CtrlK
BlogDocsLog inGet started
Tessl Logo

ainativedev/aidevcon-2026-ldn

AI Native DevCon 2026 London — all conference sessions as interactive skills

71

Quality

89%

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

Overview
Quality
Evals
Security
Files

Quality

Content

92%

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

This is a well-crafted Q&A skill that clearly defines how Claude should answer questions about a specific talk. Its strengths are the explicit grounding rules (verbatim quoting, no speculation), the clear lookup workflow across three reference files, and the concrete good/bad response example. The main limitation is the absence of bundle files to verify the referenced documents exist.

Suggestions

Include the referenced bundle files (outline.md, transcript.md, quotes.md) or confirm they exist in the bundle to fully support the progressive disclosure structure.

DimensionReasoningScore

Conciseness

The skill is lean and efficient. It doesn't explain what transcripts are, how fine-tuning works, or other concepts Claude already knows. Every section serves a clear purpose: grounding rules, response format, and reference pointers.

3 / 3

Actionability

The skill provides concrete, specific instructions: read outline.md first, quote verbatim from transcript.md, check quotes.md before the full transcript, never paraphrase as quotes. The good/bad response example makes the expected behavior unambiguous.

3 / 3

Workflow Clarity

The workflow is clearly sequenced: (1) read outline.md to locate section, (2) read relevant section of transcript.md, (3) check quotes.md for citable evidence, (4) respond with verbatim quotes and attributions. Validation is built in via the explicit rule to say so if a claim isn't in the transcript rather than speculate.

3 / 3

Progressive Disclosure

The skill references outline.md, transcript.md, and quotes.md with clear purposes for each, which is good structure. However, no bundle files were provided, so we cannot verify these references exist. The references are one-level deep and well-signaled, but the lack of actual bundle files prevents a score of 3.

2 / 3

Total

11

/

12

Passed

Description

100%

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

This is an excellent skill description that clearly identifies its niche (Brian Douglas's talk on training AI on your own code), lists specific concrete topics it covers, and provides explicit trigger guidance via a detailed 'Use when' clause. The description is rich in natural trigger terms that users would actually use, and its specificity to a particular speaker and set of tools makes it highly distinctive.

DimensionReasoningScore

Specificity

Lists multiple specific concrete actions and concepts: capturing agent sessions, extracting skills from traces, fine-tuning small local models, tapes/steros tooling, SFT vs DPO decisions, agent telemetry, QLoRA, parallel agents. These are highly specific capabilities.

3 / 3

Completeness

Clearly answers both what ('Answers questions about Brian Douglas's talk on training AI on your own code') and when ('Use when a user asks about Brian Douglas's pipeline...or wants to apply his agent telemetry and training data approach'). The 'Use when' clause is explicit and detailed.

3 / 3

Trigger Term Quality

Excellent coverage of natural terms a user would say: 'Brian Douglas', 'agent sessions', 'fine-tuning', 'training data', 'Claude Code', 'QLoRA', 'SFT vs DPO', 'tapes/steros', 'parallel agents'. These are the exact terms someone familiar with this talk would use.

3 / 3

Distinctiveness Conflict Risk

Highly distinctive — scoped to a specific person's talk with unique tooling names (tapes/steros), specific techniques (SFT vs DPO, QLoRA), and a clear niche around agent telemetry and training data pipelines. Very unlikely to conflict with other skills.

3 / 3

Total

12

/

12

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.

Validation11 / 11 Passed

Validation for skill structure

No warnings or errors.

Reviewed

Table of Contents