| name | release-debug |
| description | Diagnose deployment and release failures using Raptor CLI. Use when user mentions deployment failed, release error, release debugging, deployment logs, or terraform errors in releases. |
| allowed-tools | Bash, Read, Glob, Grep |
Release Debugger
Diagnose deployment failures and provide actionable solutions using Raptor CLI.
Platform Context
- Developer self-service platform using JSON configuration validated against schemas
- Generates Terraform from validated configurations
- All changes go through managed releases (no direct infrastructure access)
- Use "trigger release" not "run terraform", "environment" not "cluster"
Release Types
- Full/Selective Release: Available to regular users (with allow_destroy/refresh options)
- Custom Release: Operations team only (for Terraform commands)
- Selective releases are NOT
terraform -target - Facets handles dependencies safely
Available Raptor Commands
Fetch logs:
raptor logs release --project PROJECT --environment ENV RELEASE_ID
raptor logs release --project PROJECT --environment ENV -f RELEASE_ID
Get releases:
raptor get releases --project PROJECT --environment ENV -o json
Inspect resource configuration:
raptor get resources --project PROJECT TYPE/NAME -o json
raptor get resource-overrides --project PROJECT --environment ENV TYPE/NAME -o json
raptor get resource-status --project PROJECT --environment ENV TYPE/NAME
raptor get resource-outputs --project PROJECT --environment ENV TYPE/NAME -o json
Validate against schema:
raptor get resource-type-schema TYPE/FLAVOR/VERSION
raptor get resource-type-inputs TYPE/FLAVOR/VERSION
raptor get resource-type-outputs TYPE/FLAVOR/VERSION
Check valid expressions:
raptor get resource-output-expressions --project PROJECT
Download module code (for debugging):
raptor get iac-module TYPE/FLAVOR/VERSION
raptor get iac-module TYPE/FLAVOR/VERSION -o ./modules/
Subagent Delegation
This skill delegates to two subagent skills:
/raptor — For ALL raptor CLI operations (fetching logs, getting resources, schemas, modules). Invoke /raptor <task> to delegate raptor commands.
/analyze-logs — For log analysis after fetching logs. Invoke /analyze-logs <log file path> to get structured error diagnosis.
Investigation Workflow
1. Extract Error Evidence
Delegate to /raptor to fetch deployment logs:
/raptor fetch release logs for project PROJECT environment ENV release RELEASE_ID and save to /tmp/deployment_logs.txt
Then delegate to /analyze-logs for structured analysis:
/analyze-logs /tmp/deployment_logs.txt
The log analyzer returns: critical failure, error classification, root cause chain, and evidence.
2. Analyze Resource Configuration
Delegate to /raptor to retrieve configuration:
/raptor get resource config and overrides for TYPE/NAME in project PROJECT environment ENV
Check:
- What values were provided?
- Missing or incorrect fields?
- Configuration vs operational problem?
3. Validate Against Schema
Delegate to /raptor to get schema:
/raptor get resource-type-schema for TYPE/FLAVOR/VERSION
Confirm:
- Field exists in schema
- Data type matches
- Required fields present
- Never guess configuration options
4. Download Module Code (When Needed)
Delegate to /raptor for module download:
/raptor download iac-module TYPE/FLAVOR/VERSION to ./debug-modules/
Then analyze with file tools (read, grep, glob):
- Check outputs.tf for valid attribute references
- Examine main.tf for resource definitions
- Review variables.tf for expected input structure
5. Classify Root Cause
| Error Type | Solution | Example |
|---|
| Configuration error | Schema-validated JSON changes | Wrong field type, missing required field |
| Operational error | Infrastructure actions | Timeout, resource unavailable |
| State issue | Release options or custom release | State lock, state drift |
| Module bug | Module fix via registry workflow | Bad attribute reference, missing resource |
Common Patterns
State locks:
Error acquiring the state lock
-> Go to Releases -> ellipsis menu -> Unlock State (ONLY for this specific error)
Helm conflicts:
Error: cannot re-use a name that is still in use
-> Delete conflicting release or use --with-refresh option
Attribute errors:
has no attribute "X"
-> Download module, check outputs.tf for valid attributes
Resource timeouts:
-> Fix the underlying resource issue, not the Helm timeout config
Expression errors:
invalid reference: ${type.name.out.field}
-> Check valid expressions:
raptor get resource-output-expressions --project PROJECT
Response Format
Structure your diagnosis clearly:
Deployment Status: [FAILED/SUCCEEDED/IN_PROGRESS]
Issue Summary
What failed and the specific error.
Root Cause
Whether this is configuration, operational, state, or module issue.
Solution
Actionable steps:
- Configuration Changes: Schema-validated JSON with navigation path
- Infrastructure Actions: Release options or operational commands
- Module Fixes: Changes needed via module registry workflow
Always provide evidence for conclusions. Be direct and actionable.
Debugging Checklist
- Get the release ID and fetch logs
- Identify the failing resource(s)
- Parse the error message type
- Get current resource config + overrides
- Validate against schema
- Download module if attribute/code issue
- Classify root cause
- Provide evidence-based fix