+ lHj aRt 0tRt^RIt^RIt^RIt/t]^k]P !4t^tRt RRlt RRRllt R#) urskill_view repeat-view dedup registry: per-task cache of (resolved skill name, file_path) -> digest of the skill_view result served. A repeat view whose result is byte-identical returns a short stub — the earlier tool result already carries the content verbatim. Content digests need no file timestamps, so a same-size rewrite in one coarse mtime tick (or on a host whose clock disagrees with the filesystem's) is never mistaken for "unchanged". Cleared via ``reset_skill_view_dedup()`` on context compression AND on a committed proactive tool-result prune, because both replace the original content with a one-line marker. NuSkill content unchanged since it was loaded earlier in this conversation — refer to the earlier skill_view result; it is still current and complete. (Re-issued after context compression, this returns the full content again.)cJV^8dQhR\R\R\R,/#)payloadresultreturnN)dictstr)formats"22{{-..@7;;v&..$/H __" %C ^^FMM'2 3 = = ?F ! !#..s7|R@ 99S>V #::4; //Z$@RTY46EJK " !  #tc %j00d4;'(1 "  " ! s*AE*E*AE** E; c6V^8dQhR\R,RR/#)rr*Nr)r)r s"r r r 2s88C$J8$8r c\;_uu_4Vf\P4M\P\ V4R4RRR4R# +'giR#;i)zVClear the dedup cache (all tasks when task_id is None); called on context compression.N)rr clearr%r)r*s&r reset_skill_view_dedupr42s< ! ! ?  % % '  # #CL$ 7 " ! ! !s 9A A# cV^8dQh/^\9d,\\\\\3,3,;R&#)rr )__conditional_annotations__rrtuple)r s"r r r s165T#tE3J//05r )N) r6__doc__rr" threadingr Lockrr'r$r0r4r )r6s@r r;sK 355$>>+'288r