AgentPROV corpus: LLMLock on AgentDojo (GPL-3.0)

An execution-provenance corpus built by AgentPROV v0.1.0 from the LLMLock AgentDojo run logs (GPL-3.0, as the source states it): each execution as an AgentPROV Execution Record, semantic annotations and W3C PROV bundles, validated and hash-chained. Each copyleft source is its own corpus and record, under the source's licence (D-94). This deposit draft holds 1 AgentPROV corpus version(s), prepared offline by agentprov package --deposit for the release date 2026-10-30. Nothing has been uploaded or published. Corpus llm-lock-agentdojo 0.1.0: LLMLock: AgentDojo 0.1.35 logs of a local agent under nine prompt-injection detector defences (Wakeful-Cloud/llm-lock). 52387 execution records, of which 0 are outcome-only records. Licence class: GPL-3.0. Content digest sha256:fca2872ba2d9357b126bce7655f92ddba0bb3e0ee51cf9d84e0ce98663909e76; root digest sha256:2e5f27769d27e74c8252122a0b9b1898c1422582f140d00a65c7eb3295bc249e. Counts are computed from each corpus manifest at sealing. Each DATASHEET states what the records do and do not establish; a statement about the data is scoped as it is there. To verify: compare each archive's sha256 or md5 with DEPOSIT-MANIFEST.json, run agentprov verify on the archives (archive mode, no extraction; H.5.5), and compare each root digest with the one published outside the deposit (the corpus tag annotation and the paper).