diff --git a/CLAUDE.md b/CLAUDE.md
index 648db4d2a..e6ba079de 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -97,3 +97,67 @@ Network and Lightning code is heavily async (asyncio). Background tasks run as c
 - Implementation sweeps: `@needs_test_with_all_aes_implementations`, `@needs_test_with_all_chacha20_implementations`
 - `FAST_TESTS = True` skips slow implementation variants
 - Test vectors live alongside tests as JSON files (e.g., `tests/test_vectors/`)
+
+---
+
+## BitcoinPurple (BTCP) Support
+
+BitcoinPurple is a Bitcoin fork with 1-minute blocks and a shorter difficulty retarget window. The Electrum client has been extended to support it as a first-class network alongside mainnet and testnet.
+
+### Key Parameters (differ from Bitcoin)
+
+| Constant | Bitcoin | BTCP |
+|----------|---------|------|
+| `DIFFICULTY_ADJUSTMENT_INTERVAL` | 2016 blocks | **120 blocks** |
+| `POW_TARGET_TIMESPAN` | 1,209,600 s | **7,200 s** |
+| `MAX_TARGET` | `0x00000000ffff…` (compact `0x1d00ffff`) | `0x00000ffff…` (compact `0x1e0fffff`) |
+| `POW_GENESIS_BITS` | `None` (derived from MAX_TARGET) | **`0x1e0ffff0`** (genesis nBits differ from powLimit) |
+| `MAX_ADJUSTMENT_FACTOR` | 4 | 4 |
+
+`POW_GENESIS_BITS` is non-`None` for BTCP because the genesis block's `nBits` (`0x1e0ffff0`) is stricter than the `powLimit` compact (`0x1e0fffff`). `get_target(-1)` in `blockchain.py` returns `bits_to_target(POW_GENESIS_BITS)` when this is set, so period-0 header verification passes the exact equality check.
+
+### Retarget Clamping (BTCP only)
+
+At each 120-block boundary:
+```
+actual_timespan = clamp(last.time - first.time, 1800, 28800)   # [7200/4, 7200*4]
+new_target = old_target * actual_timespan / 7200
+new_target = min(new_target, MAX_TARGET)
+```
+
+### Critical `blockchain.py` Logic
+
+`verify_chunk()` has two paths:
+- `adj_interval == CHUNK_SIZE` (Bitcoin, 2016): reads old targets from on-disk headers.
+- `adj_interval < CHUNK_SIZE` (BTCP, 120): retargets happen *within* a chunk, so headers are still in RAM (the buffer being verified). Uses an internal `_read_hdr(data, i)` helper to read from the buffer instead of `read_header()`.
+
+### Files
+
+| Path | Purpose |
+|------|---------|
+| `electrum/constants.py` | `BitcoinPurple` and `BitcoinPurpleTestnet` classes with all chain params |
+| `electrum/blockchain.py` | `verify_chunk`, `get_target`, `can_connect` — all generalised for per-chain PoW constants |
+| `electrum/chains/bitcoinpurple/servers.json` | Known ElectrumX servers (TCP 51001, SSL 51002) |
+| `electrum/chains/bitcoinpurple/checkpoints.json` | SPV checkpoints (one entry = one 2016-block chunk) |
+| `tests/test_bitcoinpurple.py` | 46 BTCP-specific tests (address encoding, difficulty, header verification) |
+| `tecnichal-data.md` | Complete BTCP parameter reference (ports, genesis, PoW, LN, ElectrumX) |
+
+### Running with BitcoinPurple
+
+```bash
+# Mainnet
+./run_electrum --bitcoinpurple
+
+# Testnet
+./run_electrum --bitcoinpurple_testnet
+
+# BitcoinPurple tests only
+pytest tests/test_bitcoinpurple.py -v
+
+# Blockchain + bitcoin + BitcoinPurple together
+pytest tests/test_blockchain.py tests/test_bitcoin.py tests/test_bitcoinpurple.py -v
+```
+
+### LN Block-Scaled Timeouts
+
+BTCP has 1-minute blocks (10× faster than Bitcoin). All LN timeout values expressed in blocks must be scaled ×10 to preserve the same real-world security windows (e.g. `to_self_delay` 144 → 1440, `cltv_expiry_delta` 40 → 400). See `tecnichal-data.md §8.3` for the full table.