Retrieval Traceability

March 6, 2026 · View on GitHub

🧭 Quick Return to Map

You are in a sub-page of MemoryLongContext.
To reorient, go back here:

MemoryLongContext — extended context windows and memory retention

WFGY Global Fix Map — main Emergency Room, 300+ structured fixes

WFGY Problem Map 1.0 — 16 reproducible failure modes

Think of this page as a desk within a ward.
If you need the full triage and all prescriptions, return to the Emergency Room lobby.

Citations that look right can still hide silent drift.
This guardrail defines how to enforce traceability schemas so that every claim links back to a stable, reproducible snippet.

When to use

Answers cite a source but the snippet cannot be located.
Two runs over the same corpus produce different citations.
A fact is quoted but not aligned to any section anchor.
Long-context threads degrade and snippets blur into paraphrase.
Multi-agent systems pass partial context and lose attribution.

Root causes

Orphan citations: snippet ID missing or fabricated.
Boundary drift: citation spans cross section joins.
Silent truncation: tokens dropped at cut points.
Cache overwrite: citation schema lost after session reload.
Free-text cites: URLs or titles given without offsets.

Core acceptance targets

Each claim must include snippet_id, section_id, start_line, end_line, source_url.
ΔS(question, retrieved) ≤ 0.45 overall.
Joins between snippets ≤ 0.50 ΔS.
λ convergent across 3 paraphrases.
Audit trail reproducible from log alone.

Structural fixes

Snippet table schema
Require {snippet_id | section_id | start_line | end_line | citation}.
Fence joins
Split at section boundaries. Reject cross-section reuse.
Trace log
Store {ΔS, λ_state, mem_rev, mem_hash} per step.
Contract lock
Apply Data Contracts for payload validation.

Fix in 60 seconds

Enforce snippet table with unique IDs and line ranges.
Verify ΔS across each join ≤ 0.50.
Echo λ states at retrieval, assembly, reasoning.
Reject orphan claims (no snippet_id).
Log trail so same inputs → same citations.

Copy-paste prompt


You have TXT OS and the WFGY Problem Map.

Goal: Ensure every claim links to a reproducible snippet.

Protocol:

1. Build a Snippet Table {snippet\_id, section\_id, start\_line, end\_line, citation}.
2. Require cite-then-answer.
3. Forbid cross-section reuse.
4. If a claim has no snippet\_id, stop and request citation.
5. Report ΔS(question,retrieved), joins ΔS, and λ states.
6. Store {mem\_rev, mem\_hash, task\_id} for audit trail.
7. Answer only with snippets present in the table.

Common failure signals

Citations alternate across runs → missing trace schema.
URL without offsets → orphan citation.
Facts cited but no snippet_id → schema lock failed.
Session reload erases citations → ghost cache in memory.

🔗 Quick-Start Downloads (60 sec)

Tool	Link	3-Step Setup
WFGY 1.0 PDF	Engine Paper	1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>”
TXT OS (plain-text OS)	TXTOS.txt	1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly

Explore More

Layer	Page	What it’s for
⭐ Proof	WFGY Recognition Map	External citations, integrations, and ecosystem proof
⚙️ Engine	WFGY 1.0	Original PDF tension engine and early logic sketch (legacy reference)
⚙️ Engine	WFGY 2.0	Production tension kernel for RAG and agent systems
⚙️ Engine	WFGY 3.0	TXT based Singularity tension engine (131 S class set)
🗺️ Map	Problem Map 1.0	Flagship 16 problem RAG failure taxonomy and fix map
🗺️ Map	Problem Map 2.0	Global Debug Card for RAG and agent pipeline diagnosis
🗺️ Map	Problem Map 3.0	Global AI troubleshooting atlas and failure pattern map
🧰 App	TXT OS	.txt semantic OS with fast bootstrap
🧰 App	Blah Blah Blah	Abstract and paradox Q&A built on TXT OS
🧰 App	Blur Blur Blur	Text to image generation with semantic control
🏡 Onboarding	Starter Village	Guided entry point for new users

If this repository helped, starring it improves discovery so more builders can find the docs and tools.

要直接開始 data-contracts.md 嗎？