|
1 | 1 | # Changelog |
2 | 2 |
|
3 | | -All notable changes to this project are documented here. |
| 3 | +All notable changes are documented here. |
4 | 4 |
|
5 | | -## [0.1.0] - 2026-07-16 |
| 5 | +## [0.2.0] - 2026-07-17 |
6 | 6 |
|
7 | 7 | ### Added |
8 | 8 |
|
9 | | -- medium-neutral `design-workflow` Agent Skill |
10 | | -- explicit preservation versus redesign boundary |
11 | | -- routing for preserve, expand, create, redesign, critique, brand-check, |
12 | | - translate, and profile tasks |
13 | | -- progressive reference files and design-profile templates |
14 | | -- realistic routing and trigger evaluation cases |
15 | | -- dependency-free validation scripts |
16 | | -- Apache-2.0 license, source audit, and third-party notices |
17 | | -- GitHub Actions CI |
18 | | -- public release checklist and contributor/security documentation |
19 | | - |
20 | | -### Removed from the distributable package |
21 | | - |
22 | | -- project-specific names, palettes, layouts, products, and organizational context |
23 | | -- internal routing identifiers not required by the Agent Skills standard |
24 | | -- unverified Canva attribution |
| 9 | +- installable `design-workflow/` package whose directory matches the skill name |
| 10 | +- generated `agents/openai.yaml` interface metadata |
| 11 | +- mixed-route, partial-evidence, and instruction-security guidance |
| 12 | +- dependency-free route-contract validator and example contract |
| 13 | +- balanced trigger cases and a client-gradeable semantic smoke suite |
| 14 | +- recorded independent 16-case forward test covering all eight routes |
| 15 | +- project validator, unit tests, deterministic skill/project ZIP packaging, and SHA-256 output |
| 16 | +- CI with official `agentskills` validation and tagged GitHub Release automation |
| 17 | + |
| 18 | +### Changed |
| 19 | + |
| 20 | +- clarified `expand` versus `translate` and dimension-specific preservation |
| 21 | +- narrowed activation to tasks that require visual design judgment |
| 22 | +- separated human repository documentation and evals from the runtime skill |
| 23 | +- replaced committed self-reported dry-run results with a reproducible grading |
| 24 | + protocol |
| 25 | + |
| 26 | +### Removed |
| 27 | + |
| 28 | +- root-level `SKILL.md` layout that produced an invalid default installation name |
| 29 | +- unrelated runtime traceback from the old routing report |
| 30 | +- CI claims that fixture consistency was semantic model validation |
| 31 | + |
| 32 | +## [0.1.0] - 2026-07-16 |
| 33 | + |
| 34 | +- initial medium-neutral routing skill with eight routes |
| 35 | +- preservation-versus-redesign boundary, references, templates, and provenance |
0 commit comments