Full Pipeline Chain Tracing

Trace claims backward through pipeline stages to primary evidence sources.

Updated Apr 11, 2026
One-click install
npx skills add https://github.com/gozonerd/shadow-ai-assessment --skill full-pipeline-chain-tracing-gozonerd
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Full Pipeline Chain Tracing
Source: https://github.com/gozonerd/shadow-ai-assessment/tree/main/.claude/skills/dmis/task_type/TT-12_Full_Pipeline_Chain_Tracing
Command: npx skills add https://github.com/gozonerd/shadow-ai-assessment --skill full-pipeline-chain-tracing-gozonerd

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyyaml, and includes scripts (resource) and references (resource) components.

What problem does it solve? Final integrated outputs in multi-stage analysis pipelines often contain claims whose origins are unclear, making it impossible to verify that every assertion is grounded in primary source material. This Skill documents the complete provenance chain for each claim so reviewers can audit how evidence flowed through every processing stage. ## Core Features & Use Cases - Backward Provenance Tracing: Maps each finalized claim from P4 integration back through P3 synthesis, P2 analysis, and P1 raw processing to original source documents. - Convergence Validation: Verifies that QA and methodology convergence scores meet required thresholds (α ≥ 0.70 / 0.75) at every stage transition. - Gap Detection and Documentation: Flags broken or incomplete trace chains with severity ratings and remediation guidance instead of silently skipping them. - Use Case: During final review of a DATS assessment report, trace the claim that CommCare was deployed in StrongMinds Zambia back through all pipeline stages to confirm it rests on documented operational records and interviews. ## Quick Start Trace every primary claim in the integrated P4 output back to its primary sources and produce a validated provenance report conforming to the TT-12 output schema.

Frequently Asked Questions about Full Pipeline Chain Tracing

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I trace a claim back to its source documents in a multi-stage pipeline?

Start from the finalized claim in the P4 integrated output, then map where it appeared in P3 synthesis and P2 analysis, and finally link it to P1 raw threads and original source identifiers. Document thread counts and convergence scores at each stage transition.

How do I validate a provenance trace output against the TT-12 schema?

Run the validate_output.py script with your output YAML file as the argument. It checks required fields like provenance_traces, trace_chain stages, primary_sources, and trace_status, and reports specific validation errors.

What convergence thresholds must be met during chain tracing?

QA convergence must reach α ≥ 0.70 and methodology convergence must reach α ≥ 0.75 at each decision point. Traces showing scores below these thresholds are flagged in the convergence_validation section of the output.

What happens when a claim cannot be fully traced to primary sources?

The trace is marked as Partial or Broken in the trace_status field, and each gap is documented with a gap_id, affected stage, severity, and remediation guidance. Gaps are never silently omitted from the provenance report.

What are the limitations of backward pipeline chain tracing?

Tracing depends on complete audit trail documentation from earlier P1, P2, and P3 stages; missing intermediate records produce Partial traces. It also cannot verify claims whose source material was never assigned reference identifiers.