You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
chore(catalog): re-pin llmslim to v0.12.0 - #145645
Re-pins the llmslim context engine to its v0.12.0 release.
Change
plugin-catalog/llmslim.yaml:
pin 12a88324 → 34c6f80a, the v0.12.0 release commit;
version 0.11.2 → 0.12.0;
the disclosure now says it installs llmslim==0.12.0 from PyPI.
Capabilities are unchanged: no tools, hooks, middleware or env vars.
Upstream range
plugins/hermes between the two pins: only the version and the llmslim==0.12.0 dependency pin in plugin.yaml, and the same pin in its README.
The PyPI package moves from 0.11.2 to 0.12.0 (changelog).
The context engine's compression path (llmslim/focus.py): per-text term lists are now cached between calls, which speeds up repeated ranking with the same output. rank also takes an optional candidates argument whose default is unchanged.
Everything else: the rest of 0.12.0 changes llmslim.memory recall, which the engine does not use.
Ran the structural gate of .github/workflows/plugin-catalog-ci.yml against this head (356cfc6, fetched to a named ref, SHA matches the PR head): python scripts/validate_plugin_catalog.py plugin-catalog/ → OK: 861 file(s) valid, no error lines. That job only fires on upstream branches, so it never runs for a fork head — this is the local equivalent. The second gate (pinned-source-validate, which clones the pinned sha and runs hermes plugins validate) was not run.
@Thanatos9404 thanks for the llmslim bump — approved at 34c6f80a (plugin delta is the version pin; we read the llmslim 0.11.2 → 0.12.0 library delta, which is pure text ranking/selection code).
It lands automatically after 2026-10-30 17:08 UTC, when the exactly pinned llmslim==0.12.0 it installs is 3 days old (our dependency-age rule). No action needed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
area/compressionContext compression and continuation sessionsci-reviewedapplied to manually approve dangerous changescomp/pluginsPlugin system and bundled pluginsP3Low — cosmetic, nice to have
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
alt-glitch
added
area/compression
Re-pins the llmslim context engine to its v0.12.0 release.
Change
plugin-catalog/llmslim.yaml:12a88324→34c6f80a, the v0.12.0 release commit;llmslim==0.12.0from PyPI.Capabilities are unchanged: no tools, hooks, middleware or env vars.
Upstream range
plugins/hermesbetween the two pins: only the version and thellmslim==0.12.0dependency pin inplugin.yaml, and the same pin in its README.llmslim/focus.py): per-text term lists are now cached between calls, which speeds up repeated ranking with the same output.rankalso takes an optionalcandidatesargument whose default is unchanged.llmslim.memoryrecall, which the engine does not use.I maintain LLMSlim.