golem15/msd-core
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
5,862 Commits 4 Branches 2 Tags
4aba50db8d2326d306ff682b3b3efeff140aa29a
Commit Graph

6 Commits

Author SHA1 Message Date
Tom Boucher
1110c3b4ee fix(#4709): stop minting the retired gemini runtime id in shipped surfaces (#4711) 
* test(#4709): assert no shipped surface mints a retired runtime id

Extends the #1928 removal guard to the surfaces it structurally could not
reach. Its own docblock scopes it to the installer CLI contract and the
runtime-name-policy exports; it spawns the installer and inspects module
exports, and never reads gsd-core/workflows/**, commands/** or skills/**.

Four structural assertions, all RED on next:
- every RUNTIME= assignment must name a canonical runtime
- the runtime->model-tier table must name only model-catalog runtimes
- runtime selection menus must offer only canonical runtimes
- config-set runtime / model_profile_overrides examples must be canonical

Structural, not textual: each asserts the literal is canonical or the runtime
exists as a catalog key, never that the string "gemini" is absent. That string
is load-bearing across Antigravity's real on-disk contract, so a fifth test
pins that contract from the descriptor (not from a resolved path, which would
read $ANTIGRAVITY_CONFIG_DIR and the real $HOME -- the #4312 defect class).
An over-broad gemini -> antigravity replacement fails there rather than ships.

The menu assertion is scoped by the nearest preceding `question:` matching
/runtime/i, because the same file carries a provider menu (anthropic, openai)
and a budget menu (high, medium, low) whose labels are single lowercase tokens
too and name neither a runtime nor anything the policy should judge.

Refs #4709

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(#4709): stop minting the retired gemini runtime id in workflow text

#1928 removed the gemini runtime after Google sunset Gemini CLI on 2026-06-18,
but the removal stopped at the installer boundary. Runtime-loaded workflow text
kept assigning the id, and the name policy's unknown-id fallbacks then applied a
default designed for a never-known FUTURE runtime to an id GSD itself retired:
getRuntimeLabel('gemini') is 'Claude Code', getProjectInstructionFile('gemini')
is 'AGENTS.md', getGlobalConfigDir('gemini') is ~/.claude. A stale id produced a
plausible wrong answer instead of an error.

Those fallbacks are DELIBERATE and are left untouched here -- four docblocks
document them, src/runtime-name-policy.cts:220-222 calls the label default "the
always-safe default, fail-closed", and an existing test in this very suite pins
getProjectInstructionFile('gemini') === 'AGENTS.md'. This commit removes the
REACHABILITY of the retired id instead:

- new-project.md, ingest-docs.md: the runtime-detection cascade mapped
  /.gemini/ and $GEMINI_CONFIG_DIR to RUNTIME=gemini. Both now map
  /.gemini/antigravity{,-ide,-cli}/ and $ANTIGRAVITY_CONFIG_DIR to
  RUNTIME=antigravity, the documented successor. ingest-docs.md was not in the
  original report; the new structural test found it.
- settings-advanced.md: dropped the `gemini` row from the runtime->model-tier
  table. The model catalog has no gemini runtime (runtimeTierDefaults has 18
  keys, none of them gemini), so the row advertised built-in defaults for a
  runtime whose config key is ignored. Its three model IDs were copied from the
  `google` PROVIDER preset -- a provider axis rendered as a runtime axis.
- settings-advanced.md: removed the `gemini` / "Gemini CLI." runtime menu
  option and its group listing, so no menu offers a runtime GSD cannot install.
- settings-advanced.md: repointed the config examples from `runtime gemini` to
  `runtime antigravity`, which ships no built-in tier defaults and is therefore
  the case those overrides actually exist for.
- reapply-patches.md: $GEMINI_CONFIG_DIR -> $ANTIGRAVITY_CONFIG_DIR,
  ~/.gemini/gsd-local-patches -> ~/.gemini/antigravity/gsd-local-patches, and
  the local scan's bare .gemini -> .agents (Antigravity's localConfigDir). This
  file is hand-written, so `npm run sync:launcher` never reached it.
- update.md: bare ~/.gemini and ./.gemini as GSD config dirs -> the real
  ~/.gemini/antigravity and ./.agents.

Antigravity's own Gemini-family surfaces are untouched by design: ~/.gemini as
its configHome parent, ~/.gemini/config for global skills/agents (#3738),
hookEvents "gemini", GEMINI.md as its projectInstructionFile, the
~/.gemini/antigravity{,-ide,-cli} ambiguity probes (#1441), and every gemini-*
model ID. The launcher's own GEMINI_CONFIG_DIR arm is left to #4632, which
absorbed #4347 for it.

Refs #4709

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* docs(#4709): record the Gemini -> Antigravity migration research

Primary-source research note behind #4709: the sense taxonomy that separates a
runtime-axis `gemini` (stale) from Antigravity's on-disk contract, Google's
model IDs, and release history (all load-bearing); the PRESERVE table; the
guard-gap analysis; and the per-file inventory with file:line citations.

Refs #4709

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(#4709): keep the legacy patches probe, and stop tripping two lint gates

Three review findings, fixed inline.

1. Dropping the global ~/.gemini/gsd-local-patches probe was a regression: a
   pre-#1928 Gemini CLI install put patches there, and a stranded patches dir is
   still the user's work. Restored as an explicitly-labelled legacy arm probed
   AFTER Antigravity, so a live install always wins. This is a directory probe,
   not a runtime home -- it assigns no runtime id, so it does not reintroduce
   the defect this PR closes. The $GEMINI_CONFIG_DIR env probe is deliberately
   NOT restored: that names a runtime config home, which
   tests/declarative-reference-antigravity.test.cjs:307 pins as ignored.

2. The comment added in (1) originally contained the literal string that the new
   structural test matches, so the test flagged its own fix's comment as a mint.
   Reworded. The test was right; a comment in shipped workflow text is as
   readable to a matcher as code is.

3. docs/research/gemini-to-antigravity-migration.md used the colon slash-form
   inside a quoted manifest description. lint-docs-command-form rejects it:
   docs are never passed through the install-time converters, so the colon form
   names a command no runtime registers. Normalised to the hyphen form.

Also verified, rather than assumed: the local scan's .agents entry is
unambiguous. Antigravity is the ONLY runtime declaring localConfigDir '.agents'
across all 19 capability manifests; grok and codex use ~/.agents as a GLOBAL
home, and this scan is local (./$dir).

Refs #4709

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(#4709): retire Gemini CLI from the PR templates, and close two review gaps

Adversarial review findings, all fixed inline.

1. All three .github/PULL_REQUEST_TEMPLATE/*.md still offered "Gemini CLI" under
   "Runtimes tested", and none offered Antigravity. #1928's follow-up dropped
   Gemini CLI from .github/ISSUE_TEMPLATE/*.yml but missed the PR templates, so
   every contributor opening a fix/feature/enhancement PR has been asked for two
   releases which runtime they tested and offered a retired one. Now Antigravity.
   Guarded by a new assertion: runtime checklist labels in the PR templates must
   appear in the runtime label table. Proven non-vacuous by reverting one
   template line and watching the probe report the offender.

2. The #4709 scanning corpus excluded agents/, which also ships runtime-loaded
   markdown including .compact.md variants. Widened: 318 -> 382 files (+64), zero
   new offenders, so the gap was coverage rather than a live defect.

3. gsd-core/workflows/sync-skills.md said "grok and gemini have no dedicated
   installer flag — they alias the codex and claude skills roots respectively."
   The gemini half is wrong twice over: the runtime is retired, and it never
   aliased claude -- canonicalizeRuntimeName returns null for it and the caller's
   fail-closed default merely happens to be claude. Describing that as designed
   aliasing is exactly the confusion this issue is about. Reduced to grok, which
   genuinely does alias the codex skills root.

4. The changeset said the runtime was removed in 1.11. It shipped in 1.8.0
   (CHANGELOG.md:1023 is the enclosing release heading for the #1928 entry at
   :1124). Corrected.

Refs #4709

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* test(#4709): follow the corrected sync-skills prose, and refresh the compact baseline

Three GREEN-run failures, all caused by this PR's own edits.

1. tests/sync-skills-cross-runtime-refuse.test.cjs pinned the literal phrase
   "grok and gemini have no dedicated installer flag" — a test REQUIRING shipped
   text to name a runtime retired in 1.8.0, which is the exact class #4709
   exists to remove. The assertion and its rationale comment now track the
   corrected prose ("grok has no dedicated installer flag"), and the docblock's
   runtime list drops gemini. The remaining assertions in that file — the
   guard's exit, the installer pointer, the $DEST reference, guard-before-copy
   ordering — are untouched, so #3025's contract is otherwise intact.

2. tests/fixtures/compact-content-benchmark-baseline.json drifted because the
   new-project.md edits changed its compacted size (split "new-project": off
   14279 -> 14308, on 12335 -> 12364; aggregate off 107411 -> 107440).
   Refreshed with `node scripts/benchmark-compact-content.cjs --write`, which
   is that script's own documented remedy.

3. emitted-attribution reported four grown workflow files with no
   acknowledgment. Acked below as commit trailers per ADR-3942, which moved the
   acknowledgment out of tests/emitted-drift-acks/*.json fragments and into the
   PR's own commit range (read with three-dot base...head). Exactly the four
   files the gate named are acked — settings-advanced.md and sync-skills.md
   shrank and are deliberately absent, since a trailer no delta consumed is a
   staleAcks error.

Refs #4709

Emitted-Drift-Ack-Growth: ingest-docs.md — the runtime-detection cascade now names Antigravity's three real directories (/.gemini/antigravity{,-ide,-cli}/) and $ANTIGRAVITY_CONFIG_DIR in place of the single retired /.gemini/ arm and $GEMINI_CONFIG_DIR; three correct paths cost more bytes than the one wrong path they replace.
Emitted-Drift-Ack-Growth: new-project.md — same runtime-detection correction as ingest-docs.md, plus dropping "gemini/" from the two GEMINI.md instruction-file sentences so the prose stops contradicting getProjectInstructionFile, which returns AGENTS.md for that retired id.
Emitted-Drift-Ack-Growth: reapply-patches.md — restores the legacy ~/.gemini/gsd-local-patches probe as an explicitly-labelled arm after an adversarial-review finding that dropping it stranded a pre-#1928 user's patches, and repoints the env/global probes at Antigravity; the four-line comment is load-bearing, since a bare retired-runtime path with no explanation is exactly what the next reader would delete.
Emitted-Drift-Ack-Growth: update.md — bare ~/.gemini and ./.gemini as GSD config dirs are replaced by the real ~/.gemini/antigravity and ./.agents, which are longer strings; no content was added beyond the corrected paths.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* chore(#4709): backfill the changeset PR number

pr: 0 -> 4711, now that the PR exists. Never guessed ahead of the number.

Refs #4709

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

---------

Co-authored-by: sim <sim@local>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
 2026-09-13 23:08:20 -04:00
Tom Boucher
557a0b2876 test(#4516): migrate per-runtime install/upgrade adapters to named timeout constants (#4582) 
Batch 5 of the ad hoc timeout literal migration (epic #4445). Replaces
every bare numeric timeout/timeoutMs object-literal property across
tests/augment-upgrades.test.cjs, tests/shared-hooks-dir-resolution.test.cjs,
tests/antigravity-upgrades.test.cjs, tests/cursor-hooks.test.cjs,
tests/cursor-hook-workspace-roots.test.cjs, tests/gemini-runtime-removed.test.cjs,
tests/kilo-upgrades.test.cjs, tests/kimi-upgrades.test.cjs,
tests/kimi-variant-disambiguation.test.cjs, tests/opencode-plugin-adapter.test.cjs,
tests/windsurf-hooks-bridge.test.cjs, tests/effort-sync-installed-runtime.test.cjs,
and tests/hooks-commonjs-marker.test.cjs with a named constant, per
eslint-rules/no-adhoc-timeout-literal.cjs. Removes these 13 files from the
rule's allowlist.

Adds one new shared constant to tests/helpers/timeouts.cjs for spawning a
single already-staged hook script directly (a heavier class than the
existing quick-spawn norm), shared across three files in this batch. One
new file-local constant covers a property-test driver process distinct
from any existing class. Every other site reuses an existing shared norm.
No src/bin file touched, no numeric timeout value changed anywhere.

Co-authored-by: sim <sim@local>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
 2026-09-10 14:11:26 -04:00
Tom Boucher
3fac6e629f test(#3145): bound the installer/runtime cluster onto the process seam (#3176) 
* test(#3145): bound the installer/runtime cluster onto the process seam

Migrates 156 unbounded sync spawn sites across 47 files. Allowlist 120 to 73.

Timeouts are sized from evidence already in the tree rather than a house
default, because this wave spawns installers rather than git plumbing and an
undersized bound does not catch a hang -- it manufactures CI flake, which is
worse, since a flake gets re-run instead of investigated. install.test.cjs
records a real spawnSync ETIMEDOUT at a 60000ms cap on a loaded bench while
another lane passed the same commit in 12.7s, so full installs are bound at
120000ms against that recorded incident.

Also adds an auditable escape to the guard's timeout ceiling. The 600000ms
cap was set in #3143 from partial evidence, but fragment-single-edit-
propagation carries a documented, load-tested 900000ms bound on a run that
chains a full build plus eight generators -- the guard would have rejected a
correct timeout the moment that file left the allowlist. A value above the
ceiling is now permitted only with an inline allow-spawn-timeout-ceiling
marker carrying a non-empty reason. It raises the ceiling; it never waives
the requirement for a bound, which is asserted directly.

install-shared.cjs keeps its hand-rolled assert rather than routing through
throwIfFailed: its message embeds both streams, and throwIfFailed carries
only a trimmed stderr. The message now also names the outcome, so a bounded
timeout reads as such across its 38 importers instead of as
expected null to equal 0.

* test(#3145): extract class-norm timeouts and correct the build-hooks sizing

A pre-PR review found 52 copies of four class-norm timeout constants across
this wave. These are not per-suite fixture bindings -- they are shared facts
about how long a class of subprocess takes, derived from a recorded bench
incident. That norm already moved once (60000 to 120000 after a real
ETIMEDOUT), and 52 copies would have drifted the next time it moved.

Extracts tests/helpers/timeouts.cjs, where each norm is justified once, and
converts the copies. A site that genuinely differs -- a real tsc compile, or
regen:derived -- keeps its own local constant with its own justification.

Also corrects a misclassification: scripts/build-hooks.js was sized as a
build at 120000 in twelve places and 60000 in another, but it compiles and
bundles nothing. Its own header says no bundling needed; it copies pre-built
files and syntax-checks them with vm. Three different values bounded one
script; now there is one.

* test(#3145): fix red CI — lint self-match and a Windows chunk overrun

Two failures on PR 3176.

lint-allow-test-rule-refs read a RuleTester fixture as a real exemption. The
fixture exists to prove an unrelated marker does NOT suppress the rule, so it
carries that marker's literal text as test data. Split via concatenation, the
same idiom no-unbounded-spawn-allowlist.test.cjs already uses for its own
self-match problem. The explanatory comment needed the same treatment.

The Windows shard 3/3 chunk was killed at its 600000ms budget. Output stopped
seven minutes before the kill, so this was an overrun rather than a slow
chunk: regenDerivedPropagatesSingleFragmentEditWithNoSecondSourceSurface runs
regen:derived bounded at 900000ms, which is larger than the whole chunk
budget, so the chunk killer always fires first and it can never complete
there. Both the test and that bound predate this change; modifying the file
pulled it into the Windows targeted set and exposed it. Skipped on Windows
with the reason recorded; the Linux lanes cover it. The 900000 bound and its
ceiling marker are unchanged -- they are correct.

* test(#3145): refresh the stale test-timings cost table

The Windows shard was killed at its 600000ms per-chunk budget. run-tests.cjs
packs chunks by measured duration from tests/test-timings.json, and an
unknown file falls back to the table's median weight -- advisory by design,
but it silently underweights exactly the files that matter.

Four of the failing chunk's 22 files were absent from the table, including
the two heaviest: fragment-single-edit-propagation.install.test.cjs at 230s
(it runs regen:derived) and agent-fragments-emission.install.test.cjs at 79s.
Both were weighted as average, so the chunk's total weight read 53.68 against
a budget of 60 and the packer produced a single chunk.

Regenerated from a passing full-suite run, per the remedy the script itself
documents. 700 to 770 entries, 70 added, 0 dropped -- verified, since
gen-test-timings.cjs replaces the table wholesale rather than merging.

Proven against the real packer: the same 22 files now weigh 103.91 and split
into two chunks. No logic, budget, or timeout was changed; raising a budget
to make a red gate pass is not a fix.

---------

Co-authored-by: sim <sim@local>
 2026-08-07 15:18:18 -04:00
Tom Boucher
a0fafedfa0 feat(#2103): drive VS Code through the Embeddable Orchestration System (ADR-1239) 
VS Code is a net-new EoS runtime that — unlike every prior migration — is NOT
CLI-installed (Marketplace/VSIX extension). It has zero runtime==='vscode'
branches in bin/install.js and stays that way (regression-guarded); it is driven
entirely through the negotiated imperative Host-Integration adapter.

Registry + validator (the hard part):
- capabilities/vscode/capability.json (role:runtime): full hostIntegration block
  (imperative / palette / active vscode.lm model / engine hook bus /
  sandboxed-storage / mcp transport / sandboxed-web runtime; dispatch nested,
  maxDepth 5 per VS Code's documented subagent depth).
- capability-validator.cjs extended so a role:runtime capability can legitimately
  declare "extension-distributed, no config directory": new configHome.kind:'none'
  + installSurface:'none' (+ GATE-A pairing + the parity maps), with localConfigDir
  and configHome.name made conditional on kind!=='none'. All 18 runtimes still
  validate; getDirName returns a distinct sentinel (not '.claude') for a no-config
  runtime.
- The add-a-registry-runtime tax: NON_INSTALLABLE_RUNTIMES exemption in the
  runtime-flags drift guard, vscode added to global-config-home SPECIAL_CASED,
  EXPECTED_PROFILES.vscode='ide', and the config-adapter/derivation/pin-count
  guards updated. No golden-install fixture, model-catalog, or CONFIGURATION rows
  (vscode never enters allRuntimes).

Dispatch + extension surface:
- Fixed vscode/extension.js's createHub()-no-args bug (every dispatch was
  UnknownCommand, masked by a vacuous reachability test) — now reuses the shared
  dispatchGsdCommand subprocess-shim (Node/desktop); the reachability test is
  tightened to assert real dispatch.
- Promoted the #1933 host binding to a shipped vscode/host-binding.js; activate()
  now composes the model/hookBus/stateIO seams through it. Corrected the model
  seam to VS Code's real API (vscode.lm.selectChatModels() -> model.sendRequest();
  vscode.lm.sendRequest does not exist) so the binding actually composes on real
  desktop VS Code instead of throwing.
- New vscode/browser.js Web Extension entry with ZERO Node APIs (the engine's
  config/capability loading is Node-bound, so the web entry registers the surface
  and directs full dispatch to the native MCP server — honestly documented).
- UPGRADE 1: GSD skills as native Language Model Tools (contributes.languageModelTools
  + vscode.lm.registerTool), invoke() dispatching through the hub.
- UPGRADE 2: native subagent dispatch wired onto #runSubagent /
  chat.subagents.allowInvocationsFromSubagents (fail-soft on API availability,
  maxDepth 5 enforced).
- vscode/package.json: browser entry, engines.vscode ^1.105, chatParticipants +
  languageModelTools contributions; fixed a stale activationPoints->activationEvents
  manifest key. Added "vscode" to the package files array.

Docs (## vscode matrix section) + changeset (Added).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-07-11 23:24:42 -04:00
Tom Boucher
7acbdc71b8 test(#1925): register zcode across installer surfaces + fluidify count pins 
Resolve the zcode test cascade exposed by gsd-test:
- model-catalog.json: add zcode runtimeTierDefaults (null tiers, matching the
  other profile-marker-only runtimes) so KNOWN_RUNTIMES stays parity with allRuntimes.
- runtime-name-policy: add zcode to GLOBAL_CONFIG_HOME_FRAGMENTS (~/.zcode) so
  getGlobalConfigHomeFragment stops falling through to the .claude default.
- installer-migration-install: add the zcode fresh-install contract (flat-skills,
  no settings, no package.json — same shape as trae).
- golden-install-parity: capture the zcode fixture via a standalone generator
  script (not node --test — the gate stays gsd-test).
- capability-matrix.md: regenerate so zcode appears as a first-party row.

Fluidify the remaining count-pinned guards so adding a runtime no longer trips a
hand-pinned snapshot: gemini-runtime-removed (flag count derives from the
registry), non-claude-runtimes-registry-derivation (golden list derived from the
registry).
 2026-07-06 08:31:52 -04:00
Tom Boucher
8f2ebbe9bf feat(#1928): remove sunset Gemini CLI runtime, redirect to Antigravity (#1996) 
* feat(#1928): remove sunset gemini cli runtime, redirect to antigravity

Google sunset Gemini CLI on 2026-06-18; Antigravity CLI is its official successor (already a first-class GSD runtime). Remove the gemini runtime from the enum (16->15), aliases, labels, config-home fragment, install path, converters (convertClaudeToGemini{Markdown,Toml,Agent}, convertSlashCommandsToGeminiMentions), capability descriptor, gemini-extension.json, RULESET.GEMINI.*, and the interactive menu (renumbered, no gap).

--gemini now prints an explicit deprecation notice citing the 2026-06-18 sunset and redirects to --antigravity (no silent alias, per the issue's Hyrum's-Law rejection). Antigravity is preserved throughout: its GEMINI.md contextFileName, .gemini/antigravity config home, the shared convertGeminiToolName/claudeToGeminiTools tool vocabulary, and the 'gemini' hookEvents dialect it declares. GEMINI.md retargeted as Antigravity's context file.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(#1928): backfill changeset PR number (#1996)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(#1928): drop Gemini CLI from issue templates (review nit)

Removes the sunset Gemini CLI runtime from the two GitHub issue-template
runtime lists that the removal PR missed, per @davesienkowski's review nit:
- feature_request.yml: 'Applicable runtimes' checkbox (a user could otherwise
  request a feature for a runtime GSD no longer supports)
- bug_report.yml: 'Runtime' dropdown + the stale ~/.gemini/settings.json
  retrieval-help line

Leaves the post-removal templates fully consistent with the Antigravity redirect.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-07-04 13:32:51 -04:00