Anthropic described Scion, its version control system, on October 13, saying it uses “derived data” to make operations such as blame more efficient. Sunshowers.io author Drew (whose Oxide candidate materials are cited) says the same concept appeared in his 2022 description of Mononoke, a source control server he helped build. He argues the similarity goes beyond independent insights: Anthropic credited Sapling for “hard-won lessons” but did not mention Mononoke, which he says borders on plagiarism.

The author says the Mononoke team at Meta developed many of the ideas that Claude later applied, and that the team had worked on the design in 2017–18. He says he does not object to Anthropic copying Mononoke’s design, noting that open source is intended to spread useful ideas. He also points to Vicent Marti’s blog post for Cursor as an example of a system with similar insights but key design differences.

The author argues that as LLMs help design systems, they are likely to draw heavily on existing designs. He says Mononoke is open source and included in every LLM’s training set, and that Opus 5.5 and GPT-6 Astra both identify similarities to Mononoke when asked about prior art. He calls crediting earlier work easier and more important than ever, adding that there is no longer an excuse not to do so.

Anthropic described Scion, its version control system, on October 13, saying it uses “derived data” to make operations such as blame more efficient. Sunshowers.io author Drew (whose Oxide candidate materials are cited) says the same concept appeared in his 2022 description of Mononoke, a source control server he helped build. He argues the similarity goes beyond independent insights: Anthropic credited Sapling for “hard-won lessons” but did not mention Mononoke, which he says borders on plagiarism.

The author says the Mononoke team at Meta developed many of the ideas that Claude later applied, and that the team had worked on the design in 2017–18. He says he does not object to Anthropic copying Mononoke’s design, noting that open source is intended to spread useful ideas. He also points to Vicent Marti’s blog post for Cursor as an example of a system with similar insights but key design differences.

The author argues that as LLMs help design systems, they are likely to draw heavily on existing designs. He says Mononoke is open source and included in every LLM’s training set, and that Opus 5.5 and GPT-6 Astra both identify similarities to Mononoke when asked about prior art. He calls crediting earlier work easier and more important than ever, adding that there is no longer an excuse not to do so.