browse-and-evaluate

1# Browse And Evaluate
2 
3## Goal
4 
5Find the right skill for a task without flooding the context window or installing blindly.
6 
7## Guardrails
8 
9- Always use `--fields` on list/search/info to keep output small. Default: `--fields name,tier,workArea,description`.
10- Always use `--dry-run` before installing anything.
11- Never install more than 3 skills at once without explicit user confirmation.
12- Prefer `--format json` in non-interactive pipelines. The CLI defaults to JSON when stdout is not a TTY.
13- Use `--limit` when browsing large catalogs. Start with `--limit 10`.
14 
15## Workflow
16 
171. Search or browse the catalog.
18 
19```bash
20npx ai-agent-skills search <query> --fields name,tier,workArea,description --limit 10
21```
22 
232. Get details on a candidate.
24 
25```bash
26npx ai-agent-skills info <skill-name> --fields name,description,tags,collections,installCommands
27```
28 
293. Preview the skill content.
30 
31```bash
32npx ai-agent-skills preview <skill-name>
33```
34 
354. Dry-run the install.
36 
37```bash
38npx ai-agent-skills install <skill-name> --dry-run
39```
40 
415. Install only after reviewing the dry-run output.
42 
43```bash
44npx ai-agent-skills install <skill-name>
45```
46 
47## Gotchas
48 
49- The `preview` command sanitizes skill content to strip prompt injection patterns. If content looks truncated, check if suspicious patterns were removed.
50- Collection installs pull multiple skills. Always `--list` or `--dry-run` a collection before installing.
51…