mirror of
https://github.com/agent0ai/agent-zero.git
synced 2026-05-19 07:59:34 +00:00
refactor: align skills and tool guidance
Rename high-impact skills to task-oriented names and move plugin-owned skills into their owning plugin folders.\n\nAlign renamed skill frontmatter with the official SKILL.md standard by keeping trigger language in name/description metadata, replacing the old create-skill wizard with build-skill, and updating browser, A0 connector, computer-use, CLI setup, and scheduler skill references.\n\nTighten the recurring cross-provider guidance gaps surfaced by the evidence sweeps: memory requests now avoid promptinclude-file routing, scheduler prompts distinguish cron schedules from planned ISO dates, document questions prefer document_query, skills_tool search/read_file usage is clearer, normal notifications set info/priority 10, and local/host text editors preserve patch intent.\n\nUpdate regression tests for the renamed skills, plugin ownership, prompt budget reality, and standard frontmatter shape.
This commit is contained in:
Alessandro
parent
79fe46cd7b
commit
6d29268cbd
29 changed files with 243 additions and 424 deletions
|
|
@ -229,7 +229,7 @@ from your own browser:
|
|||
If another agent is helping with setup, do not paste a whole checklist. Paste one line:
|
||||
|
||||
```text
|
||||
Set up the A0 CLI connector for Agent Zero on this machine using the a0-setup-cli Skill.
|
||||
Set up the A0 CLI connector for Agent Zero on this machine using the setup-a0-cli Skill.
|
||||
```
|
||||
|
||||
## Troubleshooting
|
||||
|
|
|
|||
|
|
@ -4,7 +4,7 @@ Runs shell-backed execution on the machine where a connected A0 CLI is running.
|
|||
Use this tool, not `code_execution_tool`, when the user asks for the connected
|
||||
local terminal, the A0 CLI host, their local machine, or explicitly says not to
|
||||
use Docker/server/container execution.
|
||||
For complex local project work, optionally load skill `code-execution-remote`.
|
||||
For complex local project work, optionally load skill `host-code-execution`.
|
||||
|
||||
Availability and permissions are checked when the tool runs. If no CLI is
|
||||
connected, remote execution is disabled, or local access is not Read&Write for a
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
Reads, writes, and patches files on the machine where a connected A0 CLI is
|
||||
running. Use this tool, not server-side file tools, when the user asks for files
|
||||
on the connected local machine, A0 CLI host, or explicitly says not to use
|
||||
Docker/server files. For complex remote edits, optionally load skill `text-editor-remote`.
|
||||
Docker/server files. For complex remote edits, optionally load skill `host-file-editing`.
|
||||
|
||||
Availability and permissions are checked when the tool runs. If no CLI is
|
||||
connected, remote file access is disabled, or a write/patch needs Read&Write,
|
||||
|
|
@ -18,6 +18,7 @@ report that to the user instead of falling back to server-side file tools.
|
|||
|
||||
## Notes
|
||||
- Prefer `read` before line-number edits.
|
||||
- If the user says patch, change without rewriting, or don't rewrite, use `action: "patch"` instead of `write`.
|
||||
- Prefer `patch_text` for context-anchored changes and `edits` only for fresh, surgical line ranges.
|
||||
- If freshness checks reject a line patch, reread the file and retry with updated ranges.
|
||||
- Relative paths are relative to the CLI host filesystem. Do not rewrite them to
|
||||
|
|
|
|||
|
|
@ -1,28 +1,9 @@
|
|||
---
|
||||
name: code-execution-remote
|
||||
description: Guide safe use of code_execution_remote for shell-backed execution on the connected A0 CLI host machine.
|
||||
version: 1.0.0
|
||||
author: Agent Zero Team
|
||||
tags: ["agent-zero", "a0", "cli", "connector", "remote-execution", "shell"]
|
||||
trigger_patterns:
|
||||
- "code_execution_remote"
|
||||
- "remote shell execution"
|
||||
- "run commands on the cli host"
|
||||
- "run python on the cli host"
|
||||
- "run node on the cli host"
|
||||
- "connected local terminal"
|
||||
- "connected terminal"
|
||||
- "local terminal"
|
||||
- "my terminal"
|
||||
- "a0 cli"
|
||||
- "cli host"
|
||||
- "not docker"
|
||||
- "not the docker terminal"
|
||||
allowed_tools:
|
||||
- code_execution_remote
|
||||
name: host-code-execution
|
||||
description: Guide safe use of code_execution_remote for shell-backed execution on the connected A0 CLI host. Use when the user asks for their computer, local terminal, CLI host, remote shell, not Docker, or host-side Python/Node/terminal commands.
|
||||
---
|
||||
|
||||
# Code Execution Remote
|
||||
# Host Code Execution
|
||||
|
||||
## Boundary
|
||||
|
||||
|
|
@ -1,22 +1,9 @@
|
|||
---
|
||||
name: computer-use-remote
|
||||
description: Beta local desktop control through a connected A0 CLI host; use for screenshots, menus, browser chrome, and native UI tasks.
|
||||
version: 1.1.0
|
||||
author: Agent Zero Team
|
||||
tags: ["computer-use", "desktop", "local-ui", "screenshots", "native-ui"]
|
||||
trigger_patterns:
|
||||
- "computer use"
|
||||
- "computer-use"
|
||||
- "computer_use_remote"
|
||||
- "local desktop control"
|
||||
- "control local browser"
|
||||
- "click on screen"
|
||||
- "native ui"
|
||||
allowed_tools:
|
||||
- computer_use_remote
|
||||
name: host-computer-use
|
||||
description: Beta desktop control through the connected A0 CLI host. Use when the user asks for computer use, screenshots, screen inspection, menus, browser chrome, native UI tasks, clicking, scrolling, typing, or checking computer_use_remote status.
|
||||
---
|
||||
|
||||
# Computer Use Remote
|
||||
# Host Computer Use
|
||||
|
||||
This skill unlocks the beta `computer_use_remote` tool for connected local desktop control through A0 CLI.
|
||||
|
||||
|
|
@ -26,7 +13,7 @@ Load this skill before using `computer_use_remote` for local desktop and native
|
|||
|
||||
If the task is browser-only and the user is flexible, prefer direct browser tooling because it is usually more reliable and token-efficient than screenshot-driven desktop control.
|
||||
|
||||
If the task needs shell execution on the CLI host, load `code-execution-remote` separately rather than treating desktop control and shell execution as one affordance.
|
||||
If the task needs shell execution on the CLI host, load `host-code-execution` separately rather than treating desktop control and shell execution as one affordance.
|
||||
|
||||
## Tool Contract
|
||||
|
||||
|
|
@ -1,25 +1,9 @@
|
|||
---
|
||||
name: text-editor-remote
|
||||
description: Guide safe use of text_editor_remote for reading, writing, and patching files on the connected A0 CLI host machine.
|
||||
version: 1.0.0
|
||||
author: Agent Zero Team
|
||||
tags: ["agent-zero", "a0", "cli", "connector", "remote-files", "file-editing"]
|
||||
trigger_patterns:
|
||||
- "text_editor_remote"
|
||||
- "remote file editing"
|
||||
- "edit my local files through a0 cli"
|
||||
- "read files on the cli host"
|
||||
- "patch files on the cli host"
|
||||
- "connected local files"
|
||||
- "connected local machine files"
|
||||
- "local files not docker"
|
||||
- "a0 cli files"
|
||||
- "cli host files"
|
||||
allowed_tools:
|
||||
- text_editor_remote
|
||||
name: host-file-editing
|
||||
description: Guide safe use of text_editor_remote for reading, writing, and patching files on the connected A0 CLI host. Use when the user asks to inspect or edit files on their computer, local files, CLI host files, or files that are explicitly not Docker/server files.
|
||||
---
|
||||
|
||||
# Text Editor Remote
|
||||
# Host File Editing
|
||||
|
||||
## Boundary
|
||||
|
||||
|
|
@ -1,20 +1,6 @@
|
|||
---
|
||||
name: a0-setup-cli
|
||||
description: Guide the user through installing and connecting the A0 CLI on the host machine so Dockerized Agent Zero can work on real local files. Use when asked to install A0, set up the CLI connector, connect Agent Zero to local files, or troubleshoot host-vs-container setup confusion.
|
||||
version: 1.1.0
|
||||
author: Agent Zero Team
|
||||
tags: ["agent-zero", "a0", "cli", "connector", "docker", "setup", "local-files"]
|
||||
trigger_patterns:
|
||||
- "install a0"
|
||||
- "install a0 cli"
|
||||
- "set up a0 cli"
|
||||
- "set up a0 cli connector"
|
||||
- "connect a0 to agent zero"
|
||||
- "connect agent zero to local files"
|
||||
- "let agent zero work on my local files"
|
||||
- "install the terminal connector"
|
||||
- "help me set up the a0 cli connector"
|
||||
- "install a0 connector"
|
||||
name: setup-a0-cli
|
||||
description: Guide installing, connecting, or troubleshooting the A0 CLI connector on the user's host machine so Dockerized Agent Zero can work on real local files. Use for install A0, set up A0 CLI, connect local files, remote tools, host-vs-container confusion, or CLI connector setup problems.
|
||||
---
|
||||
|
||||
# A0 CLI Host Setup
|
||||
|
|
@ -84,7 +84,7 @@ def install_chrome_web_store_extension(source: str) -> dict[str, Any]:
|
|||
extension_id = parse_chrome_web_store_extension_id(source)
|
||||
target = get_extensions_root() / "chrome-web-store" / extension_id
|
||||
|
||||
with tempfile.TemporaryDirectory(prefix="a0-browser-ext-") as tmp:
|
||||
with tempfile.TemporaryDirectory(prefix="browser-extension-control-") as tmp:
|
||||
archive_path = Path(tmp) / f"{extension_id}.crx"
|
||||
_download_crx(extension_id, archive_path)
|
||||
payload_path = Path(tmp) / f"{extension_id}.zip"
|
||||
|
|
|
|||
|
|
@ -1,11 +1,11 @@
|
|||
### browser
|
||||
Rendered browser automation for pages that need interaction, JavaScript, forms, downloads, screenshots, or visual inspection.
|
||||
|
||||
Prefer `search_engine` or `document_query` for plain text research. Use the browser headlessly unless the user opens the Browser surface or asks for a visible browser.
|
||||
Prefer `search_engine` or `document_query` for plain text research. The tool must not open a Browser surface automatically. Use the tool headlessly unless the user opens the Browser surface or asks for the optional visible WebUI viewer.
|
||||
|
||||
The browser may run in Docker container mode or A0 CLI host-browser mode depending on settings. Container-mode paths resolve inside Agent Zero; host-mode paths resolve on the connected A0 CLI host.
|
||||
|
||||
For complex browser workflows, load skill `browser-tool`. For fragile forms, load skill `browser-forms`.
|
||||
For complex browser workflows, load skill `browser-automation`. For fragile forms, load skill `browser-form-workflows`.
|
||||
|
||||
Actions: `open`, `list`, `state`, `set_active`, `navigate`, `back`, `forward`, `reload`, `content`, `detail`, `screenshot`, `click`, `hover`, `double_click`, `right_click`, `drag`, `type`, `submit`, `type_submit`, `scroll`, `evaluate`, `key_chord`, `mouse`, `wheel`, `keyboard`, `clipboard`, `set_viewport`, `select_option`, `set_checked`, `upload_file`, `multi`, `close`, `close_all`.
|
||||
|
||||
|
|
@ -16,7 +16,7 @@ Workflow:
|
|||
- `content` returns markdown with refs like `[link 3]`, `[button 6]`, `[input text 8]`.
|
||||
- Interactions use refs from the latest `content` capture.
|
||||
- `navigate` reuses an existing `browser_id` and is preferred for serial browsing.
|
||||
- Screenshots are explicit only; call `vision_load` with the returned path before reasoning visually.
|
||||
- Screenshots are explicit only; the browser does not automatically load screenshots. Call `vision_load` with the returned path before reasoning visually.
|
||||
- Keep the tab set small; close pages after extracting what you need.
|
||||
|
||||
`multi` is only a browser action: use `tool_name: "browser"` with `tool_args.action: "multi"`. Never use `tool_name: "multi"`.
|
||||
|
|
|
|||
|
|
@ -1,5 +1,5 @@
|
|||
---
|
||||
name: browser-tool
|
||||
name: browser-automation
|
||||
description: Use for complex Agent Zero browser automation, including multi-tab browsing, screenshots, forms, uploads, raw pointer/keyboard actions, host-vs-container browser mode, and visual verification workflows.
|
||||
---
|
||||
|
||||
|
|
@ -39,7 +39,7 @@ Screenshot args include `quality`, `full_page`, and optional `path`. PNG is used
|
|||
- `select_option` works for native selects and detectable ARIA listbox/combobox controls.
|
||||
- `set_checked` works for checkbox, radio, switch, and toggle-like refs.
|
||||
- `upload_file` works for file input refs or associated labels; verify the file exists in the active browser environment.
|
||||
- For fragile forms, load skill `browser-forms`.
|
||||
- For fragile forms, load skill `browser-form-workflows`.
|
||||
|
||||
## Pointer And Keyboard
|
||||
|
||||
|
|
@ -1,7 +1,6 @@
|
|||
---
|
||||
name: a0-browser-ext
|
||||
description: Create, inspect, install, and safely maintain Chrome extensions for Agent Zero's built-in Browser plugin.
|
||||
tags: ["agent-zero", "browser", "chrome-extension", "playwright", "manifest-v3"]
|
||||
name: browser-extension-control
|
||||
description: Create, inspect, install, and safely maintain Chrome extensions for Agent Zero's built-in Browser plugin. Use when the user asks to build a browser extension, modify an existing extension, install a Chrome Web Store extension, or review extension permissions.
|
||||
---
|
||||
|
||||
# Agent Zero Browser Extensions
|
||||
|
|
@ -1,5 +1,5 @@
|
|||
---
|
||||
name: browser-forms
|
||||
name: browser-form-workflows
|
||||
description: Use for complex Agent Zero Browser form workflows involving selects, checkboxes, radios, file uploads, contenteditable fields, multi-step validation, or visually verified submission.
|
||||
---
|
||||
|
||||
|
|
@ -446,7 +446,7 @@ const model = {
|
|||
createExtensionWithAgent() {
|
||||
this._prefillAgentPrompt(
|
||||
[
|
||||
"Use the a0-browser-ext skill to create a new Chrome extension for Agent Zero's Browser.",
|
||||
"Use the browser-extension-control skill to create a new Chrome extension for Agent Zero's Browser.",
|
||||
"Start by asking me for the extension name, purpose, target websites, and required permissions.",
|
||||
`Create it under ${this.extensionsRoot || EXTENSIONS_ROOT}/<extension-slug> and keep permissions minimal.`,
|
||||
].join("\n")
|
||||
|
|
@ -457,12 +457,12 @@ const model = {
|
|||
const url = String(this.extensionInstallUrl || "").trim();
|
||||
const prompt = url
|
||||
? [
|
||||
"Use the a0-browser-ext skill to review and optionally install this Chrome Web Store extension for Agent Zero's Browser.",
|
||||
"Use the browser-extension-control skill to review and optionally install this Chrome Web Store extension for Agent Zero's Browser.",
|
||||
`Chrome Web Store URL or id: ${url}`,
|
||||
"Explain the permissions and any sandbox risk before enabling it.",
|
||||
].join("\n")
|
||||
: [
|
||||
"Use the a0-browser-ext skill to help me install and review a Chrome Web Store extension for Agent Zero's Browser.",
|
||||
"Use the browser-extension-control skill to help me install and review a Chrome Web Store extension for Agent Zero's Browser.",
|
||||
"Ask me for the Chrome Web Store URL or extension id first.",
|
||||
"Explain the permissions and any sandbox risk before enabling it.",
|
||||
].join("\n");
|
||||
|
|
|
|||
|
|
@ -1,9 +1,10 @@
|
|||
# Behavioral prompt includes
|
||||
"{{name_pattern}}" files in workdir auto-injected into system prompt
|
||||
create/edit/delete persist across conversations
|
||||
preference change/remember/note > MUST persist via text_editor before responding
|
||||
never just acknowledge verbally always persist to file
|
||||
use for persistent notes knowledge project context
|
||||
preference changes, instruction files, project notes, and prompt includes > persist via text_editor before responding
|
||||
explicit memory requests like "remember this", "what did I ask you to remember", or "forget this" > use memory tools, not promptinclude files, unless the user asks to edit a file
|
||||
never just acknowledge durable project/instruction changes verbally; persist them to file when the user asks for a file/instruction/preference change
|
||||
use promptinclude files for persistent project context and behavioral instructions
|
||||
recursive search alphabetical by full path
|
||||
{{if includes}}
|
||||
### includes
|
||||
|
|
|
|||
|
|
@ -45,6 +45,7 @@ usage:
|
|||
|
||||
#### patch
|
||||
edit existing file. prefer patch_text; use edits only right after read for tiny line edits
|
||||
if the user says patch, change without rewriting, or don't rewrite, use action patch instead of write
|
||||
args path plus exactly one of: patch_text string OR edits [{from to content}]
|
||||
patch_text uses current file content, no prior read required
|
||||
patch_text update-only forms:
|
||||
|
|
|
|||
|
|
@ -1,5 +1,6 @@
|
|||
## skills
|
||||
use `skills_tool` action `search` when the user's wording sounds like a task, trigger phrase, or keyword match for a skill
|
||||
if the user asks to find/search a skill, search first before loading
|
||||
use `skills_tool` action `list` when you need a broader catalog view
|
||||
use `skills_tool` action `load` before following a skill
|
||||
loaded skills may document beta/specialized tools not in the always-on tool list; use them only after loading the skill
|
||||
|
|
|
|||
|
|
@ -1,5 +1,6 @@
|
|||
### document_query
|
||||
read local or remote documents or answer questions about them
|
||||
prefer this over text_editor when the user says "check/read this document" or asks a question about a document path/url, even for Markdown files
|
||||
args:
|
||||
- `document`: url path or list of them
|
||||
- `queries`: optional list of questions
|
||||
|
|
|
|||
|
|
@ -3,4 +3,6 @@ send an out-of-band notification without ending the current task
|
|||
args: `message`, optional `title`, `detail`, `type`, `priority`, `timeout`
|
||||
types: `info`, `success`, `warning`, `error`, `progress`
|
||||
priority values: `20` high urgency, `10` normal urgency; omit for high
|
||||
normal note/notification -> set `type: "info"` and `priority: 10`
|
||||
use `success` only for a completed success message, not for a generic note
|
||||
use for progress or alerts, not as the final answer
|
||||
|
|
|
|||
|
|
@ -1,5 +1,5 @@
|
|||
### scheduler
|
||||
Manage saved tasks and schedules. For complex task work, load skill `scheduler-tasks`.
|
||||
Manage saved tasks and schedules. For complex task work, load skill `scheduled-tasks`.
|
||||
|
||||
Actions: `list_tasks`, `find_task_by_name`, `show_task`, `run_task`, `update_task`, `delete_task`, `create_scheduled_task`, `create_adhoc_task`, `create_planned_task`, `wait_for_task`.
|
||||
|
||||
|
|
@ -10,4 +10,7 @@ Rules:
|
|||
- Do not run scheduled/planned tasks unless the user asks to run now.
|
||||
- Do not create recursive task prompts that schedule more tasks.
|
||||
- New tasks use a dedicated context unless `dedicated_context` is `false`.
|
||||
- Use IANA timezones like `Europe/Rome`; omit timezone to use the current user timezone.
|
||||
- Use `create_scheduled_task` for recurring/cron tasks; `schedule` must be cron fields, not an ISO datetime.
|
||||
- For one planned date/time, use `create_planned_task` with `plan: ["YYYY-MM-DDTHH:MM:SS"]`.
|
||||
- Use IANA timezones like `Europe/Rome`; include timezone when the user names a timezone.
|
||||
- For "tomorrow at 9:15 Rome time", scheduled shape is `schedule: {"minute":"15","hour":"9","day":"11","month":"5","weekday":"*","timezone":"Europe/Rome"}`.
|
||||
|
|
|
|||
|
|
@ -7,6 +7,8 @@ workflow:
|
|||
- action `list`: discover available skills
|
||||
- action `load`: load one skill by `skill_name`
|
||||
- action `read_file`: open one file inside a loaded skill directory
|
||||
if the user says "find/search a skill", call `search` before `load` even when the likely skill name seems obvious
|
||||
`read_file` requires both `skill_name` and `file_path`; load the skill first, then read `SKILL.md` or the named relative file
|
||||
after loading a skill, follow its instructions and use referenced files or scripts with other tools
|
||||
reload a skill if its instructions are no longer in context
|
||||
example:
|
||||
|
|
|
|||
|
|
@ -33,7 +33,7 @@ This skill provides comprehensive, accurate guidance for extending and building
|
|||
- Build **API Endpoints** for the Web UI
|
||||
- Create **Agent Profiles** (subordinates) with custom prompts
|
||||
- Understand and extend the **Prompt System**
|
||||
- Create **Skills** (see the dedicated `create-skill` skill for the full wizard)
|
||||
- Create **Skills** (see the dedicated `build-skill` skill for the full wizard)
|
||||
- Work with **Projects** and workspace configuration
|
||||
|
||||
> **Path convention:** Throughout this guide, `/a0/` refers to the framework root — this is `/a0/` inside Docker, or your local repository root in development. All paths are relative to this root.
|
||||
|
|
@ -41,7 +41,7 @@ This skill provides comprehensive, accurate guidance for extending and building
|
|||
> [!IMPORTANT]
|
||||
> **Plugins are the primary way to extend Agent Zero.** Most new tools, extensions, and prompts should be packaged as plugins. For all plugin tasks (create, review, manage, debug, contribute), load the `a0-plugin-router` skill which routes to the appropriate specialist. This guide covers the underlying framework patterns that plugins build upon.
|
||||
|
||||
Related skills: `a0-plugin-router` (plugin tasks) | `create-skill` (skill creation wizard) | `a0-create-plugin` | `a0-review-plugin` | `a0-manage-plugin` | `a0-contribute-plugin` | `a0-debug-plugin`
|
||||
Related skills: `a0-plugin-router` (plugin tasks) | `build-skill` (skill creation wizard) | `a0-create-plugin` | `a0-review-plugin` | `a0-manage-plugin` | `a0-contribute-plugin` | `a0-debug-plugin`
|
||||
|
||||
---
|
||||
|
||||
|
|
@ -631,7 +631,7 @@ The agent interacts with skills through JSON tool calls:
|
|||
{"tool_name": "skills_tool:load", "tool_args": {"skill_name": "my-skill"}}
|
||||
```
|
||||
|
||||
> For the complete skill creation wizard — including SKILL.md format, frontmatter fields, directory structure, best practices, and examples — load the `create-skill` skill.
|
||||
> For the complete skill creation wizard — including SKILL.md format, frontmatter fields, directory structure, best practices, and examples — load the `build-skill` skill.
|
||||
|
||||
---
|
||||
|
||||
|
|
|
|||
72
skills/build-skill/SKILL.md
Normal file
72
skills/build-skill/SKILL.md
Normal file
|
|
@ -0,0 +1,72 @@
|
|||
---
|
||||
name: build-skill
|
||||
description: Build or improve Agent Zero skills following the official SKILL.md standard. Use when the user asks to create, rename, move, audit, test, or refactor a skill, or when a workflow should be packaged as reusable skill instructions.
|
||||
---
|
||||
|
||||
# Build Skill
|
||||
|
||||
Skills are small folders that teach Agent Zero a repeatable workflow. Keep the always-visible metadata precise, keep `SKILL.md` lean, and move detailed material into scripts, references, or assets only when the task needs it.
|
||||
|
||||
## Standard Shape
|
||||
|
||||
Every skill folder must be named exactly like the skill and contain `SKILL.md`.
|
||||
|
||||
```text
|
||||
skill-name/
|
||||
├── SKILL.md
|
||||
├── scripts/ # optional deterministic helpers
|
||||
├── references/ # optional details loaded only when needed
|
||||
└── assets/ # optional output resources or templates
|
||||
```
|
||||
|
||||
The frontmatter should contain only `name` and `description`:
|
||||
|
||||
```yaml
|
||||
---
|
||||
name: skill-name
|
||||
description: What the skill does and when to use it. Include trigger wording here because this is the part visible before the skill loads.
|
||||
---
|
||||
```
|
||||
|
||||
Use lowercase letters, digits, and hyphens. Prefer short verb-led names such as `build-skill`, `review-plugin`, or `host-file-editing`.
|
||||
|
||||
## Workflow
|
||||
|
||||
1. Identify two or three real user requests that should trigger the skill.
|
||||
2. Decide whether the skill belongs in core `skills/` or inside a plugin's `plugins/<plugin>/skills/` directory.
|
||||
3. Write the frontmatter description with all trigger conditions and key contexts.
|
||||
4. Keep the body focused on procedure, contracts, failure handling, and the files/scripts to load next.
|
||||
5. Move long examples, schemas, policies, or variant-specific detail to one-level-deep `references/` files.
|
||||
6. Add scripts only for deterministic or repeatedly rewritten operations, and test representative scripts.
|
||||
7. Validate by searching, loading, and using the skill on a median-user prompt.
|
||||
|
||||
## Placement
|
||||
|
||||
Use plugin-scoped skills when the skill exists to explain a plugin-owned tool or UI surface. Examples: Browser workflows belong under `_browser`; A0 CLI host tools belong under `_a0_connector`; Desktop canvas workflows belong under `_desktop`.
|
||||
|
||||
Use root `skills/` for Agent Zero framework workflows that are not owned by one plugin, such as building skills, developing core features, or managing community plugins.
|
||||
|
||||
## Writing Rules
|
||||
|
||||
- Put trigger language in `description`, not in a body section. The body is loaded only after the skill is already selected.
|
||||
- Do not add README, changelog, quick reference, or install guide files inside the skill.
|
||||
- Avoid compatibility aliases in prompts or skill bodies when renaming; update references to the new name.
|
||||
- Prefer concise examples over long prose.
|
||||
- Do not duplicate the same guidance in both `SKILL.md` and references.
|
||||
- When multiple skills could apply, keep each skill's responsibility narrow and name the handoff clearly.
|
||||
|
||||
## Validation
|
||||
|
||||
Run targeted checks after edits:
|
||||
|
||||
```bash
|
||||
conda run -n a0 pytest tests/test_skills_runtime.py tests/test_tool_action_contracts.py -q
|
||||
```
|
||||
|
||||
Also exercise the live path when the skill changes agent-facing tool behavior:
|
||||
|
||||
```text
|
||||
1. Ask a short ordinary prompt that should discover the skill.
|
||||
2. Confirm the agent uses skills_tool search/load when appropriate.
|
||||
3. Confirm it calls the intended tool only after the skill is loaded when the tool is skill-gated.
|
||||
```
|
||||
|
|
@ -1,288 +0,0 @@
|
|||
---
|
||||
name: "create-skill"
|
||||
description: "Wizard for creating new Agent Zero skills. Guides users through creating well-structured SKILL.md files. Use when users want to create custom skills."
|
||||
version: "1.0.0"
|
||||
author: "Agent Zero Team"
|
||||
tags: ["meta", "wizard", "creation", "tutorial", "skills"]
|
||||
trigger_patterns:
|
||||
- "create skill"
|
||||
- "new skill"
|
||||
- "make skill"
|
||||
- "add skill"
|
||||
- "skill wizard"
|
||||
---
|
||||
|
||||
# Create Skill Wizard
|
||||
|
||||
This skill helps you create new Agent Zero skills that follow the SKILL.md standard.
|
||||
|
||||
## Quick Start
|
||||
|
||||
To create a new skill, I'll guide you through these steps:
|
||||
|
||||
1. **Name & Purpose** - What should this skill do?
|
||||
2. **Trigger Patterns** - When should this skill activate?
|
||||
3. **Content Structure** - What instructions should the agent follow?
|
||||
4. **Supporting Files** - Any scripts or templates needed?
|
||||
|
||||
## SKILL.md Format
|
||||
|
||||
Every skill needs a `SKILL.md` file with YAML frontmatter:
|
||||
|
||||
```yaml
|
||||
---
|
||||
name: "skill-name"
|
||||
description: "Clear description of what this skill does and when to use it"
|
||||
version: "1.0.0"
|
||||
author: "Your Name"
|
||||
tags: ["category1", "category2"]
|
||||
trigger_patterns:
|
||||
- "keyword1"
|
||||
- "phrase that triggers this"
|
||||
---
|
||||
|
||||
# Skill Title
|
||||
|
||||
Your skill instructions go here...
|
||||
```
|
||||
|
||||
## Required Fields
|
||||
|
||||
| Field | Description | Example |
|
||||
|-------|-------------|---------|
|
||||
| `name` | Unique identifier (lowercase, hyphens) | `"code-review"` |
|
||||
| `description` | When/why to use this skill | `"Review code for quality and security issues"` |
|
||||
|
||||
## Optional Fields
|
||||
|
||||
| Field | Description | Example |
|
||||
|-------|-------------|---------|
|
||||
| `version` | Semantic version | `"1.0.0"` |
|
||||
| `author` | Creator name | `"Jane Developer"` |
|
||||
| `tags` | Categorization keywords | `["review", "quality"]` |
|
||||
| `trigger_patterns` | Words/phrases that activate skill | `["review", "check code"]` |
|
||||
| `allowed_tools` | Tools this skill can use | `["code_execution", "web_search"]` |
|
||||
|
||||
## Skill Directory Structure
|
||||
|
||||
```
|
||||
/a0/usr/skills/
|
||||
└── my-skill/
|
||||
├── SKILL.md # Required: Main skill file
|
||||
├── scripts/ # Optional: Helper scripts
|
||||
│ ├── helper.py
|
||||
│ └── process.sh
|
||||
├── templates/ # Optional: Templates
|
||||
│ └── output.md
|
||||
└── docs/ # Optional: Additional docs
|
||||
└── examples.md
|
||||
```
|
||||
|
||||
## Writing Good Skill Instructions
|
||||
|
||||
### Be Specific and Actionable
|
||||
|
||||
```markdown
|
||||
# Good
|
||||
When reviewing code:
|
||||
1. Check for security vulnerabilities
|
||||
2. Verify error handling
|
||||
3. Assess test coverage
|
||||
|
||||
# Bad
|
||||
Review the code and make it better.
|
||||
```
|
||||
|
||||
### Include Examples
|
||||
|
||||
```markdown
|
||||
## Example Usage
|
||||
|
||||
**User**: "Review my Python function for issues"
|
||||
|
||||
**Agent Response**:
|
||||
> I'll review your function using the code review checklist:
|
||||
>
|
||||
> 1. **Security**: No user input validation detected
|
||||
> 2. **Error Handling**: Missing try-catch for file operations
|
||||
> 3. **Testing**: Function is testable but no tests found
|
||||
```
|
||||
|
||||
### Provide Checklists
|
||||
|
||||
```markdown
|
||||
## Review Checklist
|
||||
- [ ] Input validation present
|
||||
- [ ] Error handling complete
|
||||
- [ ] Tests included
|
||||
- [ ] Documentation updated
|
||||
```
|
||||
|
||||
## Creating Your Skill: Step by Step
|
||||
|
||||
### Step 1: Define Purpose
|
||||
|
||||
Answer these questions:
|
||||
- What problem does this skill solve?
|
||||
- When should the agent use it?
|
||||
- What's the expected output?
|
||||
|
||||
### Step 2: Choose a Name
|
||||
|
||||
- Use lowercase letters and hyphens
|
||||
- Be descriptive but concise
|
||||
- Examples: `code-review`, `data-analysis`, `deploy-helper`
|
||||
|
||||
### Step 3: Write Trigger Patterns
|
||||
|
||||
List words/phrases that should activate this skill:
|
||||
|
||||
```yaml
|
||||
trigger_patterns:
|
||||
- "review"
|
||||
- "check code"
|
||||
- "code quality"
|
||||
- "pull request"
|
||||
```
|
||||
|
||||
### Step 4: Structure Your Content
|
||||
|
||||
Organize with clear sections:
|
||||
|
||||
```markdown
|
||||
# Skill Title
|
||||
|
||||
## When to Use
|
||||
Describe the trigger conditions
|
||||
|
||||
## The Process
|
||||
Step-by-step instructions
|
||||
|
||||
## Examples
|
||||
Show sample interactions
|
||||
|
||||
## Tips
|
||||
Additional guidance
|
||||
```
|
||||
|
||||
### Step 5: Add Supporting Files (Optional)
|
||||
|
||||
If your skill needs scripts or templates:
|
||||
|
||||
```bash
|
||||
# Create directory structure
|
||||
mkdir -p /a0/usr/skills/my-skill/{scripts,templates,docs}
|
||||
```
|
||||
|
||||
## Example: Complete Skill
|
||||
|
||||
```yaml
|
||||
---
|
||||
name: "python-optimizer"
|
||||
description: "Optimize Python code for performance and readability. Use when asked to improve or optimize Python code."
|
||||
version: "1.0.0"
|
||||
author: "Agent Zero Team"
|
||||
tags: ["python", "optimization", "performance"]
|
||||
trigger_patterns:
|
||||
- "optimize python"
|
||||
- "improve performance"
|
||||
- "make faster"
|
||||
- "python optimization"
|
||||
---
|
||||
|
||||
# Python Optimizer
|
||||
|
||||
## When to Use
|
||||
Activate when user asks to optimize, improve, or speed up Python code.
|
||||
|
||||
## Optimization Process
|
||||
|
||||
### Step 1: Profile First
|
||||
Before optimizing, understand where time is spent:
|
||||
```python
|
||||
import cProfile
|
||||
cProfile.run('your_function()')
|
||||
```
|
||||
|
||||
### Step 2: Common Optimizations
|
||||
|
||||
1. **Use List Comprehensions**
|
||||
```python
|
||||
# Slow
|
||||
result = []
|
||||
for x in data:
|
||||
result.append(x * 2)
|
||||
|
||||
# Fast
|
||||
result = [x * 2 for x in data]
|
||||
```
|
||||
|
||||
2. **Use Sets for Lookups**
|
||||
```python
|
||||
# Slow: O(n)
|
||||
if item in large_list:
|
||||
|
||||
# Fast: O(1)
|
||||
if item in large_set:
|
||||
```
|
||||
|
||||
3. **Use Generators for Large Data**
|
||||
```python
|
||||
# Memory-heavy
|
||||
data = [process(x) for x in huge_list]
|
||||
|
||||
# Memory-efficient
|
||||
data = (process(x) for x in huge_list)
|
||||
```
|
||||
|
||||
### Step 3: Verify Improvement
|
||||
Always measure before and after:
|
||||
```python
|
||||
import time
|
||||
start = time.perf_counter()
|
||||
# code to measure
|
||||
elapsed = time.perf_counter() - start
|
||||
print(f"Took {elapsed:.4f} seconds")
|
||||
```
|
||||
|
||||
## Anti-Patterns to Avoid
|
||||
- Premature optimization
|
||||
- Optimizing without profiling
|
||||
- Sacrificing readability for tiny gains
|
||||
```
|
||||
|
||||
## Skill Installation
|
||||
|
||||
### Local Installation
|
||||
|
||||
1. Create skill directory:
|
||||
```bash
|
||||
mkdir -p /a0/usr/skills/my-skill
|
||||
```
|
||||
|
||||
2. Create SKILL.md:
|
||||
```bash
|
||||
touch /a0/usr/skills/my-skill/SKILL.md
|
||||
```
|
||||
|
||||
3. Add content and save
|
||||
|
||||
4. Skills are automatically loaded on next agent initialization
|
||||
|
||||
### Sharing Skills
|
||||
|
||||
To share skills with others:
|
||||
|
||||
1. Create a GitHub repository
|
||||
2. Include the skill directory structure
|
||||
3. Add a README with installation instructions
|
||||
4. Users can copy to their `/a0/usr/skills/` directory
|
||||
|
||||
## Testing Your Skill
|
||||
|
||||
After creating a skill:
|
||||
|
||||
1. Start a new conversation
|
||||
2. Use one of your trigger patterns
|
||||
3. Verify the agent follows your instructions
|
||||
4. Iterate and improve based on results
|
||||
|
|
@ -1,5 +1,5 @@
|
|||
---
|
||||
name: scheduler-tasks
|
||||
name: scheduled-tasks
|
||||
description: Use for complex Agent Zero scheduler work, including creating, updating, deleting, running, waiting for, timezone-correcting, or auditing scheduled, planned, and adhoc tasks.
|
||||
---
|
||||
|
||||
|
|
@ -22,7 +22,7 @@ Use the `scheduler` tool to manage saved tasks. Always inspect existing tasks be
|
|||
|
||||
## Schedule Fields
|
||||
|
||||
Schedules use cron-like fields:
|
||||
Schedules use cron-like fields. Do not put ISO datetimes into `schedule`.
|
||||
|
||||
- `minute`
|
||||
- `hour`
|
||||
|
|
@ -31,7 +31,38 @@ Schedules use cron-like fields:
|
|||
- `weekday`
|
||||
- `timezone`
|
||||
|
||||
Use IANA timezones such as `Europe/Rome`. Omit timezone to use the current user timezone. Planned task datetimes should be ISO strings such as `2026-05-09T18:25:00`.
|
||||
Use IANA timezones such as `Europe/Rome`. Omit timezone to use the current user timezone. Planned task datetimes go in `plan`, not `schedule`, and should be ISO strings such as `2026-05-09T18:25:00`.
|
||||
|
||||
For one future reminder, prefer `create_planned_task` with:
|
||||
|
||||
```json
|
||||
{
|
||||
"action": "create_planned_task",
|
||||
"name": "drink water",
|
||||
"prompt": "Remind the user to drink water.",
|
||||
"plan": ["2026-05-11T09:15:00"],
|
||||
"dedicated_context": true
|
||||
}
|
||||
```
|
||||
|
||||
For a recurring or cron-shaped scheduled task, use `create_scheduled_task` with:
|
||||
|
||||
```json
|
||||
{
|
||||
"action": "create_scheduled_task",
|
||||
"name": "weekday stretch",
|
||||
"prompt": "Remind the user to stretch.",
|
||||
"schedule": {
|
||||
"minute": "15",
|
||||
"hour": "9",
|
||||
"day": "*",
|
||||
"month": "*",
|
||||
"weekday": "1-5",
|
||||
"timezone": "Europe/Rome"
|
||||
},
|
||||
"dedicated_context": true
|
||||
}
|
||||
```
|
||||
|
||||
## Safety
|
||||
|
||||
|
|
@ -133,7 +133,14 @@ def test_remote_file_and_exec_tools_are_standard_tool_prompts_independent_from_c
|
|||
|
||||
|
||||
def test_beta_computer_use_remote_is_skill_only_not_standard_tool_prompt():
|
||||
skill = PROJECT_ROOT / "skills" / "computer-use-remote" / "SKILL.md"
|
||||
skill = (
|
||||
PROJECT_ROOT
|
||||
/ "plugins"
|
||||
/ "_a0_connector"
|
||||
/ "skills"
|
||||
/ "host-computer-use"
|
||||
/ "SKILL.md"
|
||||
)
|
||||
|
||||
assert not (PROMPT_ROOT / "agent.system.tool.computer_use_remote.md").exists()
|
||||
assert '"tool_name": "computer_use_remote"' in skill.read_text(encoding="utf-8")
|
||||
|
|
@ -288,7 +295,7 @@ def test_remote_affordance_skills_parse():
|
|||
/ "plugins"
|
||||
/ "_a0_connector"
|
||||
/ "skills"
|
||||
/ "text-editor-remote"
|
||||
/ "host-file-editing"
|
||||
/ "SKILL.md"
|
||||
)
|
||||
code_execution_skill = _parse_skill_frontmatter(
|
||||
|
|
@ -296,32 +303,43 @@ def test_remote_affordance_skills_parse():
|
|||
/ "plugins"
|
||||
/ "_a0_connector"
|
||||
/ "skills"
|
||||
/ "code-execution-remote"
|
||||
/ "host-code-execution"
|
||||
/ "SKILL.md"
|
||||
)
|
||||
computer_skill = _parse_skill_frontmatter(
|
||||
PROJECT_ROOT / "skills" / "computer-use-remote" / "SKILL.md"
|
||||
PROJECT_ROOT
|
||||
/ "plugins"
|
||||
/ "_a0_connector"
|
||||
/ "skills"
|
||||
/ "host-computer-use"
|
||||
/ "SKILL.md"
|
||||
)
|
||||
|
||||
assert not legacy_connector_skill.exists()
|
||||
assert text_editor_skill["name"] == "text-editor-remote"
|
||||
assert text_editor_skill["allowed_tools"] == ["text_editor_remote"]
|
||||
assert "connected local files" in text_editor_skill["trigger_patterns"]
|
||||
assert "not docker" in code_execution_skill["trigger_patterns"]
|
||||
assert "connected local terminal" in code_execution_skill["trigger_patterns"]
|
||||
assert code_execution_skill["name"] == "code-execution-remote"
|
||||
assert code_execution_skill["allowed_tools"] == ["code_execution_remote"]
|
||||
assert computer_skill["name"] == "computer-use-remote"
|
||||
assert computer_skill["allowed_tools"] == ["computer_use_remote"]
|
||||
assert text_editor_skill["name"] == "host-file-editing"
|
||||
assert "text_editor_remote" in text_editor_skill["description"]
|
||||
assert "not Docker/server files" in text_editor_skill["description"]
|
||||
assert code_execution_skill["name"] == "host-code-execution"
|
||||
assert "code_execution_remote" in code_execution_skill["description"]
|
||||
assert "not Docker" in code_execution_skill["description"]
|
||||
assert computer_skill["name"] == "host-computer-use"
|
||||
assert "computer_use_remote" in computer_skill["description"]
|
||||
|
||||
|
||||
def test_remote_tool_stubs_are_self_contained_and_reference_per_tool_skills():
|
||||
text_stub = (PROMPT_ROOT / "agent.system.tool.text_editor_remote.md").read_text(encoding="utf-8")
|
||||
exec_stub = (PROMPT_ROOT / "agent.system.tool.code_execution_remote.md").read_text(encoding="utf-8")
|
||||
computer_skill = (PROJECT_ROOT / "skills" / "computer-use-remote" / "SKILL.md").read_text(encoding="utf-8")
|
||||
computer_skill = (
|
||||
PROJECT_ROOT
|
||||
/ "plugins"
|
||||
/ "_a0_connector"
|
||||
/ "skills"
|
||||
/ "host-computer-use"
|
||||
/ "SKILL.md"
|
||||
).read_text(encoding="utf-8")
|
||||
|
||||
assert "optionally load skill `text-editor-remote`" in text_stub
|
||||
assert "optionally load skill `code-execution-remote`" in exec_stub
|
||||
assert "optionally load skill `host-file-editing`" in text_stub
|
||||
assert "optionally load skill `host-code-execution`" in exec_stub
|
||||
assert '"tool_name": "text_editor_remote"' in text_stub
|
||||
assert '"tool_name": "code_execution_remote"' in exec_stub
|
||||
assert '"tool_name": "computer_use_remote"' in computer_skill
|
||||
|
|
|
|||
|
|
@ -528,7 +528,14 @@ def test_browser_extension_menu_exposes_agent_and_url_paths():
|
|||
html = (PROJECT_ROOT / "plugins" / "_browser" / "webui" / "browser-panel.html").read_text(
|
||||
encoding="utf-8"
|
||||
)
|
||||
skill = PROJECT_ROOT / "skills" / "a0-browser-ext" / "SKILL.md"
|
||||
skill = (
|
||||
PROJECT_ROOT
|
||||
/ "plugins"
|
||||
/ "_browser"
|
||||
/ "skills"
|
||||
/ "browser-extension-control"
|
||||
/ "SKILL.md"
|
||||
)
|
||||
|
||||
assert "Create New Extension with A0" in html
|
||||
assert "+ Create New with A0" not in html
|
||||
|
|
@ -902,19 +909,19 @@ def test_browser_tool_does_not_auto_open_canvas_policy_is_documented():
|
|||
assert "select_option" in prompt
|
||||
assert "set_checked" in prompt
|
||||
assert "upload_file" in prompt
|
||||
assert "browser-forms" in prompt
|
||||
assert "browser-form-workflows" in prompt
|
||||
assert "does not automatically load screenshots" in prompt
|
||||
assert "already open" in config
|
||||
assert "already-open Browser surface" in config_html
|
||||
|
||||
|
||||
def test_browser_forms_skill_is_plugin_owned_and_discoverable():
|
||||
skill_path = PROJECT_ROOT / "plugins" / "_browser" / "skills" / "browser-forms" / "SKILL.md"
|
||||
skill_path = PROJECT_ROOT / "plugins" / "_browser" / "skills" / "browser-form-workflows" / "SKILL.md"
|
||||
assert skill_path.exists()
|
||||
skill = skill_path.read_text(encoding="utf-8")
|
||||
assert skill.startswith("---\n")
|
||||
frontmatter = skill.split("---", 2)[1]
|
||||
assert "name: browser-forms" in frontmatter
|
||||
assert "name: browser-form-workflows" in frontmatter
|
||||
assert "description:" in frontmatter
|
||||
assert "select_option" in skill
|
||||
assert "set_checked" in skill
|
||||
|
|
|
|||
|
|
@ -46,9 +46,13 @@ async def _build_system_text(profile: str = "agent0") -> str:
|
|||
async def test_default_agent0_prompt_budget_and_guardrails():
|
||||
system_text = await _build_system_text()
|
||||
|
||||
assert tokens.approximate_tokens(system_text) <= 3000
|
||||
assert "tool_name` must exactly match a listed tool name" in system_text
|
||||
assert "tool_args` must stay a json object" in system_text
|
||||
# The default prompt now intentionally includes the compact always-on tool
|
||||
# surface plus skill metadata. Keep the guardrail close to the observed
|
||||
# budget so prompt creep remains visible without pretending this surface is
|
||||
# a tiny single-tool prompt.
|
||||
assert tokens.approximate_tokens(system_text) <= 8500
|
||||
assert "`tool_name` must be one listed tool name" in system_text
|
||||
assert "- tool_args: key value pairs tool arguments" in system_text
|
||||
assert '"tool_name": "call_subordinate"' in system_text
|
||||
assert '"reset": true' in system_text
|
||||
assert '"tool_name": "text_editor"' in system_text
|
||||
|
|
@ -59,7 +63,7 @@ async def test_default_agent0_prompt_budget_and_guardrails():
|
|||
assert '"tool_name": "code_execution_remote"' in system_text
|
||||
assert '"tool_name": "text_editor_remote"' in system_text
|
||||
assert '"tool_name": "computer_use_remote"' not in system_text
|
||||
assert "computer-use-remote" in system_text
|
||||
assert "host-computer-use" in system_text
|
||||
|
||||
|
||||
def test_a0_small_profile_removed_and_prompt_text_generic():
|
||||
|
|
|
|||
|
|
@ -198,13 +198,13 @@ def test_reactivating_name_only_scope_default_by_path_clears_hidden_override(mon
|
|||
def test_loaded_skill_entries_come_from_agent_data():
|
||||
agent = DummyAgent()
|
||||
agent.data[runtime.AGENT_DATA_NAME_LOADED_SKILLS] = [
|
||||
"computer-use-remote",
|
||||
"host-computer-use",
|
||||
"",
|
||||
"a0-development",
|
||||
]
|
||||
|
||||
assert runtime.get_loaded_skill_entries(agent) == [
|
||||
{"name": "computer-use-remote"},
|
||||
{"name": "host-computer-use"},
|
||||
{"name": "a0-development"},
|
||||
]
|
||||
|
||||
|
|
@ -276,16 +276,41 @@ def test_a0_manage_plugin_skill_frontmatter_is_valid_yaml():
|
|||
assert "Agent Zero Plugin Management" in body
|
||||
|
||||
|
||||
def test_renamed_skills_use_standard_frontmatter_only():
|
||||
skill_paths = [
|
||||
PROJECT_ROOT / "skills" / "build-skill" / "SKILL.md",
|
||||
PROJECT_ROOT / "skills" / "scheduled-tasks" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_a0_connector" / "skills" / "host-code-execution" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_a0_connector" / "skills" / "host-computer-use" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_a0_connector" / "skills" / "host-file-editing" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_a0_connector" / "skills" / "setup-a0-cli" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_browser" / "skills" / "browser-automation" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_browser" / "skills" / "browser-extension-control" / "SKILL.md",
|
||||
PROJECT_ROOT / "plugins" / "_browser" / "skills" / "browser-form-workflows" / "SKILL.md",
|
||||
]
|
||||
|
||||
for path in skill_paths:
|
||||
frontmatter, body, errors = runtime.split_frontmatter(path.read_text(encoding="utf-8"))
|
||||
assert errors == []
|
||||
assert set(frontmatter) == {"name", "description"}
|
||||
assert frontmatter["name"] == path.parent.name
|
||||
assert frontmatter["description"]
|
||||
assert body
|
||||
|
||||
|
||||
def test_unload_agent_skill_removes_loaded_skill_by_name():
|
||||
agent = DummyAgent()
|
||||
agent.data[runtime.AGENT_DATA_NAME_LOADED_SKILLS] = [
|
||||
"computer-use-remote",
|
||||
"host-computer-use",
|
||||
"a0-development",
|
||||
]
|
||||
|
||||
removed = runtime.unload_agent_skill(
|
||||
agent,
|
||||
{"name": "computer-use-remote", "path": "/a0/skills/computer-use-remote"},
|
||||
{
|
||||
"name": "host-computer-use",
|
||||
"path": "/a0/plugins/_a0_connector/skills/host-computer-use",
|
||||
},
|
||||
)
|
||||
|
||||
assert removed is True
|
||||
|
|
|
|||
|
|
@ -66,7 +66,7 @@ def _load_skills_tool(monkeypatch, skill_root: Path):
|
|||
skills_stub.AGENT_DATA_NAME_LOADED_SKILLS = "loaded_skills"
|
||||
skills_stub.MAX_ACTIVE_SKILLS = 20
|
||||
fake_skill = _FakeSkill(
|
||||
name="browser-forms",
|
||||
name="browser-form-workflows",
|
||||
description="Use for complex browser forms.",
|
||||
path=skill_root,
|
||||
tags=[],
|
||||
|
|
@ -99,13 +99,13 @@ def test_skills_tool_accepts_action_alias_for_search(monkeypatch, tmp_path: Path
|
|||
|
||||
response = asyncio.run(tool.execute(**tool.args))
|
||||
|
||||
assert "browser-forms" in response.message
|
||||
assert "browser-form-workflows" in response.message
|
||||
|
||||
|
||||
def test_skills_tool_read_file_action_reads_inside_skill_dir(
|
||||
monkeypatch, tmp_path: Path
|
||||
):
|
||||
skill_root = tmp_path / "browser-forms"
|
||||
skill_root = tmp_path / "browser-form-workflows"
|
||||
skill_root.mkdir()
|
||||
(skill_root / "notes.md").write_text("Use labels before typing.\n", encoding="utf-8")
|
||||
module = _load_skills_tool(monkeypatch, skill_root)
|
||||
|
|
@ -115,7 +115,7 @@ def test_skills_tool_read_file_action_reads_inside_skill_dir(
|
|||
None,
|
||||
{
|
||||
"action": "read_file",
|
||||
"skill_name": "browser-forms",
|
||||
"skill_name": "browser-form-workflows",
|
||||
"file_path": "notes.md",
|
||||
},
|
||||
"",
|
||||
|
|
@ -124,7 +124,7 @@ def test_skills_tool_read_file_action_reads_inside_skill_dir(
|
|||
|
||||
response = asyncio.run(tool.execute(**tool.args))
|
||||
|
||||
assert "Skill file: browser-forms/notes.md" in response.message
|
||||
assert "Skill file: browser-form-workflows/notes.md" in response.message
|
||||
assert "Use labels before typing." in response.message
|
||||
|
||||
|
||||
|
|
@ -464,9 +464,10 @@ def test_computer_use_remote_is_skill_gated():
|
|||
/ "plugins/_a0_connector/prompts/agent.system.tool.computer_use_remote.md"
|
||||
)
|
||||
skill_text = (
|
||||
project_root / "skills/computer-use-remote/SKILL.md"
|
||||
project_root
|
||||
/ "plugins/_a0_connector/skills/host-computer-use/SKILL.md"
|
||||
).read_text(encoding="utf-8")
|
||||
|
||||
assert not prompt_path.exists()
|
||||
assert '"tool_name": "computer_use_remote"' in skill_text
|
||||
assert "Beta local desktop control" in skill_text
|
||||
assert "Beta desktop control" in skill_text
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue