abhay-codes07andClaude Opus 4.8 4fff67d219 fix(skill): stamp _origin on the semantic tier so update cannot delete it (#2843) ...

build._is_ast_tier reads _origin when present and otherwise guesses the tier
from the shape of source_location (L<line> means AST). extract() stamps its AST
items _origin:"ast", but nothing stamped the semantic side, so a semantic node
whose location happened to look like "L12" read as AST — and the next
`graphify update` deleted it along with the re-extracted code file it seemed to
belong to. Part C, extract_corpus_parallel, and the CLI merge now setdefault
_origin:"semantic" on every semantic node/edge before the merge, so a semantic
item is never mistaken for an AST one. AST items (stamped "ast") are untouched;
setdefault preserves any existing stamp.

Landed from PR #3110. Skill artifacts regenerated fresh from the updated
fragments (not the PR's stale committed blobs); all five skillgen validators
pass.

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

2026-10-12 12:18:15 +01:00

..

skillgen

fix(skill): stamp _origin on the semantic tier so update cannot delete it (#2843)

2026-10-12 12:18:15 +01:00

__init__.py

feat(skills): progressive-disclosure split for all platforms (generator + drift fence) (#1121)

2026-06-02 20:48:13 +01:00