aiconfig-update
Scannednpx machina-cli add skill launchdarkly/agent-skills/aiconfig-update --openclawAI Config Update & Lifecycle
You're using a skill that will guide you through updating, archiving, and deleting AI Configs and their variations. Your job is to understand the current state of the config, make the changes, and verify the result.
Prerequisites
- Existing AI Config to modify
- LaunchDarkly API access token or MCP server
Core Principles
- Fetch Before Changing: Always check the current state before modifying
- Verify After Changing: Fetch the config again to confirm updates were applied
- Archive Before Deleting: Archival is reversible; deletion is not
API Key Detection
- Check environment variables —
LAUNCHDARKLY_API_KEY,LAUNCHDARKLY_API_TOKEN,LD_API_KEY - Check MCP config — If applicable
- Prompt user — Only if detection fails
Workflow
Step 1: Understand Current State
Fetch the config to see what exists before changing anything:
GET /projects/{projectKey}/ai-configs/{configKey}
Step 2: Make the Update
Follow API Quick Start:
- Update instructions/messages — PATCH variation
- Switch model — PATCH variation with modelConfigKey and model
- Tune parameters — PATCH variation with model.parameters
- Archive config — PATCH config with
{"archived": true} - Delete — DELETE config or variation (irreversible)
Step 4: Verify
-
Fetch updated config:
GET /projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey} -
Confirm the response shows your updated values
-
Report results:
- ✓ Update applied successfully
- ✓ Config reflects changes
- ⚠️ Flag any issues or rollback if needed
What NOT to Do
- Don't update production directly without testing
- Don't change multiple things at once
- Don't skip verification
- Don't delete without user confirmation
Related Skills
aiconfig-variations— Create variations to test changesaiconfig-tools— Update tools
References
Source
git clone https://github.com/launchdarkly/agent-skills/blob/main/skills/ai-configs/aiconfig-update/SKILL.mdView on GitHub Overview
This skill guides you through updating LaunchDarkly AI Configs and their variations. It supports changing config properties, model parameters, or messages, and archiving unused configs. Deletion is irreversible, so it should be used with caution.
How This Skill Works
The skill starts by fetching the current AI Config state, then applies changes via PATCH or DELETE requests as needed, and finally verifies the result by re-fetching the updated config or variation. It includes API key detection and follows a defined lifecycle: fetch, update, archive if needed, delete only when appropriate, and verify the outcome.
When to Use It
- Modify config properties or instructions/messages in a variation
- Switch the underlying model for a variation
- Tune model parameters within a variation
- Archive an AI Config or variation that is no longer in use
- Permanently delete a config or variation when it is retired
Quick Start
- Step 1: Fetch current state using GET /projects/{projectKey}/ai-configs/{configKey}
- Step 2: Apply changes with PATCH (update instructions/messages, switch model, or tune parameters) or DELETE for removal, and PATCH with {"archived": true} to archive
- Step 3: Verify by re-fetching the updated config/variation and confirming the changes
Best Practices
- Fetch the current state before making any changes
- Verify updates by fetching the config/variation again
- Archive before deleting; archival is reversible while deletion is not
- Avoid changing multiple items at once to reduce risk
- Confirm with the user before performing irreversible deletion
Example Use Cases
- Update the instructions for a variation to reflect new guidance for users
- Switch a variation to a newer model configuration and validate performance
- Tune parameters like temperature or max tokens in a variation to improve outputs
- Archive an AI Config that is deprecated but may need future restoration
- Permanently delete a variation after confirming it is no longer required
