 | Auto-refresh hooks, AGENTS.md, OpenSkills format |
158
+
159
+ ## 🧩 Skills
160
+
161
+ Caliber searches three community registries and scores results against your project:
162
+
163
+ - 🌐 [skills.sh](https://skills.sh) — OpenSkills registry
164
+ - 🔧 [tessl.io](https://tessl.io) — Tessl skill registry
165
+ - 📚 [Awesome Claude Code](https://github.com/hesreallyhim/awesome-claude-code) — Curated list
166
+
167
+ ```bash
168
+ caliber skills
169
+ ```
170
+
171
+ Skills are scored by LLM relevance (0–100) based on your project's actual tech stack, then you pick which ones to install via an interactive selector. Installed skills follow the [OpenSkills](https://agentskills.io) standard with YAML frontmatter.
172
+
173
+ ## 🔄 Auto-Refresh
174
+
175
+ Keep your agent configs in sync with your codebase automatically:
176
+
177
+ | Hook | Trigger | What it does |
178
+ |------|---------|--------------|
179
+ | 🤖 **Claude Code** | End of each session | Runs `caliber refresh` and updates docs |
180
+ | 📝 **Git pre-commit** | Before each commit | Refreshes docs and stages updated files |
181
+
182
+ Set up hooks interactively with `caliber hooks`, or non-interactively:
183
+
184
+ ```bash
185
+ caliber hooks --install # Enable all hooks
186
+ caliber hooks --remove # Disable all hooks
187
+ ```
188
+
189
+ The refresh command analyzes your git diff (committed, staged, and unstaged changes) and updates your config files to reflect what changed. It works across multiple repos if run from a parent directory.
190
+
191
+ ## 🔌 LLM Providers
192
+
193
+ | Provider | Setup | Default Model |
194
+ |----------|-------|---------------|
195
+ | 🟣 **Claude Code** (your seat) | `caliber config` → Claude Code | Inherited from Claude Code |
196
+ | 🔵 **Cursor** (your seat) | `caliber config` → Cursor | Inherited from Cursor |
197
+ | 🟠 **Anthropic** | `export ANTHROPIC_API_KEY=sk-ant-...` | `claude-sonnet-4-6` |
198
+ | 🟢 **OpenAI** | `export OPENAI_API_KEY=sk-...` | `gpt-4.1` |
199
+ | 🔴 **Vertex AI** | `export VERTEX_PROJECT_ID=my-project` | `claude-sonnet-4-6` |
200
+ | ⚪ **Custom endpoint** | `OPENAI_API_KEY` + `OPENAI_BASE_URL` | `gpt-4.1` |
96
201
 
97
202
  Override the model for any provider: `export CALIBER_MODEL=<model-name>` or use `caliber config`.
98
203
 
204
+ Configuration is stored in `~/.caliber/config.json` with restricted permissions (`0600`). 🔒 API keys are never written to project files.
205
+
99
206
  <details>
100
207
  <summary>Vertex AI advanced setup</summary>
101
208
 
@@ -115,12 +222,30 @@ export GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account.json
115
222
 
116
223
  </details>
117
224
 
118
- ## Requirements
225
+ <details>
226
+ <summary>Environment variables reference</summary>
227
+
228
+ | Variable | Purpose |
229
+ |----------|---------|
230
+ | `ANTHROPIC_API_KEY` | Anthropic API key |
231
+ | `OPENAI_API_KEY` | OpenAI API key |
232
+ | `OPENAI_BASE_URL` | Custom OpenAI-compatible endpoint |
233
+ | `VERTEX_PROJECT_ID` | GCP project ID for Vertex AI |
234
+ | `VERTEX_REGION` | Vertex AI region (default: `us-east5`) |
235
+ | `VERTEX_SA_CREDENTIALS` | Service account JSON (inline) |
236
+ | `GOOGLE_APPLICATION_CREDENTIALS` | Service account JSON file path |
237
+ | `CALIBER_USE_CLAUDE_CLI` | Use Claude Code CLI (`1` to enable) |
238
+ | `CALIBER_USE_CURSOR_SEAT` | Use Cursor subscription (`1` to enable) |
239
+ | `CALIBER_MODEL` | Override model for any provider |
240
+
241
+ </details>
242
+
243
+ ## 📋 Requirements
119
244
 
120
- - Node.js >= 20
121
- - One LLM provider: your **Claude Code** or **Cursor** subscription (no API key), or an API key for Anthropic / OpenAI / Vertex AI
245
+ - **Node.js** >= 20
246
+ - **One LLM provider:** your **Claude Code** or **Cursor** subscription (no API key), or an API key for Anthropic / OpenAI / Vertex AI
122
247
 
123
- ## Contributing
248
+ ## 🤝 Contributing
124
249
 
125
250
  See [CONTRIBUTING.md](./CONTRIBUTING.md) for detailed guidelines.
126
251
 
@@ -135,6 +260,6 @@ npm run build # Compile
135
260
 
136
261
  Uses [conventional commits](https://www.conventionalcommits.org/) — `feat:` for features, `fix:` for bug fixes.
137
262
 
138
- ## License
263
+ ## 📄 License
139
264
 
140
- MIT
265
+ MIT
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@rely-ai/caliber",
3
- "version": "1.5.3",
3
+ "version": "1.5.5",
4
4
  "description": "Analyze your codebase and generate optimized AI agent configs (CLAUDE.md, .cursorrules, skills) — no API key needed",
5
5
  "type": "module",
6
6
  "bin": {