Skip to content

v2.22.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 07 Sep 06:26
· 51 commits to main since this release
v2.22.0
3f0f457

v2.22.0 (2026-09-07)

Bug Fixes

  • Add paramiko skill (b7dfd82)

  • Add planning-stream JSONDecodeError recovery and remove dead RetryConfig (c9831a7)

  • Agent graph wrong edges (426a644)

  • D-orbit not smooth (ebbaf39)

  • Doubling of agent cards and markdown blocks (c8d387a)

  • Edge animation not on the lines (2b80314)

  • Graph not showing flash mode tasks (4c0a484)

  • Images centering (9f3d78e)

  • Not showing chat bubble after refreshing/session switching (30fcdc7)

  • Settings issues (78a3279)

  • Skill not return assets'info (57396d0)

  • Waiting after finished & doubling markdown renders (da35b4f)

  • skills: Address review - add --use-pretrain-script and --output to DPA-4c train; default epochs back to 50 (da9783a)

  • skills: Lower DPA-4c fine-tuning start_lr to 1e-4 per official usage note (3e06e2e)

Chores

  • Merge duplicate eos and equation-of-states skills (36ee500)

Documentation

  • eos: Spell out the third-order Birch-Murnaghan equation (3598f37)

  • skills: Atom_modify map yes only needs to precede the pair_style deepmd line (3cc32f9)

  • skills: Require atom_modify map yes for DPA-4c (and DPA-4) in LAMMPS inputs (8e87118)

Features

  • Add lenient JSON parsing for tool-call arguments and tests (9e2978f)

  • Flash agent graph (7eabd24)

  • Frontend refactor (8695db8)

  • skills: Merge DPA-4c preparation into deepmd_prepare.py with official OMat24 variants (ba1dfa6)

Refactoring

  • Fatier d orbit in waiting animation (b2e57c4)

  • Skill creation skill with kdg info (c93eb48)

  • Waiting animation (73ea271)

Detailed Changes: v2.21.0...v2.22.0

Assets 2
Loading