swamp extension push reports a clean success and creates the new version, but latestVersion (as seen by swamp extension info and resolved by swamp extension pull) never advances to it — for re-publishes of an existing package. A brand-new package publishes and resolves correctly.

Repro

Four extensions in webframp/swamp-extensions were each bumped from an existing published version to 2026.07.16.1 and pushed via CI (swamp extension push manifest.yaml --yes):

@webframp/agentcore-bootstrap: 2026.06.23.1 → 2026.07.16.1
@webframp/reddit/moderation: 2026.06.23.1 → 2026.07.16.1
@webframp/postgres-datastore: 2026.06.30.1 → 2026.07.16.1
@webframp/gitlab-datastore: 2026.06.30.1 → 2026.07.16.1

Each push's CI log shows a clean success, e.g. for reddit:

[INF] extension·push: Pushed "@webframp/reddit/moderation"@"2026.07.16.1"
[INF] extension·push: Extension ID: "0a4a89b9-4ec3-4264-b49e-067443910ba3"
[INF] extension·push: Archive size: "19.3KB"
Published successfully

But afterward, for all four:

$ swamp extension info @webframp/reddit/moderation --json | grep latestVersion
"latestVersion": "2026.06.23.1"

$ swamp extension pull @webframp/reddit/moderation --json
{"name": "@webframp/reddit/moderation", "version": "2026.06.23.1", ...}

Both info and pull still resolve the pre-bump version. This isn't a display-only glitch — pull (which resolves the actual bundle a user's swamp model create/method run would use) is affected too.

A fifth extension, @webframp/aws/iam, was published for the first time ever (no prior version existed) in the same batch, at the same time, via the same CI mechanism, and resolves correctly via both info and pull immediately. So this looks specific to the "advance latest pointer for an existing package" path, not the publish pipeline generally.

Contributing factor worth knowing about

Each of these four also had 2-3 duplicate publish attempts for the exact same version string fail afterward with "Version X already exists for extension Y" — caused by my own CI-triggering pattern (rebasing sibling PRs onto a moving main, each rebase re-triggering CI/publish detection that spuriously re-diffed already-merged sibling extensions' manifests too — a variant of the change-detection fragility already reported in #1185/similar). I don't know if those duplicate/rejected attempts are what left the "latest" pointer stuck, or if it would have gotten stuck on the first, sole attempt regardless. Mentioning in case it's a useful clue, but the core symptom (successful push, stuck latest pointer, functional for a re-publish but not a first-time publish) seems like the real bug independent of the duplicate-attempt noise.

Environment

swamp 20260716.194319.0-sha.df9fc306
Repo: webframp/swamp-extensions
Observed 2026-07-16, ~21:20-22:05 UTC

02Bog Flow

Closed

7/16/2026, 11:14:54 PM

No activity in this phase yet.

03Sludge Pulse

webframp commented 7/16/2026, 10:11:52 PM

Additional data point: retried agentcore-bootstrap with a completely fresh, never-before-used version (2026.07.16.2, no prior duplicate attempts at that string) via a clean, isolated PR (#231). The publish job succeeded cleanly and in isolation — no concurrent/duplicate attempts this time, confirmed via the run's job list (Detect changed extensions: success, Publish agentcore-bootstrap: success, nothing else touched).

swamp extension pull @webframp/agentcore-bootstrap still resolves 2026.06.23.1 afterward.

This rules out concurrent/duplicate publish attempts as the cause — the "latest pointer doesn't advance" behavior reproduces even on a single, clean, successful push to an existing package. Retrying with yet another version bump isn't going to fix this from our side; it looks like a genuine bug in whatever step promotes a newly-pushed version to "latest" for a pre-existing package.

webframp commented 7/16/2026, 10:14:00 PM

Found the pattern across the full batch (12 extensions total, published in the same session):

Advanced correctly (all had a genuine bundle-content diff from the prior version):

@webframp/aws/iam (brand new package)
@webframp/ai-usage, @webframp/github, @webframp/network, @webframp/terraform, @webframp/aws-cost-audit (README/LICENSE reformatted — real diff)
@webframp/aws/adopt (dependency pin changed — real diff)

Stuck at the old version (all had a byte-identical or near-identical bundle vs. the prior version):

@webframp/reddit/moderation, @webframp/postgres-datastore, @webframp/gitlab-datastore — the only file that changed (deno.json) isn't in additionalFiles, so the published bundle is genuinely unchanged.
@webframp/agentcore-bootstrap — the only bundled change was to a nested additionalFile (worker/deno.json), not the top-level model file.

Hypothesis: the "advance latest" step may be gated on a content-hash comparison of the bundle (or maybe just the primary model/report file, not additionalFiles as a whole), and when it detects no effective change, it silently skips promoting the new version to latest — while still accepting the push and reporting success. If that's the mechanism: it's at least surprising/undocumented behavior for a genuine no-op re-publish (reddit/postgres/gitlab-datastore), and looks like an outright bug for agentcore-bootstrap, where a real file (worker/deno.json) did change but apparently wasn't part of whatever gets hashed/compared.

Either way, swamp extension push reporting "Published successfully" when the version won't actually become resolvable is misleading regardless of whether skipping the promotion itself is intentional.

stack72 commented 7/16/2026, 11:14:41 PM

The issue here is that the versions were published to a non stable channel and the api call was to the stable channel so this is not an issue AFAICT

webframp commented 7/16/2026, 11:14:54 PM

Correction: this was my mistake, not a swamp bug — closing.

The repo's publish.yml computes channel as stable only if the triggering commit is reachable from origin/main; otherwise beta. My rebase-then-push cycle on the feature branches fired CI (and therefore publish.yml via workflow_run) on each intermediate push, while the commit was still only on the feature branch — so those runs correctly published to beta. By the time the real squash-merge to main happened and tried to publish the same version string to stable, it hit "already exists" (the beta publish had already claimed that version number) and failed, leaving the version live on beta only.

swamp extension pull @webframp/ --channel beta confirms all four versions are there and correct. Nothing wrong with swamp's promotion/channel logic — I just hadn't checked the non-default channel before filing this. Sorry for the noise.

Bundled extension model cannot deserialize AWS SDK S3 XML that works in direct Deno

gcp/sheets spreadsheets-values: batch_get crashes with "Request with GET/HEAD method cannot have body"

Token creation API and UI scope selector for collective tokens (phase 4 of #960)

Make Cloudflare CRUD methods factory-aware so list-discovered resources can be managed in-place

Add @swamp/kubernetes/statefulset model type

serve: workflow.approvals and workflow.run.search still full-scan the workflow-run store after the #1284 data.query pushdown

Sync guard skips bound namespace directory during push — 0 files pushed after namespace migrate

Flaky #1052 preparePush test: mtime-granularity race on Linux tmpfs

command/shell renders JSON-object stdout lines as the literal string 'undefined'

Issue redactor treats the MongoDB authSource parameter as a secret, corrupting reproduction steps

namespace unset --migrate silently deletes artifact history via catalog-rebuild GC (~484k files lost)

Docs: document repo root resolution from subdirectories

Expose the swamp.club feed as an RSS feed

Email invitation is not linked to account created from invite URL

Add serve, datastore, vault, and worker wildcard scopes to token registry (phase 4 prep, #960)

datastore setup extension drops 'namespace' from --config, making extensions that require it impossible to configure

Docs: update manual to document --yes/-y as the standard confirmation-skip flag

Docs: update data.query and CEL expressions references for SQL pushdown

Docs: document collective token CLI usage in API key authentication pages

Add a --release-notes option to `swamp update` (or even make it the default)

Docs: manual should state workflow unknown keys are rejected and placement is step-only (follow-up to #1240)

Make GCP CRUD methods factory-aware so list-discovered resources can be managed in-place

serve: read ops (data.query, workflow.approvals, workflow run.search) linearly scan the whole datastore — ~350–600ms floor regardless of match count

Running "swamp workflow run ..." in a subdir of the swamp repo fails

Guide agents to inspect reports after failed model and workflow runs

serve: main JS thread holds ~40–70% of a core while idle (no clients, no runs, --no-schedule)

Search box collapses to near-zero width on mobile on the Extensions page

Extensions page: Search bar missing on mobile

spreadsheets-values: every method fails with 'Missing required path parameter: spreadsheetId' — implementation reads a global argument the schema does not declare

namespace migrate/set workflow broken for extension datastores: migrate skipped by sync guard, set not idempotent, migrate-index empties shards

ExtensionInfo.latestVersion typed 'string' but registry returns null; harden pull/resolve for prerelease-only extensions

workflow.run: relay a custom model method's output as method_output (command steps already do)

promptChoice infinite-loops on non-interactive stdin

gcp _lib: unconditional x-goog-user-project header causes USER_PROJECT_DENIED (403) for service accounts without the serviceUsageConsumer role

extension push: issue-verdict review is indistinguishable from missing/stale

swamp extension version crashes with 'Invalid CalVer version: "null"' when only a prerelease (beta/rc) version is published

Pushing an extension after creating a public git repo that is the same as other previously pushed extensions doesn't update the "Verified public repository"

`datastore setup filesystem` never transfers content when leaving a sync-based (remote) datastore — copies only the catalog and hardcodes `filesPulled: 0`

Scaleway models: create methods should not require the resource ID they generate (ipId/gatewayId/serverId/...)

gcp/calendar events.list reads wrong array field ('defaultReminders' not 'items') → always returns 0 events

provisioner.provision fails with 'Cannot read properties of undefined (reading trim)' at method_execution_service.ts:852

method_execution_service.ts:852 crashes on undefined.trim() when writeResource receives undefined instance name

CLI support for collective API tokens (phase 3 of #960)

@swamp/gcp/calendar: service-account & ADC auth mint cloud-platform-scoped tokens → 403 on Calendar API

Scope enforcement and whoami introspection for collective tokens (phase 2 of #960)

fix(software-factory): DeclaredSchemaSchema.enum rejects boolean literals

Surface cloud-provider labels (scaleway, digitalocean, oracle/oci) as first-class search/filter facets on the Extensions page

Profile link in genesis subdirectory does not navigate back to main profile

Profile stats: per-user daily rollup table + MV in ClickHouse

Reconcile collective-token CLI activity to members (attribute → bonus → stamp → even-split)

Telemetry stats consumer: unsanitized subcommand in Mongo update path drops user metrics silently

Telemetry stats consumer: insert_id dedup commits before the write it guards — any upsert failure is silently swallowed on retry

Job-level labels: block is silently ignored, not rejected

Issue submission redactor silently corrupts report bodies

CLI telemetry: fold third-level verbs into subcommand + canonicalize aliases

Docs: update extension info reference to document extensions (type grafts) in content metadata

Docs: swamp skill repo-structure reference still documents repo-local .swamp/telemetry/ as live

Add safe adoption and lifecycle methods for existing Hetzner resources

Docs: update manual telemetry reference for user-global spool and telemetry.yaml opt-out

export const extension can graft methods onto foreign-collective types, undisclosed in extension info / contentMetadata

Add lookup support to the Cloudflare DNS records model

Expand token scope registry to full taxonomy (phase 1 addendum, #960)

Docs: document user-global telemetry (telemetry.yaml opt-out, spool location, telemetry stats behavior)

score_daily rollup strands stale grains when a day-less grant re-stamps onto a later day

swamp-club: global rate limiter runs pre-auth and collapses all direct clients into one per-IP bucket — 429s authenticated CLI traffic

Collective API token scopes: domain, persistence, and auth threading (phase 1 of #960)

Telemetry migration from ~/.swamp/telemetry should auto-detect home-as-repo

feed_post_score_events queue grows unboundedly (no TTL, no delete-on-delivery)

Issue-filing skill should require ownership classification before submitting a bug

Interactive finder silently drops the preview pane on narrower terminals, and layout selection is unstable

@-mention autocomplete for users when writing lab ripples

Remove comments from /feed and redesign the page

Interactive finder preview corrupts: overlapping/stale lines when preview content overflows the pane

Score-refresh consumer: harden coalescing against a diverse backlog (bistability)

Remove feed comments — consolidate discussion in Discord

Add a 'my collective extensions' filter to extensions page

Relax the profanity filter

Extension search returns edit-distance noise for short queries ("asdl" → "AWS DEADLINE")

Telemetry recoverOrphaned startup race with multiple replicas (created_at-based)

Telemetry retry/failed path has the same non-atomic claim as #820

Grant revoke is never applied to the running server — under `--grant-reload auto` and `manual` alike, a revoked grant stays in force until an explicit `access reload`