Commit graph Huginn/crates/huginn-mind
Author SHA1 Message Date
Meta Vens
5074181ebf wire: SearchSources/Sources, SourceQuery with its door, SourceHits (source-search-forge-public)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-31 06:25:25 +01:00
Meta Vens
577cae23a7 huginn-mind: a persona re-put no longer poisons reads; the admission index covers pipeline documents only
The persona document is replaced by design (compare-and-swap on updatedAt), so
its second put is a second receipt naming one identity. AdmissionIndex::build
read that as a doubled A10 write and refused every view and query. The persona
type has no admission facts there now; Mind::persona reads its own receipt. A
pipeline document written by two receipts still refuses.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-31 00:46:22 +01:00
Meta Vens
9f8b927cf2 Type the rows closure of the read-path test
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 15:04:16 +01:00
Meta Vens
2023c0a1e3 Let tests replace the read path under the gate so the view comparison is pinned
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 15:03:47 +01:00
Meta Vens
4c00208f8b Plant the changed-key fault with its receipt renamed so only the re-key check refuses it
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:59:10 +01:00
Meta Vens
9d77c18546 Drop the test-module pipeline_key import the gate's move to the parent module left unused
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:55:09 +01:00
Meta Vens
242f5a7ef5 Re-encode the read-back fault payload as a wider array header, not a map the decoder refuses
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:54:09 +01:00
Meta Vens
92fb906b07 Bend stored questions through the leaf's array encoding in the read-back fault tests
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:52:32 +01:00
Meta Vens
3d6f9179d0 Aim the read-back fault tests at the last question, not the last pipeline document
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:51:42 +01:00
Meta Vens
ac139b4785 Decide the read-back gate in Mind::read_back and pin its checks with committed fixture tests
The gate body moves out of the ignored test into one function over a Mind, now
also refusing a receipted pipeline write with no stored document. The ignored
test is a thin caller; CI plants each fault on a 135-document store.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:50:50 +01:00
Meta Vens
c5fe12ce11 Drop the test-module import the deleted migration test left unused
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:12:27 +01:00
Meta Vens
27921e5653 Add Reader::view and the one stored-document read-back gate in huginn-mind
Mind::view becomes validate_ref then Reader::at(head).view(id); Reader::view is
its lookup, so a caller holding one Reader views many documents from one
snapshot. stored_documents_read_back (ignored, HUGINN_MIND_SNAPSHOT) decodes,
validates, rekeys, re-prepares and views every stored document through one
Reader: O(N) where the deleted per-document Mind::view loop was O(N^2).

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:03:11 +01:00
Meta Vens
fe9069e528 Delete the migration-era read-back tests and the leaf's overlapping real-store gate
mind::tests::snapshot_admits_and_reads_back (an O(N^2) per-document Mind::view loop
plus the six idunn-watchdog resolutions and over-64-byte withdrawals of the
2026-10-01 migration) and eureka-pipeline tests::stored_documents_read_back go,
with the leaf README's real-store section. The one gate follows in huginn-mind.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-26 14:02:51 +01:00
Meta Vens
af3efcad40 Members walk: leave descending and cursor to the Selection default
With projection gone the walk spreads Selection::default(); an explicit
descending: false was a mutant no test could tell from the default.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 22:21:58 +01:00
Meta Vens
2a98dcdff2 Members walk test: the role hop uses resolutions of a cut spec (a question cannot be superseded)
The previous commit's test admitted a Superseded resolution of a question,
which admission refuses. Cite r2 through subject and superseded_by instead.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 22:17:30 +01:00
Meta Vens
8c98b7b028 Members walk: drop the dead projection assignment; pin the cites role through members
The walk never reads projection (select reads it only in validate and the
cursor digest), so the assignment and its "headers only" comment decided
nothing. A new test cites Q1 through two roles by rows of one schema and
asserts exact members and matched for subject, superseded_by and no role.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 22:16:07 +01:00
Meta Vens
15d8efe8ab Members walk test: expected ids in the mind's order (admission batch, then key)
The previous commit's test expected plain key order and fails; this one builds
and corrects it.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 21:34:42 +01:00
Meta Vens
05347b5417 Members walk test: 599 members, so a page of 199 takes 4 pages and fails
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 21:33:27 +01:00
Meta Vens
dccf920030 Mind::members builds its walk inline and counts its pages; the tests observe the production walk
walk_selection is deleted: members is the only place the walk is spelled.
Members carries the number of select pages taken. The page-size test calls
members with caller limit 1 and document projection over 590 questions and
asserts the exact ids and exactly 3 pages. A new test pins that the cites
hop limits members and matched to the cited rows.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 21:32:56 +01:00
Meta Vens
abd2890834 Mind::members: the walk's page size is the mind's by construction
The walk selection is built field by field from the caller's filters, so a
caller's limit, projection, order or cursor cannot reach it (one select per
member at limit 1 measured 300 ms against 14 ms). A test pins it.
Unverified at this commit: built and tested once at the end of the batch.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 21:05:02 +01:00
Meta Vens
f13138c3fc Mind::members: drop the projection override the walk never read
cargo-mutants showed deleting it survives: ids come from rows whatever the
projection, so the override decided nothing.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 20:17:47 +01:00
Meta Vens
874dacde38 Test helper rank_all: the local ranked no longer shadows it
The first commit's huginn-mind test module did not compile (helper shadowed
by a local binding, and a mechanical rewrite had touched the helper body).

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 19:52:28 +01:00
Meta Vens
d7b1b1d116 A semantic query returns up to top_k members of its selection, with scores
The mind decides membership (Mind::members) before the index is asked; the
index scores exactly the members (Qdrant has_id + exact) and the page carries
one cosine score per row. Deletes OVERSAMPLE_MAX, the x4 oversample, the
unfiltered VectorIndex::search contract and check_semantic.

Not yet compiled when committed: the first Yggdrasil run follows.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 19:51:29 +01:00
Meta Vens
f8eaf50a03 campaign-repos: a campaign's repo set changes by target revision; one line_of decides supersession lineage (squash of eureka-substrate/campaign-repos c0a1f2c..4442e1a)
target.repos additive at epoch v2; Docs::repos_in_force owns membership; line_of owns lineage (targets per campaign, cut_specs per campaign+cut, stewardships per instance+repo, rulings and follow_ups global); rival_in_force owns one-assignment-in-force. Soul s5 holds (eureka-substrate:verdict:cut-campaign-repos.s5).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-22 23:38:20 +02:00
Meta Vens
5da8b06a6d Merge remote-tracking branch 'origin/main' into eureka-body/mind-launch 2026-10-10 19:16:19 +02:00
Meta Vens
126a914abb Restore LF line endings in the files my last three commits rewrote as CRLF
The repo stores LF; a scripted edit on Windows wrote CRLF. No content change:
git diff -w d5fddc3 shows only the three commits' real edits.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 10:58:14 +02:00
Meta Vens
6df2a684ef one_persona_per_mind test: a planted row is unreadable on status and door alike
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 10:20:39 +02:00
Meta Vens
6f483bbe02 Persona status goes through persona(): one decision of readability
whoami's persona field derived from stored_updated_at, which ignores the
writing receipt, while persona_get requires one. Status now matches on
persona(), so a receiptless row is Unreadable on both. The test asserts
agreement for every planted row and for a put-written row (restoring the
is_err equality weakened in c5af029).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 10:15:44 +02:00
Meta Vens
c5af029d20 Tests follow PersonaStatus: the MCP whoami answers in it, a receipt-less planted row is not a door read
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 08:53:05 +02:00
Meta Vens
ea28b8cbc0 Decoder reports field path and error class, never a caller value; status says Unreadable
The shared request decoder (envelope.rs) no longer quotes decoder text or
envelope fields: a refusal names the field path and the class. MindStatus.persona
is PersonaStatus {Absent, Stored, Unreadable}, so whoami answers in the persona
door's own terms.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 08:32:48 +02:00
Meta Vens
e95e2d57aa Merge origin/main (f003801) into huginn-persona-state
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 08:30:51 +02:00
Meta Vens
35c5b7ac24 Merge origin/main (f003801) into eureka-body/mind-launch
Mechanical adaptation: testkit's PipelineCutSpec builder gains the new reads field.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 06:50:46 +02:00
Meta Vens
8f59d4850a Persona: a put may rewrite identity fields, and every change lands
Pins identity-write-all-visible: publicName, publicDescription, presentation
and values change through put_persona with no per-field rule and no rate
limit beyond the monotone updatedAt.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 06:35:25 +02:00
Meta Vens
ef7c5e5f2f Persona: the view's receipt is the newest one that wrote the document
The match on type, key and bytes was three conditions for one fact; the
newest receipt writing the persona identity is that fact. Tests pin the
declared-name walk and the receipt of each of five consecutive puts.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 06:09:42 +02:00
Meta Vens
bf32a4955e Restore LF line endings in the files this cut touched
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 06:06:01 +02:00
Meta Vens
6fde91fe0a Persona: compile fixes and the regenerated wire schemas
The jsonschema error kind lives in jsonschema::error; the eureka-state tool
closures own their input; huginn.mind_request/response.v1 carry the persona
operations (derived by the existing published_wire_schemas test).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 05:47:00 +02:00
Meta Vens
e426ef33c1 mind-body launch: AlreadyLive's holder is viewed before it answers Busy; huginn-mind pins the holder by its literal key
The holder RunId is grammar-checked only, so the launch asks the mind for its
in-force Self-run holders and refuses a name the mind does not show as this
instance's own. One reader of the slot serves step 0 and this check. The
admission test compares the refusal's run to the literal run key.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 05:46:16 +02:00
Meta Vens
dbdb4e86f8 Huginn holds an instance's persona document, validated and receipted
put_persona replaces the mind's gamecult.persona_state.v0 whole through the
commit primitive: schema-valid against the published file, personaId is the
instance, canonical provenance, CAS on updatedAt, monotone updatedAt. The
document is an opaque seventeenth stored type, one per mind. Wire operations
persona_get and persona_put, MindStatus.persona, HuginnClient methods and the
eureka-state persona tools. Published wire schemas follow in the next commit.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 05:30:48 +02:00
Meta Vens
7375687188 Pin Rust 1.95.0 and clear the twelve clippy lints red at main
rust-toolchain.toml owns which rustc and clippy judge the tree. View in
HuginnMindResponse is boxed so the large_enum_variant allow on the wire
type is deleted; the PipelineDocument allow keeps a sharpened reason.
Type aliases, is_multiple_of, slice::from_ref and a literal refused()
clear the rest. Wire bytes and schemas are unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 05:21:13 +02:00
Meta Vens
228d1a8893 huginn-mind: name the run kind the live check asks about
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 05:17:43 +02:00
Meta Vens
5ef4f5eef6 huginn-mind: refusals name a holder by a typed RunId; a run born closed owes no live slot
AlreadyLive and AlreadyClaimed carry RunId, which parses as a run key on
every door, so a peer cannot fill it with an arbitrary string. The live
check applies only to a run the batch leaves in force. Tests pin a holder
that differs in start, host and budget, a born-closed run, and the second
live Persona turn's admission refusal in mind-body.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-10 05:17:11 +02:00
Meta Vens
11af610871 huginn-mind tests: import RunTurn for the one-live tests
The previous commit (ff259a1) does not build its tests: RunTurn was not in
scope in admission.rs's test module. This commit builds.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 22:50:41 +02:00
Meta Vens
ff259a16bb huginn-mind: admission refuses a second live run of hers of one turn (AlreadyLive)
Ruling one-live-self-run, decided where batches are serialized: run_rule
(was run_claims_rule) now also refuses an opening or reinstated run whose
instance and turn already have an in-force operated_by Mind run. The
fixture run() is an operator's run; mind_run() is hers. Unverified: this
commit has not been built when written; verification follows.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 22:48:01 +02:00
Meta Vens
821bab62ac Clippy at the base: allow large_enum_variant on two huginn-mind enums, drop a let-and-return
cargo clippy -D warnings failed at df7507b on PipelineDocument (masking these downstream); the two enums
are returned once per commit or request, and admission.rs resolution matrix returned a let binding. Builds.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 22:17:41 +02:00
Meta Vens
ea9c324663 Launch primitive: queue, breaker, open_and_launch and the run aliases
Adds the turn and operated_by aliases on the run kind, and in mind-body the
Launcher port with SystemdLauncher, the work queue, the repetition breaker and
open_and_launch. Not yet built: this commit may not compile.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 20:51:07 +02:00
Meta Vens
44c010c9da Leaf: CodeLocation.symbol and CutSpec.reads, additive at epoch v2 through the value_types! = absent marker
Adds Absent, Symbol, ReadAnchor and read_window_is_bounded, the marker in
value_types!, the workspace fixtures and the reads admission test. The
golden file and the six published schemas are not regenerated yet, so the
golden, schema and wire-schema tests fail at this commit (it builds).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 20:29:53 +02:00
Meta Vens
e1b0a5c6ff Runs: reinstatement shares the opening claim rule, liveness included
run_claims_rule is the one check for a run's claims (in force, held by no
other live run), called by opening and by the withdrawal that reinstates.
Tests: contested second claim on reinstatement, resolved claim refused and
live claims allowed, a repeated non-first claim refused at validate.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 16:47:10 +02:00
Meta Vens
0f02914b57 Runs: pin that a changed run under an admitted key is not a replay
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 16:30:37 +02:00
Meta Vens
f1a4449300 Runs: reinstating a run goes through claim exclusivity; a run claims an item once
Withdrawing a run's closure now asks docs.claim_holder, the same owner the
opening rule uses, and refuses AlreadyClaimed when another live run holds a
claim. Duplicate claims within a run are refused at validate. Tests pin the
reinstatement sequence, second-claim holders, and same-key different-content
IdentityCollision.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 16:30:14 +02:00
Meta Vens
bb9e64dca1 Mind: run admission with AlreadyClaimed, the run resolution row, the claims edge
A run's instance must be this mind. Every claim must name a document in force
(MissingReference, CitesResolvedDocument); a claim held by another run in force
is refused AlreadyClaimed { item, run }, so the claim is the one consumption
fact. A byte-identical replay of an admitted run owes its claims no fresh
freedom and answers AlreadyAdmitted. A run resolves Recorded or Withdrawn only.
Runs cite their claims under role `claims`, so cites and cited selections reach
them; the header facts carry label, operator, turn, start and claims.

The published schemas that embed PipelineKind (finding, follow_up, question,
resolution, ruling) and the two mind wire schemas are re-derived: the Run
variant, PipelineFacts::Run, AlreadyClaimed and the claims role. The leaf commit
before this one has those five stale.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 16:07:19 +02:00