CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

experiment-orchestration

Takes a product goal through a LaunchDarkly experiment and, if it wins, a guarded rollout (or a regular rollout on plans without guarded rollouts). Uses GitHub for the code change and deployment when it is connected to the agent space; otherwise hands those steps to a person.

Quality

79%

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

Fix and improve this skill with Tessl

tessl review fix ./skills/aws-devops-agent/experiment-orchestration/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

77%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 exceptionally actionable, well-sequenced 8-step workflow with strong validation gates and error-recovery loops — among the best on actionability and workflow clarity. Its weaknesses are repetition that inflates token cost and a fully monolithic structure with no reference files to move stable detail (report template, handoff URLs, retry procedure) out of the main file.

Suggestions

Move stable, rarely-changing detail — the Step 8 report template, the 'When a step can't succeed through you' UI instructions and URLs, and the retry-after-rollback procedure — into one-level-deep reference files linked from the relevant steps.

State the 'stop with control as the winner' rationale and the 'check periodically, don't tight-loop' guidance once each (in Step 7 and the preamble respectively) instead of repeating them across Steps 3, 5, 6, and 7.

Fix the stray empty list item '6.' in Step 1: Goal Clarification.

DimensionReasoningScore

Conciseness

The body is dense and operational rather than educational, but noticeably repetitive: the 'stop the iteration with control as the winner' rationale is explained in Step 6 item 3, Step 7 Phase 1, and again in the 'To stop the iteration' paragraph; the 'check periodically, don't tight-loop' guidance is restated in Steps 3, 5, and 7; and Step 1 contains a stray empty list item '6.'. This exceeds the 'minor instances' of anchor 4 and fits anchor 3 — mostly efficient but could be tightened.

3 / 5

Actionability

Fully executable instruction-only guidance: exact MCP tool calls ('call get-metrics with a query'), copy-paste-adaptable structures (the flags config, treatments, defaults JSON), precise unit conversions ('20% is 20000', '60 minutes is 3600000'), templated UI URLs, and a complete Step 8 report template cover the common cases, matching anchor 5.

5 / 5

Workflow Clarity

Eight clearly sequenced steps with explicit validation checkpoints (read the flag back after creation and verify the control default, release-readiness review before merge, confirm deployment before the experiment, first-day instrumentation check, significance plus analyzedUnitCount of at least 100 before acting) and feedback loops for error recovery (review failure fed back with a capped retry of ~3 attempts; rollback diagnosed as external vs. metric cause before retrying). This matches anchor 5.

5 / 5

Progressive Disclosure

No bundle files exist (references/, scripts/, assets/ are all absent), so all detail — the Step 8 report template, the UI handoff instructions with URLs, the retry-after-rollback procedure — is inlined in a single ~215-line file. Section and step structure is good, but content that clearly belongs in separate reference files is inline with no offloading references, fitting anchor 3; it is above anchor 2 because the organization itself is genuinely clear.

3 / 5

Total

16

/

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, domain-anchored, and highly distinct, with a clear statement of what the skill does and how it behaves with and without GitHub. Its main weakness is the absence of any explicit 'Use when...' trigger guidance, which caps completeness, plus missing the common synonyms 'A/B test' and 'feature flag'.

Suggestions

Add an explicit trigger clause, e.g. 'Use when the user wants to run an A/B test or LaunchDarkly experiment on a product change, or roll out a proven change gradually.'

Include the natural synonyms users actually say — 'A/B test', 'feature flag', 'feature rollout' — alongside 'experiment' and 'rollout' to broaden triggering.

Briefly mention the end-to-end scope (metrics, flags, monitoring, report) so the capability list is comprehensive, not just the experiment-and-rollout arc.

DimensionReasoningScore

Specificity

Names several concrete actions — 'Takes a product goal through a LaunchDarkly experiment', 'a guarded rollout (or a regular rollout on plans without guarded rollouts)', 'Uses GitHub for the code change and deployment', 'hands those steps to a person' — but omits capabilities the body covers (metric/flag creation, monitoring, reporting), so it is not comprehensively covered as anchor 5 requires.

4 / 5

Completeness

The 'what' is clear and concrete (experiment then guarded/regular rollout, GitHub for code and deployment, human handoff otherwise), but there is no 'Use when...' clause or equivalent explicit trigger guidance, which caps completeness at 3 per the judging guidelines.

3 / 5

Trigger Term Quality

'experiment', 'rollout', 'LaunchDarkly', and 'GitHub' are natural terms for this audience, giving good keyword coverage, but common synonyms like 'A/B test' and 'feature flag' are missing, keeping it below anchor 5.

4 / 5

Distinctiveness Conflict Risk

'LaunchDarkly experiment' and 'guarded rollout' define a clear niche with distinct triggers and minimal conflict risk with any other skill, matching anchor 5 exactly.

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

launchdarkly/ai-tooling

Path

skills/aws-devops-agent/experiment-orchestration/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.