Name: migrate
Rating: 73.6 (1 reviews)
Author: launchdarkly

migrate

Migrate an application with hardcoded LLM prompts to a full LaunchDarkly AgentControl implementation in five stages: audit the code, wrap the call, move the tools, add tracking, attach evaluators. Use when the user wants to externalize model/prompt configuration, move from direct provider calls (OpenAI, Anthropic, Bedrock, Gemini, Strands) to a managed config, or stage a full hardcoded-to-LaunchDarkly migration.

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Medium

Suggest reviewing before use

Quality

Content

85%Weight 40%Scale 1-3

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

A thorough, highly actionable migration skill with strong workflow sequencing, explicit validation checkpoints, and well-structured one-level-deep reference disclosure. The main weakness is redundancy: several critical warnings are restated verbatim across the intro callout, the relevant stage, and the 'What NOT to Do' section, which inflates token cost without adding clarity.

Suggestions

Consolidate the three repeated warnings (`load_chat_model` deletion, `/configs-targeting` fallthrough flip, tracker scope) into one canonical location — keep the intro callout as a pointer and the detailed rule in its stage or 'What NOT to Do', not all three verbatim.

The Edge Cases table duplicates several Stage-3/Stage-4 rules already covered in prose (Strands TS, custom StateGraph TOOLS list, multi-agent graph); consider folding table-only edge cases into their stage and keeping the table for genuinely orthogonal cases.

Trim explanatory restatements of the hand-off model — the 'do not auto-invoke siblings' instruction is repeated at nearly every stage; state it once in Core Principles and reference it.

Dimension	Reasoning	Score
Conciseness	The body is concrete and mostly earns its tokens, but several high-stakes warnings are repeated three times — the `load_chat_model` deletion warning appears in the intro callout, Stage 2 sub-step 1, and 'What NOT to Do'; the `/configs-targeting` fallthrough warning and the tracker-scope rule likewise recur — which is padding despite accuracy. Not score-3 lean; not score-1 padded with concepts Claude already knows.	2 / 3
Actionability	Provides fully executable Python and Node code blocks with exact package names and pinned versions, before/after paired snippets, and specific API method calls — copy-paste ready for the worked shapes, matching the score-3 anchor.	3 / 3
Workflow Clarity	Five stages are explicitly sequenced with a hard STOP checkpoint after Stage 1 requiring one of four explicit confirmation forms (confirm/add/fix/stop), and each stage ends with a verify sub-step and feedback loop, matching the score-3 anchor for clear sequence with explicit validation and error-recovery.	3 / 3
Progressive Disclosure	The body is an overview that points to six real, one-level-deep reference files (all verified present in ./references/) with a dedicated References section and a coverage table mapping each shape to its reference, matching the score-3 anchor for well-signaled one-level-deep navigation. The in-body length is justified by worked examples that belong in the main file.	3 / 3
	Total	11 / 12 Passed

Description

100%Weight 40%Scale 1-3

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, complete, and uses third-person imperative voice with an explicit 'Use when' trigger clause covering the realistic ways a user would voice this need. It occupies a clear, low-conflict niche. Voice is third person ('Migrate an application...') with no first/second-person penalty.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions across the five named stages — 'audit the code, wrap the call, move the tools, add tracking, attach evaluators' — matching the score-3 anchor that lists several specific concrete actions.	3 / 3
Completeness	Clearly answers both 'what' (migrate hardcoded LLM prompts to a full LaunchDarkly AgentControl implementation in five stages) and 'when' via an explicit 'Use when...' clause with three distinct triggers, matching the score-3 anchor.	3 / 3
Trigger Term Quality	Natural trigger phrases a user would say are present — 'externalize model/prompt configuration', 'move from direct provider calls (OpenAI, Anthropic, Bedrock, Gemini, Strands) to a managed config', 'stage a full hardcoded-to-LaunchDarkly migration' — giving good coverage of how the need is actually voiced.	3 / 3
Distinctiveness Conflict Risk	The niche is tightly scoped to a LaunchDarkly AgentControl migration with named providers and a five-stage process, making it unlikely to trigger for unrelated skills.	3 / 3
	Total	12 / 12 Passed

Validation

87%

Warnings & errors only

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

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (573 lines); consider splitting into references/ and linking	Warning
relative_links	Relative link issues: 10 suspicious	Warning

	Total	14 / 16 Passed

Repository: launchdarkly/ai-tooling
Path: skills/agentcontrol/migrate/SKILL.md
Commit: add614f

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

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.