Skip to content

Commit 73f2b2c

Browse files
sjnimsclaude
andauthored
docs(agent-development): improve visibility of tools vs allowed-tools disambiguation (#28)
## Summary - Added prominent callout in Overview section for early visibility - Added contextual note in tools field section with cross-reference to skill-development skill - Implements Option C from issue (both locations) ## Problem Fixes #24 The important distinction between `tools` (agents) and `allowed-tools` (skills) was buried at line ~385 in the Quick Reference section, making it easy to miss when switching between component types. ## Solution Added the disambiguation in two strategic locations: 1. **Overview section** (line 19) - Early visibility with warning emoji callout 2. **tools field section** (line 204) - Contextual help when users are configuring tool restrictions, includes cross-reference to skill-development skill This dual placement ensures users see the warning whether they're scanning the document from the top or jumping directly to the tools field documentation. ### Alternatives Considered - **Option A (tools field only)**: Would miss users scanning from top - **Option B (Overview only)**: Would miss users jumping to tools field - **Option C (both)**: Selected - ensures visibility in both contexts ## Changes - `plugins/plugin-dev/skills/agent-development/SKILL.md`: Added 2 callout boxes ## Testing - [x] Markdownlint passes - [x] Both callouts render correctly in markdown - [x] Cross-reference to skill-development skill included --- 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude <noreply@anthropic.com>
1 parent 114b52c commit 73f2b2c

1 file changed

Lines changed: 4 additions & 0 deletions

File tree

  • plugins/plugin-dev/skills/agent-development

plugins/plugin-dev/skills/agent-development/SKILL.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -16,6 +16,8 @@ Agents are autonomous subprocesses that handle complex, multi-step tasks indepen
1616
- System prompt defines agent behavior
1717
- Model and color customization
1818

19+
> **⚠️ Field Name Difference:** Agents use `tools` to restrict tool access. Skills use `allowed-tools` for the same purpose. Don't confuse these when switching between component types.
20+
1921
## When to Use Agents vs Commands vs Skills
2022

2123
| Component | Best For | Triggering | Example Use Case |
@@ -199,6 +201,8 @@ tools: ["Read", "Write", "Grep", "Bash"]
199201
- Testing: `["Read", "Bash", "Grep"]`
200202
- Full access: Omit field or use `["*"]`
201203

204+
> **Important:** Agents use `tools` while Skills use `allowed-tools`. The field names differ between component types. For skill tool restrictions, see the `skill-development` skill.
205+
202206
## System Prompt Design
203207

204208
The markdown body becomes the agent's system prompt. Write in second person, addressing the agent directly.

0 commit comments

Comments
 (0)