docs: use Python/Node.js bindings in SDK integration examples

[aphoristicartist]

Summary

All SDK integration examples were using subprocess.run() or CLI commands instead of the actual Infiniloom Python (pip install infiniloom) and Node.js (npm install infiniloom-node) bindings. Fixed to use native bindings for better DX and performance.

README.md

Before:

import subprocess
context = subprocess.run(["infiniloom", "pack", ".", "-f", "xml", ...], capture_output=True, text=True).stdout

After:

import infiniloom
context = infiniloom.pack(".", format="xml", model="claude", compression="balanced")

Same fix applied to LangChain example: CLI → infiniloom.embed() with JSON parsing.

Claude Code guide — new section

"Using the Python Binding with Claude SDK" with working examples for infiniloom.pack(), build_index(), analyze_impact(), get_diff_context(), embed().

Codex guide — new section

"Using the Node.js Binding with Vercel AI SDK / Mastra" with:

• Vercel AI SDK streaming example using pack() from infiniloom-node
• Mastra agent tool example using createTool with infiniloom-node
• Examples for pack, scan, embed, buildIndex, analyzeImpact, getDiffContext

Test plan

• Python examples use functions that exist in infiniloom package
• Node.js examples use functions that exist in infiniloom-node package
• Install commands are correct (pip install infiniloom anthropic, npm install infiniloom-node ai @ai-sdk/openai)

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[github-actions]

Benchmark Results

                        time:   [21.374 µs 21.393 µs 21.413 µs]
                        thrpt:  [467.02 Kelem/s 467.44 Kelem/s 467.85 Kelem/s]
                        time:   [85.760 µs 85.884 µs 86.043 µs]
                        thrpt:  [116.22 Kelem/s 116.44 Kelem/s 116.60 Kelem/s]
                        time:   [37.299 µs 37.318 µs 37.340 µs]
                        thrpt:  [1.3390 Melem/s 1.3398 Melem/s 1.3405 Melem/s]
                        time:   [109.68 µs 109.82 µs 109.96 µs]
                        thrpt:  [454.73 Kelem/s 455.30 Kelem/s 455.87 Kelem/s]
                        time:   [103.09 µs 103.16 µs 103.23 µs]
                        thrpt:  [1.9373 Melem/s 1.9387 Melem/s 1.9400 Melem/s]
                        time:   [210.30 µs 211.10 µs 212.19 µs]
                        thrpt:  [942.54 Kelem/s 947.43 Kelem/s 951.03 Kelem/s]
file_reading/sequential time:   [227.56 µs 227.84 µs 228.13 µs]
                        time:   [99.780 µs 100.49 µs 101.47 µs]
                        time:   [95.918 µs 96.226 µs 96.540 µs]
                        time:   [100.54 µs 100.90 µs 101.26 µs]
                        time:   [130.52 µs 132.97 µs 136.58 µs]
                        time:   [621.57 ps 621.70 ps 621.88 ps]
                        time:   [621.60 ps 621.74 ps 621.94 ps]
                        time:   [621.81 ps 622.06 ps 622.37 ps]

Benchmarks run on Ubuntu runner. See artifacts for full results.