---

name: bitcoin-l2-fractal-bitcoin description: | Fractal Bitcoin: a Bitcoin Core-derived scaling chain with ~30s blocks and Cadence Mining (part merge-mined with Bitcoin, part permissionless, part Index Mining since FIP-101). Mainnet 9 Sept 2024. FIP-102 / fractald v0.4.0 hard-forked at block 2,100,000, combining the first and second FB halvings; the chain passed that height before 15 September 2026. USE WHEN: building on Fractal, indexing BRC-20 on Fractal, reasoning about merge-mined Bitcoin-derived chains. allowed-tools: Read, Grep, Glob

Fractal Bitcoin

Not a rollup and not a sidechain with a foreign VM: Fractal runs Bitcoin Core's own codebase, retuned. The launch press release calls it "the only Bitcoin native scaling solution that uses the Bitcoin Core code itself to recursively scale unlimited layers on top", claiming "complete consensus consistency with the Bitcoin network" and "20x increase in capacity per layer".

Mainnet launched 9 September 2024. Node client is fractal-bitcoin/fractald-release; latest tag as of 15 September 2026 is v0.4.0, published 10 August 2026.

Property	Value
Block time	~30s or less
Consensus code	Bitcoin Core derivative
Native asset	FB
Halving interval	2,100,000 blocks
Core contributor	UniSat team

Cadence Mining

Blocks are not all mined the same way. At launch the split was 1:2, merge-mined : permissionless — one block in three carries a merge-mined commitment from Bitcoin miners, the other two are mined permissionlessly on Fractal itself.

FIP-101 (created 2025-12-31, status Deployed, activation height 1,500,000) changed that split to 1:1:1 — Merged Mining : Permissionless Mining : Indexing, introducing a third reward class for operators of the Fractal Standard Indexing Service. It also added Taproot-based non-custodial FB staking: users stake FB through Taproot scripts, funds stay under the user's own key, and stakers can exit or migrate between indexing instances without operator permission. Index verification and reward settlement are explicitly kept off the block-production critical path so indexing cannot stall the chain.

FIP-102 and the double halving

FIP-102 (created 2026-07-28, "Native FB on Bitcoin Mainnet"), implemented by fractald v0.4.0 as a consensus-tightening hard fork at block 2,100,000:

• The first scheduled halving fires at 2,100,000: 25 FB → 12.5 FB.
• FIP-102 brings the second halving forward to the same height: 12.5 FB → 6.25 FB.
• The 6.25 FB that the second halving would have retired is not burned — it is redirected to distribution on Bitcoin mainnet, rolled out progressively over roughly three months.
• Every subsequent reward level is advanced by one 2,100,000-block interval. Max supply and allocation are unchanged; only timing is.

Target allocation after activation:

Path	Share	FB per Fractal block equivalent
Merged Mining	1/6	~2.0833
Permissionless Mining	1/6	~2.0833
Index Mining	1/6	~2.0833
Bitcoin mainnet distribution	1/2	6.25
Total	1	12.5

The three Fractal-side figures are averages: an eligible block pays 6.25 FB, with one block in each three-block cycle going to each mechanism.

The Fractal chain tip was at height 2,121,967 on 15 September 2026, so the fork height has been passed and the combined halving is live. The eligible on-chain interactions and distribution rules for the Bitcoin-mainnet half are deferred to FIP-103, which was not yet published in fractal-bitcoin/fips on that date (the repo held FIP-30, FIP-101, FIP-102 and FIP-104).

Note the release-notes discrepancy: the v0.4.0 body text says "Release Date: September 2026" while the GitHub tag was published 2026-08-10. Trust the tag timestamp.

Upgrade posture

v0.4.0 tightens consensus. Older non-mining nodes will keep following the chain after activation but will not independently enforce the new rules — a silent-downgrade footgun. Miners and pools did not need to change mining software; node operators did need to upgrade.

Use cases

• High-throughput BRC-20 / Ordinals indexing — the reason the UniSat team built it, and why FIP-101 pays indexers directly.
• Merge-mining revenue for Bitcoin pools at no extra hash cost.
• Testing Bitcoin Core-compatible logic at ~30s block times against near-identical consensus code.

Limitations

• Not secured by Bitcoin. Only one block in three is merge-mined; the permissionless and index blocks are secured by whatever hash rate shows up. "Consensus consistency with Bitcoin" is about code, not about Bitcoin validating Fractal.
• No trust-minimised peg — FB is Fractal's own asset, not BTC.
• Governance concentration — FIPs are authored by the "Fractal Team" and the reference indexer is UniSat's. FIP-102 rewrote the emission schedule mid-flight.
• Emission changes are frequent — two consensus hard forks (FIP-101, FIP-102) inside nine months. Pin a version and read the FIP before assuming a reward number.

See also

• counterparty/SKILL.md
• drivechains-spacechains/SKILL.md
• liquid/SKILL.md
• rootstock-rsk/SKILL.md