agent: builder model: opus max_turns: 60 budget_usd: 12 —
yaml load, the tree a program reads: aliases expanded within the limit, a duplicate key an error (#928 step 4, second half)Score: #928 step 4, the next step of the package after y21-events.md (which runs just before this brief, in the same checkout; its commit is on your develop). This change touches only packages/yaml and needs no version raise and no CHANGELOG.md entry (ADR-yaml-1 D1).
Read docs/builder-facts.md, ADR-yaml-1 D4, D6, D8 (docs/specification/adr/adr-yaml-1.md), and in packages/yaml/src/main.nika the tree types at the top, Limits, Error, place_of, documents, parse, as_i64, as_f64.
rg -n "pub fn events|pub fn event_line" packages/yaml/src/main.nika shows both (y21 run 4 built). If not, build nothing here: Result: blocked, Found: y21 not built. Rebuild target/fast/nikaia first (cargo build --profile fast -p nikaia): it must hold #1070-#1072 (y24-y26) and #1085 (y28; a value of views handed back through an if/match arm, which load’s resolving will likely write too). If a view-checker refusal stops you, narrow it in two probes (the shape as return x and as let r = x; return r) and report it as Found:; do not rewrite the code around it.
pub fn load(input: ref Bytes; limits: Limits = default_limits()) -> Vec[Node] sync throws: documents (one parse), then each root resolved, one Node per document. The strings stay views of the input (D4, D6): a copied node is its structure copied (.clone() of a Node whose Strings are views is a copy of the views, not of text). The result has no Value::Alias left; anchors and tags stay on the nodes as written.
Error (unknown anchor *name, position of the alias). An alias inside the node its own anchor names (&a [ *a ]) is an Error (alias *a inside its own anchor), as libyaml-based loaders refuse a recursive alias; an earlier, finished &a does not make it legal.10 * <the document's own nodes> + limits.alias_nodes. Keep each anchor’s resolved node together with its node count, and check before copying, so a billion laughs stops at the first alias over the limit without building it. The error names the limit and its value and the alias’s position (alias expansion exceeds <n> nodes (alias_nodes limit <alias_nodes>), wording yours).Error at the second key’s position, whose message also names the first’s line and column (duplicate key at L:C, first at L:C). Keys compare by kind and value: Null = Null; Bool by its bool; Str by text() (so a and "a" are one key); Int by as_i64 (so 16 and 0x10 are one key; an Int that does not fit compares by its text); Float by as_f64 (a NaN equals no key); Seq/Map structurally, element by element and entry by entry in order. Different kinds never match (1 and "1" are two keys).<< is an ordinary key: nothing is merged (D6). A merge function on request is the next brief, not this one.parse and documents stay as they are (the text as written).
Commit.packages/yaml/tests/load.nika + .stdout: an anchor used twice (both copies printed); an alias inside a nested collection; anchors are per document (two documents, the second’s *a unknown); unknown anchor; recursive alias; a billion laughs of 9 levels with 9 aliases each stops with the limit error, and the same with a raised Limits { alias_nodes: .. } small enough to build passes; a Limits lowered so an ordinary anchor fails; duplicate keys: a/"a", 16/0x10, [1, 2] as a key twice, and 1/"1" accepted; <<: *base kept as a key << with the copied mapping as its value. Write the test before the first package run and fill .stdout with --bless in that run (one run is 3-4 minutes); check every line by hand.If a form the package needs is missing or refused in the compiler, the lowering or std (no ADR refusing it), stop: Result: blocked, Found: the gap with a probe of a few lines, Needs: architect decision. Do not write around it (a hand-written deep copy instead of .clone(), a map from anchor to index because a Map[String, Node] of views is refused, are both ways around: report them).
Run 1 built both steps on local branch wip/y27-load (eef3a2e7: main.nika load and helpers, tests/load.nika, tests/load.stdout blessed and checked line by line in a scratch copy) and stopped on #1089: let a = match m.get(k) { null => throw .. x => x } then match a.node { .. } in Loader::expansion lowers without & (rustc E0507, shown as NK1127 on every limits = default_limits()). y29-1089-match-on-a-get-bound-local.md runs just before this brief and fixes it.
Pre-flight instead of the one above: git log --oneline -5 shows y29’s commit (Fixes #1089); rebuild target/fast/nikaia. Then git checkout eef3a2e7 -- packages/yaml/src/main.nika packages/yaml/tests/load.nika packages/yaml/tests/load.stdout (only these three; main.nika of eef3a2e7 is y21’s file plus load, check with git diff HEAD -- packages/yaml/src/main.nika that nothing else differs), target/fast/nikaia check packages/yaml/src/main.nika clean, then the package gate once without --bless: load must pass with the blessed .stdout as it is. Do not rewrite expansion’s match a.node (run 1’s scratch spelling a.node?.clone() is the way around #1089). If the build still fails there, Result: blocked with rustc’s error. Commit the three files (and nikaia.contracts/nikaia.derived if they changed), Part: #928. No --bless unless check is clean (docs/builder-facts.md, Traps already met).
docs/builder-facts.md first. develop in the main checkout; git status at start: other agents’ uncommitted files are not yours, never touch or stage them.git add <files> only; commit after each step. Do not push: the driver runs the suite on your commit and pushes it (CLAUDE.md, Builders commit, the driver tests and pushes).cargo fmt, cargo clippy --all-targets -- -D warnings, scripts/pre-push-check.sh, and the gate of .claude/agents/builder.md (#852).CHANGELOG.md, badges; no new NK code. New tests in Nikaia only; never the full suite. No foreground sleep, no polling; scripts in files.std or the toolchain lacks stops the work: no code around it unless an ADR or Part I-III refuses it (name it). Else Result: blocked, Found: gap + probe, Needs: architect decision (CLAUDE.md, A missing form stops the work).NIKAIA_CACHE_LIMIT=2G; at most two clones with target/ in the machine; delete your own clone’s target/ and target/release/nikaia-scratch at the end; report GB freed.Pushed: line is local <sha> of your last commit. A report without a Pushed: line, or with a placeholder left in (TC, DISK), is not done and cuts the batch (k3a1 2026-10-29: Result: done, nothing committed).NIKAIA_CACHE_LIMIT=2G target/fast/nikaia test --project packages/yaml green, and cargo test --profile fast -p nikaia --test it -- fmt::every_nika_file (it formats every .nika under packages/). For a change only under packages/yaml/ this is the gate the line above asks for; a change under crates/ runs the gate of .claude/agents/builder.md (#852) as well. target/fast/nikaia is built by the session hook; if it is missing, cargo build --profile fast -p nikaia once.Changelog: none, unless the brief changes crates/.Files not to touch: anything outside packages/yaml/ and the brief’s own report, unless the brief names it; never docs/briefs/batch.txt or a brief of another lane.
max_turns: commit every finished step, then report Result: partial with what is left under Needs:. Uncommitted work at the end is lost work (y4 ran out at 110)./home/user/Nikaia/docs/briefs/yaml/y27-load.md.report.md)Fields of docs/architect.md, Briefs and reports: Result: done|noop|partial|blocked (the bare word alone on that line: Result: done, never a summary after it, which cuts the batch, b869a 2026-10-27; Owner: is a question for the owner or none, never a role), Files, Commit, Body (optional, one paragraph for the commit message), Fixes, Pushed, Tests, Gate (green|red), Deviation, Found, Needs, Changelog, Owner, Disk freed, and for a move Rs2nika: (the kept N of M lines of rs2nika --score, then the three costliest hand edits; none for a task that moves no Rust, none (redesign) for a brief with Shape: redesign and no Convert: line). none where nothing, and then the bare word: Deviation: none (a note) cuts the batch (batch.nika compares exactly); a note goes into the commit message. Found: is only for a defect outside your change (it cuts the batch); a process event such as another session resetting or rewriting your files is no Found:, it goes under Needs: (m3a 2026-10-28); a converter draft that came out wrong is no Found:, it is one of the costliest edits under Rs2nika:, which the converter lane reads (e12-4b 2026-10-27); how you solved your own step goes into the commit message. Deviation: is only a step of the brief built otherwise than written or not built (a choice the brief left to you, such as keep what X needs and say so, is no deviation: it goes into the commit message; nor is building two steps in one commit when every step is built, e12-3a 2026-10-27, nor a step another commit on develop already built, shown by your test failing with that commit’s change reverted, b889 2026-10-27: name the commit in Body:, nor a test’s expected .stdout/.stderr taken from what the program really prints where the brief’s sketch of it was short or wrong, b934 2026-10-29: say so in Body:, nor a lowering that follows the ADR governing it where the brief’s expected lowering contradicts that ADR, b933 2026-10-29: name the ADR in Body:, nor a lowering other than the brief’s where rustc cannot take the brief’s, b1029 2026-10-31: say why in Body:, nor re-keying an allow-list entry (KNOWN) whose hash changed for the same program and the same reason, b-fixforward-sweep2 2026-10-28: say so in Body:, nor an existing test’s expected .stdout/.stderr updated because your change rightly changes the message it prints, b1012b 2026-10-31: name the test in Body:); how you verified (e.g. a test not run against the unfixed compiler) goes under Tests:. A noop says why under Tests: (what was probed), not under Deviation:. Final answer, one line: done <sha>, report <path> (or BLOCKED: ... / PUSH REJECTED ...).
Attribution: Co-Authored-By: Claude Opus 5.5 noreply@anthropic.com Claude-Session: https://claude.ai/code/session_01UVxJDveRM5qKZKyCqq6Y24