Coder/deepseek-harness
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: split skill providers

Browse Source
This commit is contained in:
Yichen Jiang
2026-07-08 15:50:38 +08:00
parent 5fd647012e
commit f453ba77a2
77 changed files with 1673 additions and 1334 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
packages/core/tool-skill/README.md
View File

@@ -10,6 +10,6 @@ Requires `ctx.tools` and `ctx.skills` (`inject: ['tools', 'skills']`).
|---|---|---|
| `name` | string (required) | Exact kebab-case skill name from the available skills listing. |
Execution uses the calling agent's `session.header.cwd` to resolve project-local skills. A successful call returns a text block containing `<skill_content name="...">`, the skill body, the skill base directory, and relative-path guidance. Unknown names, invalid names, and skills marked `disableModelInvocation: true` return `isError` tool results through the normal tool registry error path.
Execution uses the calling agent's `session.header.cwd` so workspace-sensitive providers can resolve the right winning skill. A successful call returns a text block containing `<skill_content name="...">`, the skill body, and provider resource guidance. Local filesystem skills include a base directory for resolving relative files; remote or embedded providers can return URL or opaque provider-managed guidance instead. Unknown names, invalid names, and skills marked `disableModelInvocation: true` return `isError` tool results through the normal tool registry error path.
The tool does not call `agent.inject()` in v1. Its result is already recorded as the tool result and becomes available to the next model step without duplicating the content as synthetic context.