
Kaspa’s core developers pushed new work on DAGKnight, the consensus protocol planned to replace GHOSTDAG, over the weekend. Pull request #1168 was merged into the dagknight branch of the rusty-kaspa repository at 11:17 p.m. ET on Sunday, Nov. 15. A second pull request, #1172, which implements the protocol’s “applied” tiebreaking rule, was opened Sunday morning and remains under review.
GitHub records show #1168 is the first pull request merged into that branch since Sept. 8. The same weekend, the main branch gained a formal security policy that describes how to report vulnerabilities and how community-funded bug rewards are paid. No activation date for DAGKnight has been announced.
What DAGKnight is, and what changed this weekend
Kaspa is a proof-of-work network that uses a BlockDAG. Blocks created at the same time are not discarded as orphans, as on Bitcoin. They are kept and ordered. Today that ordering is done by GHOSTDAG, which relies on a fixed parameter, k, chosen to fit an assumed worst-case network delay.
DAGKnight is designed to remove that fixed assumption. Instead of hard-coding a delay bound, it is meant to adjust to how the network is actually performing. Developers and supporters have linked it to plans for higher block rates. Kaspa currently runs at about 10 blocks per second; api.kaspa.org data showed 864,505 blocks in the 24 hours to Monday afternoon.
PR #1168, titled “[DK] DK store rename and BFS coloring semantics,” was written and merged by the GitHub user coderofstuff. It changes 16 files, with 805 lines added and 724 removed. It renames the DagknightStore to ConflictZoneColoringStore and returns the code to coloring “the full zone bottom-up to prepare for follow-up related to shortcuts and coloring changes,” according to the PR description.
PR #1172, “Tiebreaking applied,” was opened by the GitHub user biryukovmaxim at 8:31 a.m. ET on Nov. 15. It touches 3 files, adding 627 lines and removing 324.
The description says it “replaces the Algorithm 4 witness tiebreak with the applied rule,” and changes which competing sides of the DAG reach the tiebreaking step. Tiebreaking decides the outcome when two competing groups of blocks are closely matched. The dagknight branch is now 130 commits ahead of master and 21 behind it, according to GitHub’s compare view.
Community accounts tracked the changes within hours. “DAGKnight just moved again,” wrote Kaspius Maximus (@KaspiusMaximus).
The post continued: “Overnight, PR #1168 was merged into rusty-kaspa’s dagknight branch: the first DK merge since September. Next up: #1172 ‘Tiebreaking applied’.” It added: “The goal: consensus that adapts to real network latency. No activation date yet.”
Daniel_Kaspa (@Daniel_Kaspa_KR) summarized #1172 as “OPEN | CI PASSED | NOT MERGED,” and wrote that “when competing block groups are closely matched, DAGKnight needs a deterministic way to choose.”
The pace is worth stating plainly. Before this weekend the branch had gone more than two months without a merged pull request, and two other DAGKnight test PRs (#1131 and #1132) were closed without being merged. Research-heavy consensus work tends to move in bursts. Holders waiting on a launch date have none to point to.
A security policy for the node software
On the main branch, PR #1171 added a SECURITY.md file. It was authored by GitHub user someone235 and merged by michaelsutton at 9:58 a.m. ET on Nov. 15.
The policy asks researchers to report vulnerabilities privately through GitHub’s security advisory tool and not to discuss them on GitHub issues, X, or Discord before a fix is coordinated. It also writes down a community-funded reward process. Maintainers classify confirmed bugs as Critical, High, Medium, or Low, with Critical “intended for consensus-critical vulnerabilities.” After a fixed release and a waiting period for node operators to upgrade, they open a 14-day community fundraiser and split the proceeds among researchers.
That model has a clear limit. Payouts depend on what the community donates, so researchers are not promised a fixed bounty.
Why it matters
For node operators and exchanges, DAGKnight is a future consensus change. When it ships it will need a hard fork and a coordinated upgrade, as Crescendo and Toccata did. Nothing in this weekend’s changes requires action from node operators.
For builders, DAGKnight is relevant because many plans for faster block rates depend on it. Applications that rely on quick confirmations, such as the covenant-based apps that appeared after the June 30 Toccata hard fork, would be affected by any change in block timing.
For security researchers, the new policy gives a documented route for reporting bugs in the node software and a defined, if donation-based, reward path. It also sets an expectation that bugs will not be aired publicly before a fix is out.
What to watch next
- Review and merge of PR #1172, and any further “[DK]” pull requests on the shortcut and coloring changes that #1168 prepares for.
- Michael Sutton (@michaelsuttonil), a Kaspa core researcher, is due to appear on Vlad Costea’s livestream on Tuesday, Nov. 17, at 2 p.m. ET. “He will talk about his research work at Kaspa, covenants, ZK stuff & more!” Costea (@Vladcostea) wrote. Any timeline comments on DAGKnight would be news.
- A testnet deployment of DAGKnight, which would come before any mainnet activation.
Sources
- PR #1168: https://github.com/kaspanet/rusty-kaspa/pull/1168
- PR #1172: https://github.com/kaspanet/rusty-kaspa/pull/1172
- PR #1171 (SECURITY.md): https://github.com/kaspanet/rusty-kaspa/pull/1171
- SECURITY.md: https://github.com/kaspanet/rusty-kaspa/blob/master/SECURITY.md
- dagknight vs master compare: https://github.com/kaspanet/rusty-kaspa/compare/master…dagknight
- @KaspiusMaximus: https://x.com/KaspiusMaximus/status/2122268924620218540
- @Daniel_Kaspa_KR: https://x.com/Daniel_Kaspa_KR/status/2122135214084534367
- @Vladcostea: https://x.com/Vladcostea/status/2122050088256278855
- Toccata activation (Kaspa docs): https://docs.kaspa.org/toccata
- Kaspa network data: https://api.kaspa.org/info/blockdag