| Alert Definition (AD) |
Folder-level grouping for one published fault-intelligence set. ID format: AD######. |
| Fault Signature (FS) |
YAML artifact that detects a fault from events, telemetry, or alarms and extracts variables. ID format: FS######. |
| Remediation Guide (RG) |
Markdown-only human-readable troubleshooting and remediation source document. ID format: RG######. |
| Repair Action Workflow (RAW) |
YAML artifact that encodes validation, action selection, repair actions, and terminal outcomes. ID format: RAW######. |
| Linked set |
An AD folder containing aligned FS, RG, and RAW artifacts with the same six-digit suffix. |
| Artifact group |
Same as linked set: a folder under intelligence-artifacts/ for one alert definition. |
alert_vars |
Variables passed from a matched FS or alert payload into the RAW. |
eval_cli |
RAW validation action that runs CLI and evaluates output with a pattern. |
config_cli |
RAW repair action that applies persistent configuration after approval. |
resolve |
RAW terminal action for successful repair. |
escalate |
RAW terminal action for permanent handoff to support, TAC, or RMA. |
fail |
RAW terminal action for workflow execution failure. |
| Collection List (CL) |
Future Health Intelligence YAML artifact defining what diagnostic data to collect. |
| Parser (PARSE) |
Future Health Intelligence YAML artifact defining how to parse collected data. |
| Health Check Rule (HCR) |
Future Health Intelligence YAML artifact defining health evaluation logic. |
| IA |
Intelligence Artifact. |
| FMS |
Fault Management System. |
| MCP |
Model Context Protocol. |
| Headed mode |
Interactive OpenCode TUI run where the operator pastes an alert prompt directly. |
| Headless mode |
OpenCode server plus FastAPI relay receiving alerts over HTTP. |
| Strict mode |
Agent execution mode that follows the RAW as the controlling workflow. |
| Hybrid-reasoning mode |
Agent execution mode that allows additional LLM judgment around ambiguous workflow choices. |
| Test bundle |
YAML test file that exercises RAW behavior without LLM, RADKit, Webex, or Splunk. |
| Session log |
Markdown record written under logs/troubleshooting/ for post-run review. |