mirror of
https://github.com/Tencent/BrowserSkill.git
synced 2026-09-10 13:01:23 +08:00
feat(dsh-plugin-browserskill): publish the BSK skill through ctx.skills
The plugin exposed only tools; the BrowserSkill agent skill never reached the harness's <available_skills> catalog. Register it as an embedded runtime skill (progressive disclosure: catalog entry resident, body on skill-tool invocation) via the official ctx.skills seam — silent no-op when the composition lacks it. Content is build-time assembled so there is one source of truth: skill/prelude.md (committed, dsh-specific: tool<->CLI map, owned-session semantics, plugin-only overrides) + the canonical repo-root skill/SKILL.md verbatim (the same file crates/bsk-cli/build.rs mirrors; frontmatter stripped, name/description supplied by the registration). Generated into src/skill-content.generated.ts (gitignored) — registration and every pre-step snapshot are pure in-memory reads: no disk, no process, no daemon. Scripts: build:skill chained into build/test/typecheck (same pattern as build:css); biome excludes the generated artifacts. Tests: +3 (catalog content, in-memory weight, silent degradation). 116/116 green.
This commit is contained in:
@@ -23,6 +23,19 @@ extension keep owning the actual browser control — this package is a thin, wel
|
||||
| `browser_screenshot` | `bsk screenshot` | PNG capture of the tab or a ref-cropped element; inlines the image when the deployment supports image input, otherwise returns a file path. |
|
||||
| `browser_emulate` | `bsk emulate` | Apply or clear mobile device emulation on the active tab. |
|
||||
|
||||
## Agent skill (progressive disclosure)
|
||||
|
||||
Beyond the tools, the plugin publishes the **`browser-skill` agent skill** through the harness's
|
||||
official skill seam (`ctx.skills.register`): the catalog entry (name + routing description) is
|
||||
resident in `<available_skills>`, and the body is loaded only when the model invokes the `skill`
|
||||
tool. The body is assembled at build time from two parts, so there is exactly one source of
|
||||
truth: a dsh-specific prelude (`skill/prelude.md` — tool↔CLI map, owned-session semantics,
|
||||
plugin-only overrides) followed verbatim by the canonical CLI skill (`skill/SKILL.md` at the
|
||||
repo root — the same file `crates/bsk-cli/build.rs` mirrors for the CLI package; workflows,
|
||||
stop-when-done rules, refs usage, sandbox rules). Registration and every pre-step catalog
|
||||
snapshot are pure in-memory reads (no disk/process/daemon); compositions without the skill
|
||||
seam degrade silently.
|
||||
|
||||
## Multi-session model
|
||||
|
||||
One agent conversation can drive several browser sessions at once:
|
||||
|
||||
Reference in New Issue
Block a user