Consolidate the agent publish/share surface onto the canonical KA lifecycle (#1087) by Jurij89 · Pull Request #1265 · OriginTrail/dkg

Jurij89 · 2026-06-21T01:23:02Z

Summary

One canonical way to publish. Agents now have exactly one path to put knowledge into SWM/VM — the named Knowledge-Asset lifecycle (create → write → seal → share → publish). This deletes the bypass/loose agent tools dkg_publish (direct raw-quad mint), dkg_shared_memory_publish, and dkg_share across all three adapters (MCP, OpenClaw, Hermes), and re-points every internal caller onto the canonical per-KA POST /api/knowledge-assets/:name/vm/publish.
Deletes the legacy bridge route + a latent authz bypass (Track C follow-up: numeric publishContextGraphId preflight bypass #1179). Removes POST /api/shared-memory/publish entirely — both the assertionName and selection forks — and with it the Track C follow-up: numeric publishContextGraphId preflight bypass #1179 numeric-publishContextGraphId branch that trusted a caller-supplied on-chain CG id without an ownership check. The surviving publish routes enforce the ownership check.
Adds an ergonomic, lifecycle-respecting one-shot (D3). Extends dkg_knowledge_asset_create with optional quads + alsoShareSwm so an agent can create → write → seal → share in one call. It is default-false (sharing must be explicit), stops at SWM (never mints to VM), and is byte-parallel across all three adapters — so it cannot reintroduce a one-call bypass.
Migrates the node-UI publish CTAs. The "Publish in SWM" and per-entity "Ratify" buttons now publish per-assertion via /vm/publish (named-only), with a catch 409 → seal-in-SWM → retry-once safety net; this also fixes a latent crash when a CG held ≥2 SWM roots.
Rewrites the agent docs. Both SKILL.md files teach the single canonical flow and scrub every agent-facing loose-SWM/bridge reference; .cursor/rules updated; the skill-endpoint contract test is flipped to assert the bridge route is absent.
Deliberately deferred (out of scope, tracked in Consolidate daemon-level loose SWM writes onto the named-KA lifecycle (post agent-tooling cleanup) #1260). The daemon-level loose-SWM write primitive (/api/shared-memory/write + /conditional-write) is kept — it is the staging buffer the source-worker bulk-ingest + async-LIFT pipeline depends on, and a live consumed channel (dkg_query SWM view, node-UI, gossip/sync). The SWM gossip/sync receive fabric (sync/host routes + handlers) is untouched. Three now-dead but publicly re-exported facade methods (DkgClient.publish, DkgPublisherExtension.publishVerifiableMemory/writeSharedMemory) that still call surviving routes are left for that follow-up.

This change was implemented in waves and gated by a multi-angle adversarial review (plan-conformance, over-deletion, #1179 security, correctness, doc-truth, cross-adapter parity) before opening.

Fixes Post-rc.17: deprecate the legacy /api/shared-memory/publish bridge and consolidate the one-shot publish onto the per-KA lifecycle #1087
Relates to Clarify knowledge asset lifecycle when SWM share succeeds without a seal but VM publish requires finalization #1116 (seal-on-share clarity; already shipped — reflected here in the tooling + docs)
Removes the Track C follow-up: numeric publishContextGraphId preflight bypass #1179 numeric-publishContextGraphId authz-bypass branch (with the deleted bridge route)
Follow-up: Consolidate daemon-level loose SWM writes onto the named-KA lifecycle (post agent-tooling cleanup) #1260 (consolidate the daemon-level loose-SWM writes onto the named-KA lifecycle)

Files changed

File(s)	What
`packages/cli/src/daemon/routes/memory.ts`	Delete the entire `POST /api/shared-memory/publish` route (both forks) + the #1179 numeric-CG authz-bypass branch + the helpers it solely used; keep `/write`, `/conditional-write`, `/memory/turn`, and all sync/host routes
`packages/cli/src/api-client.ts`	`publishFromFinalizedAssertion` delegates to `/vm/publish`; deleted the `publishFromSharedMemory` selection-fork shim
`packages/cli/src/commands/shared-memory.ts`, `benchmark/publish-get/*`	Re-point the CLI publish command + the publish-get benchmark onto the named-KA `/vm/publish` flow
`packages/mcp-dkg/src/tools/publish.ts` (deleted), `tools/assertions.ts`, `index.ts`, `client.ts`	Delete `dkg_publish`/`dkg_shared_memory_publish`; extend `dkg_knowledge_asset_create` (D3); scrub the surviving publish description
`packages/adapter-openclaw/src/{tools/*,DkgNodePlugin.ts,dkg-client.ts,index.ts}`	Delete `dkg_publish`/`dkg_shared_memory_publish`/`dkg_share`; extend create (D3); delete the 2 bridge-calling facade methods
`packages/adapter-hermes/hermes-plugin/{__init__.py,client.py}`, `src/types.ts`, `README.md`	Delete the 3 tool schemas + handlers + dead client methods; extend create (D3); narrow the publish-exposure gate 3→1; system-prompt surgery; keep `dkg_memory`
`packages/core/src/publisher-extension.ts`	Delete the bridge-calling `publishSharedMemory` facade method + its transport member + dead request type
`packages/node-ui/src/ui/{api.ts,views/MemoryLayerView.tsx,views/project/components/ka.tsx,hooks/useMemoryEntities.ts}`	Migrate the publish CTAs to per-KA `/vm/publish` (named-only) with a seal-retry chokepoint; remove dead loose-write code
`packages/network-sim/src/{api.ts,components/ControlPanel.tsx,server/sim-engine.ts}`	Re-point sim publish to `/vm/publish`; delete the loose-publish-by-selection helper + its button
`packages/cli/skills/dkg-node/SKILL.md`, `dkg-importer/SKILL.md`, `.cursor/rules/dkg-annotate.mdc`	Rewrite to the single canonical flow; scrub all agent-facing loose-SWM/bridge references
tests across cli/mcp/openclaw/hermes/node-ui/core	Update/flip tests; flip the skill-endpoint contract; delete the obsolete selection-fork + multi-root tests

Test plan

Notes / known risks

OpenClaw doc refresh: existing OpenClaw nodes must re-run dkg openclaw setup to pick up the new SKILL.md — the daemon serves /.well-known/skill.md live (MCP/Hermes/HTTP get it immediately), but OpenClaw copies the file into the agent workspace at setup.
Deferred / follow-up (intentionally out of scope, for transparency):
- The daemon-level loose-SWM write primitive (/api/shared-memory/write + /conditional-write) — kept; the source-worker bulk-ingest + async-LIFT pipeline depends on it. Consolidating it onto the named-KA model is tracked in Consolidate daemon-level loose SWM writes onto the named-KA lifecycle (post agent-tooling cleanup) #1260.
- devnet / ops validation scripts — the pure-assertionName callers are re-pointed to /vm/publish here (devnet/v10-core-flows/automated.test.ts, scripts/testnet-publish-stress/publish-loop.mjs, devnet/_bootstrap/bootstrap.cjs; response shape verified compatible — same engine as W1). The remaining callers are mixed (assertionName + selection in one file) or pure-selection (loose publish-by-selection) — devnet/v10-stress + rich-scenario + agent-provenance tests, scripts/redistribute-memory.mjs + drain-swm-duplicates.mjs + seed-dkg-code-project.mjs, and the .sh validators. The selection shape has no named assertion to point at, so it needs the named-KA rework (the same loose-SWM boundary as the kept /write); none are in a CI lane. Deferred with the Consolidate daemon-level loose SWM writes onto the named-KA lifecycle (post agent-tooling cleanup) #1260 / devnet-rework follow-up. (A compat handler was rejected — it would resurrect the bridge route + the Track C follow-up: numeric publishContextGraphId preflight bypass #1179 authz-bypass.)
- 2 sibling node-UI publish CTAs (entities.tsx, layer-widgets.tsx) carry the same pre-register pattern fixed in the two migrated CTAs, but it is pre-existing (not introduced here) and shared with the finalize/promote branch — separate analysis.
- 3 now-dead but publicly re-exported facade methods (DkgClient.publish, DkgPublisherExtension.publishVerifiableMemory/writeSharedMemory) that still call surviving routes — left for a future library-API cleanup.
Pre-existing env-gated suites (better-sqlite3 native bindings, live-daemon/real-chain, Windows path-separator, comma-locale render) are unrelated to this PR.

…the assertionName fork W1 of the #1087 agent-tooling cleanup. ApiClient.publishFromFinalizedAssertion now delegates to knowledgeAssetPublish (POST /api/knowledge-assets/:name/vm/publish), so `dkg shared-memory publish`, `dkg index`, and the publishAssertion composite re-point with unchanged signatures + typed returns. The assertionName fork of /api/shared-memory/publish (it duplicated /vm/publish) is removed and replaced with a 400 guard that rejects a stray `assertionName` body (so it can't fall through to the selection fork). network-sim + node-ui e2e publish helpers re-pointed. The selection fork, the #1179 numeric-publishContextGraphId branch, /write, /conditional-write, and all sync/host routes are untouched (handled in later waves). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…share; add the create one-shot [D3] W2 of the #1087 agent-tooling cleanup, across MCP, OpenClaw, Hermes. Removed the three agent tools and their dead client methods: - MCP: deleted tools/publish.ts wholesale + registerPublishTools wiring + dead client.publishQuads/publishSharedMemory (createKnowledgeAsset kept for D3). - OpenClaw: deleted dkg_publish/dkg_shared_memory_publish/dkg_share defs + handlers + tool-host wiring. The dkg-client/publisher-extension publish methods are now caller-less but KEPT (daemon/CLI call — flagged for the daemon wave). - Hermes: deleted the 3 tool schemas/handlers + dead client methods + _annotate_207_partial; narrowed the publish gate 3 tools -> 1 (the _direct_publish_allowed gate preserved for the surviving canonical publish); system-prompt surgery to teach the named-KA path. dkg_memory + shared helpers untouched. [D3] Extended dkg_knowledge_asset_create (x3 adapters) with optional quads + alsoShareSwm/also_share_swm (default FALSE, passed explicitly so the helper's seal-true default can't leak; no alsoPublishVm). One daemon call: POST /api/knowledge-assets {quads, finalize, alsoShareSwm}. Modes: draft / wm-sealed / swm-shared. Tests updated across all 3 adapters incl. D3 3-mode + default-false regression + no-alsoPublishVm guards. MCP 227 / OpenClaw 1098 / Hermes pytest 109 + vitest 90 green (excl. 2 pre-existing MCP inject-hook flakes + the pre-existing part-11 python3 host issue). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…iant at the wire Adds 4 body-level guard tests proving a default/false create never emits alsoShareSwm:true on the wire (parity with the TS adapters, where the createKnowledgeAsset client helper would otherwise force-share on seal). Locks in the no-auto-share invariant against future body-builder changes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…th the TS adapters) Aligns Hermes to the plan's create-tool contract ("ignored when there are no quads") and the two TS adapters: a bare dkg_knowledge_asset_create with also_share_swm but no quads is now a plain WM draft (no error, no share) instead of a hard-reject. All 3 adapters are now byte-parallel; the daemon's non-finalized-share guard remains the backstop. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… loose-SWM/bridge dead code W3 of the #1087 agent-tooling cleanup. - CTA-1 (MemoryLayerView "Publish all/selected in SWM"): data source switched from listSwmEntities (raw SWM roots) to listAssertions(cg,'swm') (named), publishing per-ASSERTION via /vm/publish -- fixes the latent >=2-root "exactly one root entity" throw. - CTA-2 (ka.tsx "Ratify"): named-only -- publishes via /vm/publish when the entity resolves to a named SWM assertion; un-named loose roots get no publish button. - Added knowledgeAssetPublishWithSeal: the single catch->seal->retry chokepoint (catch the /vm/publish 409 PUBLISH_NOT_FULL_SHARE|VM_PUBLISH_PRECONDITION -> wm/finalize{layer:swm} -> retry once; SWM_SUBSET_NOT_SEALABLE -> typed throw, no loop). Extended knowledgeAssetFinalize with the swm layer. - Removed now-dead loose-SWM code: publishTriples, writeSharedMemory, the publishSharedMemory bridge caller, listSwmEntities/SwmRootEntity + 3 helpers, and their tests/mocks/stale comments. None touch the /api/shared-memory/write ROUTE (kept; daemon-level, tracked in #1260). tsc --noEmit clean; ui-api-pure + affected component tests green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…1179 authz-bypass branch W3.5 of the #1087 agent-tooling cleanup -- the bridge is now caller-free (adapters deleted in W2, node-UI migrated in W3). - Deleted the entire POST /api/shared-memory/publish route handler block (the W1 stray-assertionName 400 guard + the selection fork + the #1179 numeric- publishContextGraphId authz-bypass branch -- SECURITY: gone, NOT migrated) + the helpers it solely used (resolvePublishRootEntities & friends, publishResponsePayload, MAX_PUBLISH_EPOCHS) + now-unused imports. - Deleted ApiClient.publishFromSharedMemory (the internal CLI shim for the selection fork) + its test (publish-from-swm-readboth.test.ts). - Migrated the publish-get benchmark's sync leg off the removed loose-write+selection path onto the canonical named-KA publishAssertion. KEPT (verified): engine agent.publishFromSharedMemory (hard dep of /vm/publish), agent.publishFromFinalizedAssertion, route (ii) /api/knowledge-assets/publish, ALL sibling routes (/write, /conditional-write, /memory/turn + sync/host routes), and the 5 now-dead-but-publicly-re-exported facade/transport methods (left for a future library-API cleanup). cli build green; api-client 46/46. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…crub loose-SWM/bridge refs W4 of the #1087 agent-tooling cleanup. - dkg-node + dkg-importer SKILL.md: deleted the dkg_publish/dkg_shared_memory_publish/ dkg_share rows; replaced the "three publish tools / two forks" framing with the single canonical create -> write -> seal -> share -> publish flow; removed the agent-facing /api/shared-memory/{write,publish,conditional-write} prose (now a "non-agent primitive, tracked in #1260" note); reframed route (ii) as CLI/programmatic; [D3] documented dkg_knowledge_asset_create({quads, alsoShareSwm}) (default-false, stops at SWM) as the one-shot; added the consequence note (loose SWM is non-publishable via the canonical path -> author a named KA). KEPT swm/share, dkg_query view:shared-working-memory, and the seal-error rows. - .cursor/rules/dkg-annotate.mdc: dropped dkg_share from the Write list; retired-tools note; human-gated publish -> dkg_knowledge_asset_publish. - Flipped cli/test/skill-endpoint.test.ts (canonical endpoints present; /api/shared-memory/publish absent). - Tidy-ups: 2 now-stale code comments (memory.ts, api-client.ts) + auth.test.ts's route-agnostic sample path -> a surviving route. Release note: OpenClaw copies SKILL.md into the workspace at `dkg openclaw setup`; existing OpenClaw nodes need a re-run to pick up the new docs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… doc-truth parity) Fixes the 2 majors + cosmetic stale comments surfaced by the pre-push adversarial review. - network-sim (MAJOR): deleted publishFromSharedMemory + its "Publish to Chain" ControlPanel button -- it POSTed the now-deleted /api/shared-memory/publish selection fork (TS-silent, would break at runtime). It modeled the removed loose-then-publish-by-selection flow; named-KA publish already lives in the separate Publish tab (-> /vm/publish). SharedMemoryTab is now Write+Query SWM only. - MCP + OpenClaw (MAJOR): stripped "Prefer this over dkg_shared_memory_publish" from the surviving dkg_knowledge_asset_publish description (a deleted tool named in agent-facing copy); replaced with Hermes-parallel positive wording. Restores cross-adapter parity. - Stale-comment tidy: mcp/openclaw handler comments, the CLI shared-memory publish comment, a test doc-comment, and the node-ui VM-graph comment -- repointed off the deleted route/tools. Verified: zero deleted-route/tool refs in network-sim or MCP/OpenClaw src; adapters tsc-clean + targeted tests green; network-sim build green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… the deleted bridge Pre-push follow-up to the review-fix pass. The repo-wide grep showed 2 of the 5 "kept" facade methods still POST the now-deleted /api/shared-memory/publish route, so they would 404 if used -- keeping a public method that calls a deleted route is worse than removing it. - Deleted DkgClient.publishSharedMemory (adapter-openclaw) + its 5 test blocks + the index.ts re-export. - Deleted DkgPublisherExtension.publishSharedMemory (core) + the transport-interface member + the dead SharedMemoryPublishRequest interface + the mock-transport stub; folded the shared test onto its surviving publishVerifiableMemory half. KEPT (call surviving routes): DkgClient.publish + DkgClient.share, and DkgPublisherExtension.publishVerifiableMemory + writeSharedMemory. tsc clean; core 1084/1084, OpenClaw 1093 pass/1 skip. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…eted bridge route CI caught this real-daemon-route test asserting 400 on POST /api/shared-memory/publish (deleted in this PR -> now 404). Repointed to the canonical per-KA /api/knowledge-assets/:name/vm/publish, which rejects an unknown CG with the same CONTEXT_GRAPH_NOT_FOUND via the shared resolveRequiredWriteContextGraphId preflight -- preserving unknown-CG validation coverage (no sibling test covered it on a surviving publish route). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… seal-retry recovery Addresses the PR #1265 review (otReviewAgent). - Removed the ensureContextGraphOnChain pre-call (+ unused import) from both publish CTAs (MemoryLayerView publishAssertions, ka.tsx VerifyOnDkgButton). The daemon /vm/publish route deliberately runs local preconditions FIRST and only registers the CG on the CG_NOT_REGISTERED retry path, rehydrating the STORED publishPolicy. The pre-call burned registration gas on a doomed publish and risked the #1085 default-policy (the UI register call sends no policy). The CTAs now call knowledgeAssetPublishWithSeal directly and let the daemon register at the right time. - Added 3 tests for the recovery contract: 409 PUBLISH_NOT_FULL_SHARE and VM_PUBLISH_PRECONDITION each -> wm/finalize{layer:'swm'} -> retry-once succeeds; SWM_SUBSET_NOT_SEALABLE -> typed throw, no retry loop. tsc clean; ui-api-pure 52/52. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Deep-review caught the package README still documenting dkg_publish + dkg_shared_memory_publish as live (the publish-bridges table, a prose recommendation, and the deleted src/tools/publish.ts package-layout row) with a stale "31 tools" count -- both prior reviews audited SKILL.md + tool descriptions but not the README. Collapsed the publish section to the single canonical dkg_knowledge_asset_publish + a removed-in-#1087 note; fixed the count to 29 (matches the drop-sweep.test.ts locked surface). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…deleted publish route Addresses 2 PR #1265 review comments. - publish-get benchmark: the sync-leg KA name was benchmark-sync-<runId>-<iteration>, but runIteration runs for both warmup and measured phases with the same numeric iteration, so warmup-N and measured-N collided (name/idempotency-based create) and the GET validated the wrong marker. Name now includes the phase (benchmark-sync-<runId>-<warmup|measured>-<iteration>), mirroring the payload's own warmup-/measured- root+marker uniqueness. - context-graph-write-path-validation: added a negative test that POSTs the deleted /api/shared-memory/publish with a VALID CG and asserts 404 (unambiguous route-not-found) -- guards the route deletion against accidental reintroduction. cli tsc clean; the benchmark unique-roots + marker tests pass. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…te the api mocks Addresses 2 PR #1265 review comments. - /api/knowledge-assets/:name/vm/publish returns 207 when the KA minted on-chain but the CG binding failed (contextGraphError in the body; post() treats 207 as ok, so it rides through un-thrown). Typed contextGraphError on PublishResult + knowledgeAssetPublishWithSeal, and surfaced it: ka.tsx Verify CTA renders a warning ("binding incomplete") instead of a clean check; MemoryLayerView counts a `partial` tally, prefers a clean sample, and downgrades the result card to a warning when partial>0. Added a ui-api-pure test (207+contextGraphError resolves, no seal/retry). - Completed the full vi.mock('../src/ui/api.js') in the two component tests that load the components barrel (-> ka.tsx) with the new exports it imports (knowledgeAssetPublishWithSeal + the real SwmSubsetNotSealableError class), avoiding a missing-mock module-load failure. tsc clean; ui-api-pure 53/53; component-mock suites green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Follow-up doc-truth pass from the deep adversarial review: - docs/agent-context/task-packs/publish-query.md + docs/use-dkg/publish-and-query.md: dropped dkg_shared_memory_publish from the live Node Skill tool lists (archive/ docs left as historical record). - mcp-dkg/src/client.ts: reworded the registerContextGraph JSDoc off the deleted publishSharedMemory/register_if_needed flow (vm/publish auto-registers). - knowledge-assets.ts: the migration-alias comment no longer cites the deleted MCP publishQuads / OpenClaw publish callers (the alias remains for publishAssertion + network-sim). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…d wire tests otReviewAgent nit: the 4 new no-leak guard tests passed create_assertion('ka','cg1') positionally, but the signature is (context_graph_id, name) -- so they recorded a reversed contextGraphId/name and only checked quads/alsoShareSwm. Switched to keyword args (context_graph_id='cg1', name='ka') and added assertions on the recorded contextGraphId + name so an arg-order/serialization regression now fails. (The pre-existing write_assertion test uses the distinct (assertion_name, context_graph_id) signature -- left unchanged.) pytest 113/113 (PYTHON=python). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…o false republish advice) otReviewAgent: the round-2 partial-publish warning told the user to republish, but a 207 means the KA is already minted on-chain and re-running publish won't re-attempt the failed context-graph binding (no re-bind endpoint exists). Centralized the copy in api.ts -- partialPublishWarning(contextGraphError) ("Published on-chain, but linking it into the context graph did not complete, so context-graph views may be incomplete" + the daemon detail) + PARTIAL_PUBLISH_STATUS_SUFFIX -- and wired both CTAs (ka.tsx + MemoryLayerView) to the single source. Added a unit test asserting the copy never suggests republish/retry, and stubbed the new exports in both component mocks. tsc clean; ui-api-pure 54/54. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ed") otReviewAgent nit: the part-14 create test's doc comment said also_share_swm WITHOUT quads is "rejected", contradicting the test's own assertion (and the §2.6 contract) that it is IGNORED -- no error, plain draft. Comment-only; aligns the validation evidence with the pinned contract. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…"Requires quads") otReviewAgent nit: the also_share_swm schema description claimed it "Requires quads", contradicting the handler (and the MCP/OpenClaw wording), which IGNORES the flag when no quads are supplied. Reworded to "Default false. Ignored when there are no quads." Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…) param spelling otReviewAgent: the shared dkg-node SKILL documents the D3 flag as camelCase `alsoShareSwm`, but the OpenClaw + Hermes tools expose all fields in snake_case (`also_share_swm`, `context_graph_id`, `sub_graph_name`). An agent copying the example on those adapters would pass a key they ignore -- sealing the asset but silently not sharing it. Added a parameter-spelling note to the Tool-vs-HTTP section (the systemic fix; applies to every param, not just this flag). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

otReviewAgent · 2026-06-21T03:11:44Z

-    }
-  }
-
  // POST /api/shared-memory/conditional-write  { contextGraphId, quads, conditions, subGraphName? }


🔴 Bug: Removing the /api/shared-memory/publish handler makes that endpoint fall through to 404, but active integration/devnet callers in this repo still post to it with the finalized-assertion shape ({ contextGraphId, assertionName }), for example devnet/v10-core-flows/automated.test.ts and multiple devnet/bootstrap scripts. Those flows will now fail before reaching the new /api/knowledge-assets/:name/vm/publish path. Either keep a compatibility handler for the assertion-name publish shape or update the remaining active callers in the same change.

Partly taken, partly explained — thanks, this is a fair push.

Re-pointed (6c0203c) the pure-assertionName callers — the ones with no selection sibling, so they become fully working — to /api/knowledge-assets/:name/vm/publish: devnet/v10-core-flows/automated.test.ts (the file named here), scripts/testnet-publish-stress/publish-loop.mjs, and devnet/_bootstrap/bootstrap.cjs. The .test.ts assertions hold unchanged — /vm/publish returns kaId/status/txHash/blockNumber via the same engine publishFromFinalizedAssertion as the deleted fork (verified statically; the W1 in-CI ApiClient.publishFromFinalizedAssertion → /vm/publish migration is the proof the response shape is compatible).

Still deferred (tracked with #1260): the remaining callers are either mixed (an assertionName and a selection POST in the same file — re-pointing only the assertionName half leaves the file 404-ing on the selection call) or pure-selection (loose publish-by-selection). The selection shape has no named assertion to point at, so migrating it is the named-KA rework this PR deliberately scopes out — it's the same loose-SWM boundary as the kept /api/shared-memory/write. These are devnet/ops tooling; none are in a CI lane.

On the compatibility-handler alternative — declining it deliberately: keeping an assertionName handler would re-introduce the /api/shared-memory/publish route this PR's core purpose is to delete, and the #1179 numeric-publishContextGraphId authz-bypass lived in that exact handler. The whole devnet/ops validation surface wants one coherent named-KA re-validation pass rather than a resurrected bridge; that's the deferred follow-up noted in the PR description (now updated to reflect the 3 re-pointed files).

…deleted bridge route otReviewAgent (escalated): the deleted /api/shared-memory/publish left active devnet callers 404-ing. Re-pointed the 3 PURE-assertionName files (no selection sibling, so they become fully working) to /api/knowledge-assets/:name/vm/publish, mirroring the W1 ApiClient migration: - devnet/v10-core-flows/automated.test.ts (:283, :448) - scripts/testnet-publish-stress/publish-loop.mjs (:299) - devnet/_bootstrap/bootstrap.cjs (:353, :375) Response shape verified statically: /vm/publish returns kaId/status/txHash/blockNumber via the same engine publishFromFinalizedAssertion as the deleted fork, so the test assertions hold unchanged. The remaining callers are MIXED (assertionName + selection in one file) or pure selection (loose publish-by-selection) -- they need the named-KA rework deferred to #1260; a compat handler is rejected (it would re-introduce the bridge route + the #1179 authz-bypass). Devnet-only, not in CI. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

otReviewAgent · 2026-06-21T03:27:18Z

+  // daemon must no longer recognize the route at all (route-not-found 404),
+  // rather than reach a handler.
+  it('no longer serves the removed /api/shared-memory/publish route (404)', async () => {
+    const res = await postJson(daemon, '/api/shared-memory/publish', { contextGraphId: LOCAL_CG });


🔴 Bug: This removal is not matched by a complete caller migration. Several non-archive devnet/validation paths still POST to /api/shared-memory/publish (for example scripts/v10-rc-validation.sh, devnet/agent-provenance/automated.test.ts, and devnet/rich-scenario/automated.test.ts), so after this 404 assertion those flows will fail rather than exercising the new per-KA /api/knowledge-assets/{name}/vm/publish route. Please migrate or deliberately retire those callers in the same cleanup so the route deletion is internally consistent.

Jurij Skornik and others added 9 commits June 21, 2026 01:03