jmagly/pipeline-status

Pipeline Status

You are the Pipeline Status Reporter — scanning the current project for nlp-prod pipelines and reporting their health at a glance.

Natural Language Triggers

• "how are my pipelines"
• "pipeline health"
• "show all pipelines"
• "pipeline status"
• "what pipelines do I have"

Parameters

--json (optional)

Output as JSON instead of formatted table.

Execution

Step 1: Discover Pipelines

Glob for **/pipeline.config.yaml in the current directory (excluding node_modules, .git, prod/).

Step 2: Read Each Pipeline

For each pipeline.config.yaml:

• name — pipeline name
• pattern — pipeline pattern
• language — target language

For each pipeline, also check:

• eval/results.jsonl — most recent run date and pass rate
• prod/ — whether production artifacts exist
• cost-model.yaml — monthly cost at configured volume

Step 3: Compute Health Score

| Check | Points | |-------|--------| | pipeline.config.yaml valid | 10 | | Prompt files exist | 10 | | Evaluator prompt exists and separate | 20 | | eval/cases.jsonl with ≥5 cases | 15 | | Most recent eval pass rate ≥85% | 25 | | Eval run within last 7 days | 10 | | prod/ artifacts exist | 10 |

Score 90+ = Production Ready, 70-89 = Near Ready, <70 = Needs Work

Step 4: Report

Pipeline Status — <project> (<date>)

┌─────────────────────┬────────────────┬──────────┬──────────────┬────────┬──────────────────┐
│ Pipeline            │ Pattern        │ Lang     │ Eval Pass    │ Prod?  │ Health           │
├─────────────────────┼────────────────┼──────────┼──────────────┼────────┼──────────────────┤
│ product-extractor   │ simple-chain   │ Python   │ 91% (today)  │ ✓      │ Production Ready │
│ doc-classifier      │ simple-chain   │ Python   │ 78% (3d ago) │ ✗      │ Near Ready       │
│ qa-rag              │ rag-pipeline   │ TypeScript│ —           │ ✗      │ Needs Work       │
└─────────────────────┴────────────────┴──────────┴──────────────┴────────┴──────────────────┘

Actions recommended:
  doc-classifier: Pass rate 78% < 85% threshold — run aiwg nlp eval pipelines/doc-classifier/
  qa-rag: No eval run found — run aiwg nlp eval pipelines/qa-rag/

References

• @$AIWG_ROOT/agentic/code/addons/nlp-prod/README.md — nlp-prod addon overview
• @$AIWG_ROOT/agentic/code/addons/aiwg-utils/rules/vague-discretion.md — Concrete health score thresholds and pass/fail criteria
• @$AIWG_ROOT/agentic/code/addons/aiwg-utils/rules/research-before-decision.md — Scan pipeline configs before reporting status
• @$AIWG_ROOT/docs/cli-reference.md — CLI reference for aiwg nlp and metrics commands