No description
  • Rust 98.4%
  • C++ 1.2%
  • Python 0.4%
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
Meta Vens f566fe6bb3 Re-derive what the batch derived, and stop re-raising an earlier record
Two rules Soul's follow-up pass falsified, both in admission.

A batch that derives a write could not be replayed. `derive` set every
derived sequence to `latest + 1` over an image that already held the first
admission's derived record, so the second admission produced a different
document, A8 refused it -- `AlreadyResolved` on the source side, and
`AlreadyStewarded` on the receiving one -- and A9 never saw the digest it
would have recognised. `derive` now reproduces the record it already made
when image or batch holds one, found by what the derivation writes into it: a
resolution `Answered` by this ruling, a withdrawal whose reason is this
hand-off's key, an assignment whose note is. The same document comes back
byte for byte, A9 answers with the first receipt, and only a batch that never
landed takes `latest + 1`.

A withdrawal could re-raise an earlier record over a later one, which is the
rationale Q19 A's cap rests on and the cap alone does not carry. Withdrawing
a resolution puts its subject back in force, so a repo handed away and handed
back could have the hand-off's withdrawal withdrawn and stand under two
assignments at once; a supersession withdrawn while the revision that
superseded it stood left two revisions in force, which is what the finding
rule reads for its invariant vocabulary. `WouldReinstateOverLater` refuses
that: the record the withdrawal would reinstate, against anything in force
and later in the scope that record is sequenced in -- `(instance, repo)` for
a stewardship, the subject for a resolution, the campaign for a target and
the cut for a cut spec.

Four tests, one per fix and kind: the ruling's resolution and both sides of a
hand-off replayed to `AlreadyAdmitted` with the first receipt and no second
receipt in the store; the retaken repo and the superseded revision refused,
with the later record still the only one in force.

Four more tests for rules the code already held and the suite could not tell
from weaker ones: a sequence gap refuses on both sequenced kinds rather than
only a taken sequence; the latest sequence counts the batch as well as the
image; the Q19 cap reads a chain landing in one batch; and the hand-off's
second withdrawal of one assignment takes the next sequence, not `1`.

41 tests, 0 warnings.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-16 20:21:04 +02:00
.voidbot Document Huginn repo swarm wiring 2026-06-21 15:51:46 +01:00
crates Re-derive what the batch derived, and stop re-raising an earlier record 2026-09-16 20:21:04 +02:00
tools Pin the reopened subject, the retaken repo, and six blind spots 2026-09-16 19:44:15 +02:00
.gitignore Add the empty Rust workspace 2026-09-16 15:11:05 +02:00
AGENTS.md Close the epoch record, the minted leaf refusal and the nested resolution 2026-09-16 17:53:08 +02:00
Cargo.lock Take the sequenced keys and make in force mean what it says 2026-09-16 19:28:17 +02:00
Cargo.toml Add the empty Rust workspace 2026-09-16 15:11:05 +02:00
README.md Pin the reopened subject, the retaken repo, and six blind spots 2026-09-16 19:44:15 +02:00

Huginn

Huginn is the GameCult memory organ. It owns each agent instance's mind as typed state: what an instance remembers, which of that memory is admitted to steer its next action, and the provenance of every admitted claim.

Upstream: https://github.com/GameCult/Huginn.git

Authority

  • An instance owns its mind; Huginn owns the state. Huginn is the single writer of an instance's memory documents. No other service, script, or agent writes them.
  • Memory documents are CultCache state. huginn-mind, the one live crate, persists them through CultLib's Rust CultCache into a redb store at <state_root>/minds/<instance>/mind.redb. Publishing them over CultNet is huginn-daemon's, and that crate is a stub: nothing here publishes yet.
  • Retrieval is to depend on Qdrant directly, refusing loudly when Qdrant is unreachable rather than falling back to a second store or a second writer. No crate opens that connection yet.
  • Mind state is not version-controlled. It lives in Huginn's store, not in any repository.

Generic .cc inspection is not Huginn's job. CultCache Studio inspects and edits .cc state. Huginn reads and writes minds.

Layout

A Cargo workspace of three crates:

  • crates/huginn-mind: storage, identity, and admission of memory documents.
  • crates/huginn-daemon: a stub. It will carry the CultNet surface and serve loop; today it holds neither.
  • crates/eureka-state: a stub. It will carry typed state for the Eureka pipeline; today it holds none.

huginn-mind is live: it opens one instance's store (an owned redb CultCache at <state_root>/minds/<instance>/mind.redb, locked for the mind's lifetime), and refuses a store whose instance document names another instance, whose epoch record is foreign, or whose types are not a mind's. It admits batches of pipeline documents through one commit path: the leaf's bounds, formats and keys, then the organ's cross-document rules (references, in-force status, the resolution matrix, derived resolutions and stewardships), then one compare-and-swap that lands the batch whole with a receipt naming the exact bytes it read and wrote. An exact replay answers with the stored receipt. Document shape and keys come from epiphany-pipeline; the store is CultLib's Rust CultCache. huginn-daemon and eureka-state are stubs; the campaign's cut map in Epiphany/notes/eureka-pipeline-state-cut.md owns what each crate must do next.

cargo check --workspace

Persona

Huginn's legacy repo Persona lives under .voidbot/. Its state/huginn.cc holds legacy void.* documents and its voice/identity.json names the Persona. The two disagree about jurisdiction; that migration belongs to the portable-Persona work, not to this workspace, and nothing here reads or writes .voidbot/.