crawlio-browser 1.6.0 → 1.6.2
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.
- package/README.md +11 -11
- package/dist/mcp-server/{chunk-CSSOZILQ.js → chunk-T4GKS2PG.js} +1 -1
- package/dist/mcp-server/index.js +4007 -406
- package/dist/mcp-server/{init-HPQIML7U.js → init-PFND5ZFY.js} +10 -3
- package/package.json +2 -2
- package/skills/browser-automation/SKILL.md +81 -10
- package/skills/browser-automation/reference.md +3 -2
- package/skills/clone/SKILL.md +101 -0
- package/skills/compare/SKILL.md +102 -0
- package/skills/dossier/SKILL.md +146 -0
- package/skills/extract/SKILL.md +67 -0
- package/skills/investigate/SKILL.md +97 -0
- package/skills/monitor/SKILL.md +64 -0
- package/skills/test/SKILL.md +101 -0
package/README.md
CHANGED
|
@@ -5,7 +5,7 @@
|
|
|
5
5
|
|
|
6
6
|
## [Documentation](https://docs.crawlio.app/browser-agent/overview) | [API Reference](https://docs.crawlio.app/browser-agent/tools) | [Chrome Extension](https://www.crawlio.app/browser-agent)
|
|
7
7
|
|
|
8
|
-
MCP server that gives AI full control of a live Chrome browser via CDP.
|
|
8
|
+
MCP server that gives AI full control of a live Chrome browser via CDP. 114 tools with framework-aware intelligence, typed evidence infrastructure, tracking pixel analysis, technographic fingerprinting, SEO auditing, and confidence-tracked findings — captures what static crawlers can't see.
|
|
9
9
|
|
|
10
10
|
> **Note:** This repo supersedes [`crawlio-browser-mcp`](https://github.com/AshDevFr/crawlio-browser-mcp). All development now happens here.
|
|
11
11
|
|
|
@@ -29,7 +29,7 @@ That's it. Auto-detects and configures 14 MCP clients: Claude Code, Cursor, VS C
|
|
|
29
29
|
|
|
30
30
|
```bash
|
|
31
31
|
npx crawlio-browser init # Default: code mode, stdio transport
|
|
32
|
-
npx crawlio-browser init --full # Full mode (
|
|
32
|
+
npx crawlio-browser init --full # Full mode (114 individual tools)
|
|
33
33
|
npx crawlio-browser init --portal # Portal mode (persistent HTTP server)
|
|
34
34
|
npx crawlio-browser init --cloudflare # Add Cloudflare MCP (89 tools, no wrangler)
|
|
35
35
|
npx crawlio-browser init --dry-run # Show what would happen
|
|
@@ -160,7 +160,7 @@ The JIT Context MCP Runtime is a layered execution architecture where each layer
|
|
|
160
160
|
│ │ ↳ Absorbs: connection drops, tab refreshes, │ │
|
|
161
161
|
│ │ port conflicts, extension lifecycle │ │
|
|
162
162
|
│ ├────────────────────────────────────────────────────────────┤ │
|
|
163
|
-
│ │
|
|
163
|
+
│ │ 147 RAW COMMANDS (bridge.send) │ │
|
|
164
164
|
│ │ CDP-level browser control via Chrome extension │ │
|
|
165
165
|
│ └────────────────────────────────────────────────────────────┘ │
|
|
166
166
|
└─────────────────────────────────────────────────────────────────┘
|
|
@@ -206,7 +206,7 @@ The JIT Context MCP Runtime is a layered execution architecture where each layer
|
|
|
206
206
|
| **Framework awareness** | None | None | 17 namespaces, injected JIT |
|
|
207
207
|
| **Action resilience** | Model must handle timing | N/A (no DOM) | Built-in actionability polling + settle delays |
|
|
208
208
|
| **Error recovery** | Re-call individual tool | Re-create isolate | Re-execute against same live state (Agentic REPL) |
|
|
209
|
-
| **Multi-step patterns** | Model improvises | Model writes loops |
|
|
209
|
+
| **Multi-step patterns** | Model improvises | Model writes loops | 17 tested higher-order methods + behavioral protocol |
|
|
210
210
|
|
|
211
211
|
[Read the full architecture guide →](https://docs.crawlio.app/browser-agent/overview)
|
|
212
212
|
|
|
@@ -214,7 +214,7 @@ The JIT Context MCP Runtime is a layered execution architecture where each layer
|
|
|
214
214
|
|
|
215
215
|
### Code Mode (3 tools) — default
|
|
216
216
|
|
|
217
|
-
Collapses
|
|
217
|
+
Collapses 114 tools into 3 high-level tools with ~95% schema token reduction:
|
|
218
218
|
|
|
219
219
|
| Tool | Description |
|
|
220
220
|
|------|-------------|
|
|
@@ -230,7 +230,7 @@ const screenshot = await bridge.send({ type: 'take_screenshot' }, 10000);
|
|
|
230
230
|
return screenshot;
|
|
231
231
|
```
|
|
232
232
|
|
|
233
|
-
### Full Mode (
|
|
233
|
+
### Full Mode (114 tools)
|
|
234
234
|
|
|
235
235
|
Every tool exposed directly to the LLM. Enable with `--full`:
|
|
236
236
|
|
|
@@ -425,15 +425,15 @@ When a framework is detected, the smart object exposes framework-specific helper
|
|
|
425
425
|
|
|
426
426
|
## Method Mode
|
|
427
427
|
|
|
428
|
-
Method Mode is a domain layer built on top of Code Mode. It adds higher-order methods, a typed evidence system, and a behavioral protocol to the `execute` sandbox — without changing the tool surface. The model still sees three tools. The same `smart` object. The same
|
|
428
|
+
Method Mode is a domain layer built on top of Code Mode. It adds higher-order methods, a typed evidence system, and a behavioral protocol to the `execute` sandbox — without changing the tool surface. The model still sees three tools. The same `smart` object. The same 147-command catalog underneath. What changes is what happens *inside* `execute`.
|
|
429
429
|
|
|
430
430
|
### The Maturity Ladder
|
|
431
431
|
|
|
432
432
|
| Layer | Optimizes For | Behavioral Variance | Evidence Quality |
|
|
433
433
|
|-------|---------------|---------------------|-----------------|
|
|
434
|
-
| **Raw MCP** (
|
|
434
|
+
| **Raw MCP** (114 tools) | Completeness | High — flat tool list, no composition guidance | None — unstructured text |
|
|
435
435
|
| **Code Mode** (3 tools) | Token efficiency | Medium — right primitives, ad-hoc composition | None — model-defined shapes |
|
|
436
|
-
| **Method Mode v1** (+
|
|
436
|
+
| **Method Mode v1** (+ 17 methods + protocol) | Consistency | Low — proper methods, protocol constraints | Convention — `{ finding, evidence, url }` |
|
|
437
437
|
| **Method Mode v2** (+ typed evidence + gaps + confidence) | Correctness | Minimal — typed schemas, tool-enforced findings | Structural — typed records, gap tracking, confidence propagation |
|
|
438
438
|
|
|
439
439
|
### Architecture
|
|
@@ -462,7 +462,7 @@ Method Mode is a domain layer built on top of Code Mode. It adds higher-order me
|
|
|
462
462
|
│ │ Framework Namespaces [up to 17, injected JIT] │ │
|
|
463
463
|
│ │ react · vue · angular · nextjs · shopify · ... │ │
|
|
464
464
|
│ ├──────────────────────────────────────────────────────┤ │
|
|
465
|
-
│ │ bridge.send() —
|
|
465
|
+
│ │ bridge.send() — 147 raw commands │ │
|
|
466
466
|
│ └──────────────────────────────────────────────────────┘ │
|
|
467
467
|
└────────────────────────────────────────────────────────────┘
|
|
468
468
|
```
|
|
@@ -631,7 +631,7 @@ Multi-framework detection returns a **primary** framework (meta-framework takes
|
|
|
631
631
|
## Tools Reference
|
|
632
632
|
|
|
633
633
|
<details>
|
|
634
|
-
<summary><b>All
|
|
634
|
+
<summary><b>All 114 tools</b> — Connection, Capture, Navigation, Network, Storage, Emulation, Tracking, SEO, and more</summary>
|
|
635
635
|
|
|
636
636
|
### Connection & Status
|
|
637
637
|
|