CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

talk-trieloff-browser-agents

Summarizes, explains, and applies Lars Trieloff's AI Native DevCon talk on browser-native agents. Use for browser agents, running AI in the browser, browser-as-runtime architecture, agent containment, local-versus-cloud tradeoffs, safe AI product integration, documented APIs, user consent, credential isolation, and reviewable agent actions.

Quality

79%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./Plugins/aidevcon/skills/talk-trieloff-browser-agents/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

67%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 well organized, concise, and offers genuinely actionable response templates with a clear reading sequence and conflict-resolution rule. Its critical defect is that the entire progressive-disclosure structure is broken: all three referenced grounding files are absent from the bundle, so the workflow cannot actually be executed as written.

Suggestions

Ship the referenced bundle files (outline.md, quote.md, transcript.md) or remove the grounding instructions that point to them — currently every reading step in Grounding Rules 1-2 and the 'grounding the comparison in outline.md' directive fails because none of these files exist in the bundle.

If the skill is meant to stay single-file, inline the essential advisory anchors and talk structure into SKILL.md and replace the file-lookup workflow with a self-contained one, which would also let progressive_disclosure score on the simple-skill exception.

Trim the redundant opening paragraph and the 'Core Concepts' list, both of which restate the frontmatter description, to tighten the conciseness gap to anchor 5.

DimensionReasoningScore

Conciseness

The body is lean — the two tables, response shapes, and worked example earn their tokens, and it never explains concepts Claude already knows. Minor trimmable redundancy exists: the opening paragraph ("Use this skill to summarize the talk, compare browser-native agents with cloud or desktop agents...") and the "Core Concepts" list largely restate the frontmatter description, keeping it below anchor 5.

4 / 5

Actionability

Concrete, usable templates: a four-part response shape ("Thesis... Design tension... Safe pattern... Boundary"), a six-row product-design table with an "Evidence To Require" column, a comparison table, and a worked Example Q&A. As an instruction-only skill, absent code is not penalized, but it falls short of anchor 5 because the substantive grounding depends on bundle files that are not present, and directives like "End with the next safest implementation step" stay abstract.

4 / 5

Workflow Clarity

The Grounding Rules give an explicit reading sequence ("Read `outline.md` first to locate the relevant concept", "verify details against `transcript.md`") plus an explicit conflict-resolution checkpoint ("If `outline.md`, `quote.md`, and `transcript.md` disagree, prefer the redacted transcript for safety boundaries and the outline for structure"). This exceeds anchor 3's implicit checkpoints but is not anchor 5 because there is no step verifying the grounding material was actually found and read.

4 / 5

Progressive Disclosure

Scored against the actual bundle structure: the body directs Claude to `outline.md`, `quote.md`, and `transcript.md`, but the bundle contains no such files (no references/, scripts/, or assets/ directories — only SKILL.md), so every grounding instruction points to a nonexistent file and the promised detailed layer is entirely missing. The SKILL.md's own sections are well organized, which keeps it above the monolithic/untraceable anchor 1.

2 / 5

Total

14

/

20

Passed

Description

83%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 strong: third-person voice, an explicit what-and-when pair, a named source that makes the skill distinct, and a rich set of natural trigger terms. The only gaps are a few missing synonyms (e.g., in-browser/client-side agents) and generic action verbs rather than more granular named capabilities.

DimensionReasoningScore

Specificity

"Summarizes, explains, and applies" names three concrete actions against a specific source ("Lars Trieloff's AI Native DevCon talk on browser-native agents"), going beyond the 1-2 actions of anchor 3. It is not a 5 because the verbs are generic and finer-grained capabilities shown in the body (comparing agent approaches, designing product boundaries) are absent from the description.

4 / 5

Completeness

"Summarizes, explains, and applies Lars Trieloff's AI Native DevCon talk on browser-native agents" explicitly answers what, and "Use for browser agents, running AI in the browser, browser-as-runtime architecture..." explicitly answers when with a long list of concrete trigger phrases — a direct anchor-5 match. The "Use for" clause is present, so the completeness cap of 3 does not apply.

5 / 5

Trigger Term Quality

Natural user phrasings like "browser agents", "running AI in the browser", "user consent", and "documented APIs" give good keyword coverage. Not a 5 because common synonyms such as "in-browser agents" or "client-side agents" are missing; clearly above the sparse keyword coverage of anchor 3.

4 / 5

Distinctiveness Conflict Risk

Anchoring to a named speaker and talk ("Lars Trieloff's AI Native DevCon talk") creates a clear niche with distinct triggers. Not a 5 because broad trigger terms like "user consent", "documented APIs", and "safe AI product integration" could fire for general agent-safety or product-design skills unrelated to this talk.

4 / 5

Total

17

/

20

Passed

Validation

93%

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

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

Total

15

/

16

Passed

Repository

jscraik/Agent-Skills

Path

Plugins/aidevcon/skills/talk-trieloff-browser-agents/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.