# Research notes Series: Breathe You by raregonzo, credited on the live Art Blocks page as Shane Richardson. Assignment AB5D-SERIES-083. Package status draft-for-human-review. ## Sources and access Eleven sources were inspected and all eleven are cited. One access failed and is recorded below rather than in the source ledger. Identity was confirmed before research: chain 1, contract `0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270`, project id 75, slug breathe-you-by-raregonzo. The supplied atlas position of 83 was not used as a project id. The supplied metadata snapshot was treated as a lead; every field was re-read live on 18 September 2026. The refreshed values agreed with the snapshot on invocations, maximum, mint chronology, licence label and script type. Artist credit is split across sources. The index gives raregonzo; the live project page credits Shane Richardson and links a profile under that name. Both are recorded. The assignment's dossier lead also points at the Shane Richardson artist page. Failed access: `https://shanerichardson.art`, the artist website recorded on the project. Browser navigation was denied and direct requests failed at name resolution on the access date. No archived copy was pursued. This is a gap in the record, not an inspected source, and it is not cited. No artist interview exists in the index for this project and none was located by search. The artist's project statement on the mint page is the only substantive primary statement available. Access route: the cloud container running this session has no egress to artblocks.io, the Art Blocks GraphQL index, the media proxy or the generator, all of which fail at the outbound proxy with a 403 on CONNECT. All live observation ran through the browser pane on the operator's computer. GraphQL queries were issued as HTTP POST from that browser context. ## Full script inspection Retrieved from `projects_metadata` filtered on the exact slug, with contract address and project id checked in the same response. Length 20,236 bytes. SHA-256 `84a5028bb33a85176168748921c5d94383cecfbd4cc32b322be5590cb2df4998`, computed over the UTF-8 bytes via the Web Crypto API in the browser. The record reports `script_count` 5. Both the single `script` field and the ordered concatenation of the five `scripts` rows were hashed and the digests are identical, so the five stored fragments are consecutive pieces of one program with nothing missing. Coverage: the whole script was read. Three of the five fragments are almost entirely CSS text; those were checked programmatically as well as read, by extracting every class of the form gN and confirming the set runs 1 to 180 with no gaps and no duplicates outside that range. The JavaScript portion was read in full, function by function. Functions traced: the load handler and main generation loop; the five element builders for block, outer blank, outer colour, inner blank, inner colour and pupil; the colour selector switch; the unused size clamp; and the Random class. Mechanism established from the code, recorded as direct script findings and kept separate from the artist's statement: population equals hash byte 16 plus 256; colour regime equals hash byte 15 integer-divided by two, matched against 78, 79, 100 and 42; positions drawn between minus eighty-nine pixels and the square's edge; sizes between two and sixteen percent of the square; five nested elements per circle with paired box shadows; pulse animation on every outer blank disc; rotation applied per circle by a two-way random choice; both animations infinite. Randomness: a xorshift generator seeded by parsing the first sixteen hex characters of the token hash as a Number. That value is outside the exact integer range, so the seed is a rounded double, and the first bitwise shift coerces it to thirty-two bits. The decimal output is the state modulo one thousand divided by one thousand, so there are at most one thousand distinct draw values. Both points are recorded as observations about the program rather than as criticism of the output. External dependencies: the record reports an external asset dependency count of zero, the declared library is `js@na`, and the script contains no canvas reference, no library import, no network request and no audio. It registers no event listeners. Verified by pattern search over the retrieved bytes as well as by reading. Dead code: one size clamp helper is defined and never called. No block reference was pinned; this was an index retrieval. Script content is not redistributed in this package. ## Token identity and visual observation Nine token identities were confirmed on the exact contract before use, plus a full-edition sweep of 256 rows. All four candidate identifiers supplied in the assignment exist, so no substitution was needed for the base sample. Sampling rationale beyond the supplied candidates: the artist's mint page advertises four named rare variants and explicitly leaves open whether any exist. Rather than repeat that claim, the colour regime was computed for every token in the edition from byte 15 of its hash using the script's own switch. That produced exactly eleven special tokens. Four of them were then adopted as examples, one per regime, so that the derivation could be checked visually as well as arithmetically. Two independent confirmations of the derivation were obtained. First, the platform exposes a Special attribute on exactly those eleven tokens and on no others, with values Birth, Death, Golden and Rainbow mapping onto the blank, black, gold and rainbow code regimes. Second, the images show precisely the predicted colour behaviour in each case. Token page checks: five token URLs and the collection URL were requested and all returned status 200 at 2026-09-17T23:24:29Z. No status has been fabricated. Images: eight were loaded at their native 2400 by 2400 pixels through the platform media proxy and viewed. One defect was found and is reported. The stored preview for invocation 4 is an incomplete capture, populated across the top of the frame and blank white across the lower forty-five percent. The live renderer for the same token, allowed to settle, contains all 397 circle blocks, matching the platform's own Circles attribute, with content extending past the container edge. The artwork is complete; the capture is not. Other unsampled tokens in this project may be affected, and the coordinator should verify any published still against the live renderer. ## Motion and audio Motion was measured, not watched, and not inferred from the title or the library. For invocation 233 the live renderer was opened and allowed to settle, then the document was inspected directly. Element counts: 338 circle blocks, matching the indexed Circles attribute; 338 elements carrying the pulse class; 166 carrying the rotation class, that is 49.1 percent against the even split the code implies. Computed animation on a pulsing element read "3s ease-out infinite simplePulse". Its computed transform sampled at three moments gave scale factors of 0.857, 0.962 and 0.909, consistent with the declared 0.85 to 1.0 to 0.85 cycle. Computed rotation of one spinning element sampled at five moments 0.7 seconds apart gave -89.6, -5.6, 78.4, 162.4 and -112.6 degrees, an advance of 84 degrees per 0.7 seconds, which is 120 degrees per second and one full turn in exactly 3.0 seconds against a declared 3 second duration. Audio: no audio or video element exists in the rendered document and no audio call appears anywhere in the script. No sound was heard because there is none to hear. Browser autoplay policy was therefore never engaged, and no autoplay limitation applies to this work. Interaction: the artist script registers no event listeners and the project page documents no controls. Nothing responds to input. A canvas element is present in the generator page. The artist script never references a canvas, so that element belongs to the platform wrapper and not to the artwork. This is noted to prevent a later reviewer from concluding that the work draws to a canvas. ## Header ownership and media The complete current AB5D wallet set was read from the platform's own profile record rather than from the supplied cache: three linked addresses, `0xce6dc14c5494d8a4325a81d3b257d130fa2039d1`, `0x2247247ee0eee59557e6cd14f7e047f6a9309b5c` and `0xdd19308a9731cdecb67e0d80ca6823394208c045`. Token ownership was swept across all ten projects in this batch filtering on those three addresses. For this project the sweep returned token 75000233, invocation 233, held by `0xdd19308a9731cdecb67e0d80ca6823394208c045`. The assignment's cached candidate named the same token and the same owner, and the live check confirms it rather than merely repeating it. Because the work is animated and the platform itself labels it so, the presentation file recommends the live renderer as the header medium rather than a still. The renderer autoplays with no play control. A fallback still URL is recorded but flagged as not representing the work. ## Mint chronology and calculations Events distinguished: first indexed mint 2021-06-03T21:30:52Z; activation 2021-06-04T01:27:22Z; second indexed mint 2021-06-04T12:06:23Z, duplicated in `start_datetime`; completion 2021-07-05T15:27:17Z. Note that the first mint precedes activation by roughly four hours, consistent with an artist proof, though the current owner of invocation 0 is not the artist address and present ownership proves nothing about the original recipient. One calculation is performed: second indexed mint to completion, 31 days, 3 hours, 20 minutes and 54 seconds, covering invocations 1 through 255. Labelled inferred. Explicitly not presented as a public sale duration, because the public opening is not established and `start_datetime` does not mark it independently. The page's displayed release date of 4 June 2021 corresponds to activation rather than to the first mint. Unlike some other projects on this contract, the displayed date and the UTC activation date agree, so no timezone discrepancy arises here. Price: 78000000000000000 wei, that is 0.078 ETH, recorded as the currently configured value and as reported. No primary announcement of an original price was located. The coincidence that the page's present secondary floor is also 0.0780 ETH is noted without interpretation. ## Rights and conflicts The only rights statement located is the index field `license` reading "NIFTY", with `link_to_license` null. The live project page publishes no licence statement; its expanded details panel was read in full. A contemporaneous community explainer published 10 June 2021 identifies NIFTY as shorthand for NFT License 2.0 by Dapper Labs. That is cited as a third-party interpretation, not as a grant by this artist. To test how much weight the label can carry, the `license` field was censused across all 359 projects on this contract. It returns 34 distinct strings, including at least eight that appear to denote the same NIFTY or NFT License family, plus case and whitespace variants, a label that is a bare URL and a label that is a full sentence. Only 87 of 359 projects carry any licence link. The field is unvalidated free text. This is recorded as an unresolved conflict against the artwork rights entry, because the label alone does not identify a document or a version. Artwork rights are recorded as category custom, reported label preserved exactly as "NIFTY", terms url null, status reported rather than verified. Code rights are unknown; no licence statement accompanies the script. Fully on-chain storage establishes readability, not permission. NFT ownership has not been treated as an artwork permission, and the CC0 status of this package's prose has not been extended to the artwork. ## Unresolved questions and exceptions Six exceptions are raised, all pending and assigned to the publication coordinator: missing licence terms behind an uncontrolled label; unknown code rights; the absence of any artist record beyond the mint page; the untested historical priority claim; the unverified original price; and the preview capture defect. The priority claim deserves particular care at publication. The artist states on the mint page that this is the first Art Blocks work to use only HTML, CSS and JavaScript. This research does not test that, because testing it would require auditing the dependency and script of every project released before 3 June 2021. What is verified is narrower and is what the essay says: the program uses no rendering library and creates no canvas. The essay does not assert a historical first. Unresolved leads not converted into sources: the artist website does not resolve; no interview or artist essay was found; the relationship of this project to the artist's later Billows and Ouroboros was not investigated. ## Actual model, tools, elapsed time and measured usage Model: claude-opus-5, served through the Claude desktop application in Cowork mode on the Claude Agent SDK. The batch routing label was "claude", a routing label and not a model identifier. The agent is not Astra. Tools actually used: HTTP POST to the Art Blocks public GraphQL index; a browser pane on the operator's computer for page rendering, text extraction, image inspection, live renderer execution, DOM inspection and computed-style sampling over time; the Web Crypto API for the SHA-256 digest; web search and fetch for the licensing explainer; local shell and file tools for package assembly and validation. No human reviewed or edited the prose. Word count: the coordinator waived the 1,500 to 2,500 word limit for this batch during the session. This essay runs longer than the original cap by that instruction. Elapsed work: produced within a single continuous session on 18 September 2026, Melbourne time. Per-package elapsed time is not instrumented here and is not reported as though it were. Telemetry: no token usage telemetry is exposed to the agent. Input, cached input, output, reasoning and total counts are null in usage.json with the reason given. Nothing has been estimated and labelled measured, and batch totals have not been divided into invented per-essay figures. ## Coordinator code review Complete indexed script independently read. Corrected seed precision, parent-relative sizes, shadow scope, inherited transforms, load handler and conditional probability claims. Original findings preserved in evidence. Source, visual, ownership and rights review remains pending; this is not a publication approval.