CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

moai-ref-aside-browser

Safe-use reference for the optional Aside browser agent (the `aside` command-line tool and its MCP server), which works inside the operator's own logged-in browser. Covers when the orchestrator may use it, how its permission modes work, and the rules that keep it read-only by default. Operated by the orchestrator alone. NOT for: public pages that a plain fetch can read, subagent work, or CI runs.

Quality

87%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Quality

Content

100%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is an exemplary compact reference: dense with tool-specific facts, concrete commands, explicit safety rules, and built-in failure handling, with no filler and no unnecessary external files. It models its stated constraints (read-only default, orchestrator-only operation, optional registration) precisely and consistently.

DimensionReasoningScore

Conciseness

The body is lean and every token earns its place: it contains only Aside-specific facts Claude could not know ('120-second timeout; `page` starts unset, so call `openTab(url)` or `getTabs()` first') with no padding or explanation of known concepts. Version sensitivity is handled with an explicit fallback ('Flag and tool names here reflect one observed version; if a probe or a call fails, treat Aside as absent') rather than a pinned version number.

5 / 5

Actionability

For an instruction-only skill the guidance is fully concrete: copy-paste ready commands ('moai mcp add aside --command aside --args mcp --scope user', 'aside --version', 'aside skills install') and specific executable rules ('Prefer to omit `--permission` entirely', 'restrict REPL work to navigation, reading, and screenshot operations', 'Redirect verbose REPL output to a file and bring only the exit code and a bounded tail into context') cover the common cases.

5 / 5

Workflow Clarity

This is a simple single-purpose skill under 50 lines whose action is unambiguous, and validation checkpoints are present rather than missing: an availability probe ('aside --version'), a failure fallback ('if a probe or a call fails, treat Aside as absent'), a silent-continue path when excluded ('for example when `CI=true`'), and mandatory operator confirmation before any state-changing step. Because verification steps exist, the destructive-operation cap at 3 does not apply.

5 / 5

Progressive Disclosure

The skill has no bundle files (references/, scripts/, and assets/ do not exist) and the body contains no dangling file references, so nothing needs external splitting. At under 50 lines with clear, well-organized sections (Purpose, When to use, Who operates it, Safe-use rules, Optional registration, Relation to /moai e2e) it qualifies for 5 under the rubric's simple-skill provision.

5 / 5

Total

20

/

20

Passed

Description

70%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, third-person, and highly distinctive with explicit boundary guidance, but it lacks a positive trigger clause: it tells when NOT to use the skill without saying when to use it, capping completeness. Folding the when_to_use triggers into the description would resolve the main weakness.

Suggestions

Add an explicit positive trigger clause to the description, e.g. 'Use when the operator asks to run `aside`, register the optional Aside MCP server, or explicitly requests `/moai e2e --tool aside`.'

Mention the two MCP tools (`exec` and `repl`) and the e2e entry point in the description so the 'what' coverage is comprehensive, not just the permission-mode rules.

Merge or mirror the when_to_use frontmatter content into description so the trigger guidance travels with the field most loaders actually surface.

DimensionReasoningScore

Specificity

The description names the concrete artifact ('the optional Aside browser agent (the `aside` command-line tool and its MCP server)') and enumerates several specific coverage areas ('when the orchestrator may use it, how its permission modes work, and the rules that keep it read-only by default'). It lists several specific actions with minor gaps (the exec/repl tools and the e2e entry point are not mentioned), matching anchor 4 rather than the comprehensive coverage of anchor 5 or the 1-2 actions of anchor 3.

4 / 5

Completeness

The 'what' is clear (safe-use reference covering permission modes and read-only rules for the Aside browser agent), but the only trigger guidance is negative: 'NOT for: public pages that a plain fetch can read, subagent work, or CI runs'. Per the judging guideline, a missing positive 'Use when...' clause or equivalent explicit trigger guidance caps completeness at 3; the positive triggers exist only in the separate when_to_use field, not the description.

3 / 5

Trigger Term Quality

Natural keywords are present with synonyms for the tool: 'Aside browser agent', '`aside` command-line tool', 'MCP server', 'logged-in browser'. Coverage is good but a few natural terms a user might say (e.g., 'e2e', 'browser automation', 'aside MCP') are missing from the description, so it fits anchor 4 rather than the comprehensive synonym coverage of anchor 5.

4 / 5

Distinctiveness Conflict Risk

It targets a clearly named niche tool ('the optional Aside browser agent (the `aside` command-line tool and its MCP server)') and explicitly excludes overlapping contexts ('NOT for: public pages that a plain fetch can read, subagent work, or CI runs'), giving it distinct triggers with minimal conflict risk, matching anchor 5.

5 / 5

Total

16

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

modu-ai/moai-adk

Path

.claude/skills/moai-ref-aside-browser/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.