Add Skill: local-document-research-with-traceable-citations

Extracted from: https://github.com/0verL1nk/PaperSage.git
Score: 1.0

skills/agent-builder-workflow/SKILL.md

@@ -1,110 +0,0 @@
----
-name: agent-builder-workflow
-version: 1.0.0
-description: Build no-code AI agents that connect to enterprise knowledge sources,
-  perform unified search and deep research, and generate explainable answers with
-  citations
-inputs:
-- Task description and agent objectives (e.g., answer Q&A, research specific topics,
-  generate reports)
-- Knowledge sources (documents, databases, enterprise systems) to connect to
-- Retrieval strategy configuration (graph-based knowledge graph vs. vector search)
-- Output requirements (citation format, response structure, code execution needs)
-steps:
-- 'Step 1: Define agent task and objectives - Specify what the agent should do (e.g.,
-  answer a question, perform deep research on a topic, generate a report with citations)'
-- 'Step 2: Configure knowledge sources - Connect to enterprise documents, databases,
-  or external systems that will serve as the agent''s context'
-- 'Step 3: Build LangGraph chain - Create a workflow chain using LangGraph that combines
-  retrieval (graph/vector) and LLM response generation with citation capabilities'
-- 'Step 4: Execute agent - Run the LangGraph chain to process the task and generate
-  responses with grounded citations'
-- 'Step 5: (Optional) Code execution sandbox - If the agent needs to generate or execute
-  code, deploy it to a safe sandbox environment for verification'
-outputs:
-- Agent execution logs showing retrieval steps and LLM responses
-- Grounded answers with block citations to source documents
-- Generated reports or artifacts (if code execution was performed)
-- Structured task completion status and results
-tags: []
-metadata:
-  source_repo: https://github.com/pipeshub-ai/pipeshub-ai.git
-  extracted_at: ''
-  confidence: 0.95
----
-
-# agent-builder-workflow
-
-Build no-code AI agents that connect to enterprise knowledge sources, perform unified search and deep research, and generate explainable answers with citations
-
-## Setup
-
-**Dependencies:**
-
-```text
-pip install langchain langgraph serper-dev-tool qdrant-client redis fastapi pydantic
-```
-
-**Setup steps:**
-
-1. Install dependencies: pip install langchain langgraph serper-dev-tool qdrant-client redis fastapi
-1. Configure knowledge sources in .env (graph DB connection, vector DB, document paths)
-1. Define agent task in agent_builder.py with objectives and retrieval strategy
-1. Run agent chain: python agent_chain.py --task "research_quantum_computing"
-1. For code execution: add sandbox step to agent_chain.py with code generation and safe execution
-
-## Key Files
-
-- `pipeshub-ai/backend/agent_chain.py - LangGraph chain definition for agent workflows`
-- `pipeshub-ai/backend/retrieval_pipeline.py - Knowledge graph and vector search implementation`
-- `pipeshub-ai/workflows/agent_builder.py - No-code agent creation interface`
-- `pipeshub-ai/workflows/citation_generator.py - Block citation generation from retrieved sources`
-
-## Steps
-
-1. Step 1: Define agent task and objectives - Specify what the agent should do (e.g., answer a question, perform deep research on a topic, generate a report with citations)
-2. Step 2: Configure knowledge sources - Connect to enterprise documents, databases, or external systems that will serve as the agent's context
-3. Step 3: Build LangGraph chain - Create a workflow chain using LangGraph that combines retrieval (graph/vector) and LLM response generation with citation capabilities
-4. Step 4: Execute agent - Run the LangGraph chain to process the task and generate responses with grounded citations
-5. Step 5: (Optional) Code execution sandbox - If the agent needs to generate or execute code, deploy it to a safe sandbox environment for verification
-
-## Implementation Details
-
-```python
-LangGraph chain with retrieval (graph/vector) and LLM response stages
-```
-
-```python
-Knowledge graph construction from enterprise documents
-```
-
-```python
-Citation formatting using block references to source documents
-```
-
-## Inputs
-
-- Task description and agent objectives (e.g., answer Q&A, research specific topics, generate reports)
-- Knowledge sources (documents, databases, enterprise systems) to connect to
-- Retrieval strategy configuration (graph-based knowledge graph vs. vector search)
-- Output requirements (citation format, response structure, code execution needs)
-
-## Outputs
-
-- Agent execution logs showing retrieval steps and LLM responses
-- Grounded answers with block citations to source documents
-- Generated reports or artifacts (if code execution was performed)
-- Structured task completion status and results
-
-## Failure Modes
-
-- Insufficient retrieval results due to poor knowledge graph connectivity or vector embedding quality
-- Permission errors when accessing enterprise knowledge sources
-- LLM context window overflow when generating long explanations with citations
-- Sandbox execution failures for code generation or execution tasks
-- Timeout errors during multi-step agent chain execution
-
-## Source
-
-Extracted from: [https://github.com/pipeshub-ai/pipeshub-ai.git](https://github.com/pipeshub-ai/pipeshub-ai.git)
-Confidence: 0.95

skills/agent-builder-workflow/commands.md

@@ -1,6 +0,0 @@
-# Commands: agent-builder-workflow
-
-## Available Commands
-
-- `/skill agent-builder-workflow` — Load this skill
-- `/run agent-builder-workflow` — Execute workflow

skills/agent-builder-workflow/examples.md

@@ -1,10 +0,0 @@
-# Examples: agent-builder-workflow
-
-## Usage Example
-
-```python
-# How to use this skill
-# Inputs: Task description and agent objectives (e.g., answer Q&A, research specific topics, generate reports), Knowledge sources (documents, databases, enterprise systems) to connect to, Retrieval strategy configuration (graph-based knowledge graph vs. vector search), Output requirements (citation format, response structure, code execution needs)
-# Process: Step 1: Define agent task and objectives - Specify what the agent should do (e.g., answer a question, perform deep research on a topic, generate a report with citations) → Step 2: Configure knowledge sources - Connect to enterprise documents, databases, or external systems that will serve as the agent's context → Step 3: Build LangGraph chain - Create a workflow chain using LangGraph that combines retrieval (graph/vector) and LLM response generation with citation capabilities
-# Outputs: Agent execution logs showing retrieval steps and LLM responses, Grounded answers with block citations to source documents, Generated reports or artifacts (if code execution was performed), Structured task completion status and results
-```

skills/agent-builder-workflow/metadata.json

@@ -1,35 +0,0 @@
-{
-  "name": "agent-builder-workflow",
-  "version": "1.0.0",
-  "goal": "Build no-code AI agents that connect to enterprise knowledge sources, perform unified search and deep research, and generate explainable answers with citations",
-  "inputs": [
-    "Task description and agent objectives (e.g., answer Q&A, research specific topics, generate reports)",
-    "Knowledge sources (documents, databases, enterprise systems) to connect to",
-    "Retrieval strategy configuration (graph-based knowledge graph vs. vector search)",
-    "Output requirements (citation format, response structure, code execution needs)"
-  ],
-  "steps": [
-    "Step 1: Define agent task and objectives - Specify what the agent should do (e.g., answer a question, perform deep research on a topic, generate a report with citations)",
-    "Step 2: Configure knowledge sources - Connect to enterprise documents, databases, or external systems that will serve as the agent's context",
-    "Step 3: Build LangGraph chain - Create a workflow chain using LangGraph that combines retrieval (graph/vector) and LLM response generation with citation capabilities",
-    "Step 4: Execute agent - Run the LangGraph chain to process the task and generate responses with grounded citations",
-    "Step 5: (Optional) Code execution sandbox - If the agent needs to generate or execute code, deploy it to a safe sandbox environment for verification"
-  ],
-  "outputs": [
-    "Agent execution logs showing retrieval steps and LLM responses",
-    "Grounded answers with block citations to source documents",
-    "Generated reports or artifacts (if code execution was performed)",
-    "Structured task completion status and results"
-  ],
-  "failure_modes": [
-    "Insufficient retrieval results due to poor knowledge graph connectivity or vector embedding quality",
-    "Permission errors when accessing enterprise knowledge sources",
-    "LLM context window overflow when generating long explanations with citations",
-    "Sandbox execution failures for code generation or execution tasks",
-    "Timeout errors during multi-step agent chain execution"
-  ],
-  "confidence": 0.95,
-  "explanation": "PipesHub provides a reusable agent builder workflow that combines LangGraph orchestration with graph-based and vector-based retrieval. This pattern can be adapted to any enterprise context where AI agents need to search across multiple knowledge sources, generate explainable answers with citations, and optionally execute code in a safe sandbox. The workflow is defined by specific configuration files (LangGraph chain definitions) and follows a standard pattern: task definition \u2192 knowledge source connection \u2192 retrieval strategy \u2192 response generation \u2192 optional code sandbox.",
-  "source_repo": "https://github.com/pipeshub-ai/pipeshub-ai.git",
-  "score": 1.0
-}

skills/local-document-research-with-traceable-citations/SKILL.md

@@ -0,0 +1,82 @@
+---
+name: local-document-research-with-traceable-citations
+version: 1.0.0
+description: Enable users to import local documents, asynchronously process them into
+  an indexed knowledge base, and obtain AI-generated answers that cite specific page
+  locations and OCR evidence.
+inputs:
+- Local document files (PDF, DOCX, PPTX, XLSX, images, TXT)
+- Configured LLM API endpoint and keys (via .env or settings)
+- Optional web search service config if enabled
+- Local OCR model cache (downloaded on first use)
+steps:
+- Import documents into a project; files are queued for asynchronous processing.
+- Convert non-PDF formats (DOCX, PPTX, XLSX) to PDF using native Office or LibreOffice
+  fallback.
+- Run OCR (PaddleOCR) on PDF pages/images to extract text, page numbers, polygons,
+  and confidence scores.
+- Chunk text and generate embeddings; publish to LanceDB hybrid index (dense vector
+  + full-text) only when fully processed.
+- User starts a research session or branch; Leader agent analyzes query.
+- Hybrid RAG retrieves candidate chunks from ready documents; dynamic material scope
+  ensures no half-indexed docs.
+- Leader delegates tasks to sub-agents (researcher, reviewer, writer) via constrained
+  task capability; each delegation logs start, completion, duration, and evidence.
+- Generate answer that references only actually used evidence; citations include document
+  ID, page, coordinates.
+- User clicks citation to open original document and view highlighted OCR location.
+outputs:
+- Project with indexed document library (SQLite metadata + LanceDB vectors)
+- AI answers with verifiable citations to source pages
+- Evidence preview with page image and OCR highlight polygons
+- Persistent session history, branches, and long-term memory
+tags: []
+metadata:
+  source_repo: https://github.com/0verL1nk/PaperSage.git
+  extracted_at: ''
+  confidence: 0.85
+---
+
+# local-document-research-with-traceable-citations
+
+Enable users to import local documents, asynchronously process them into an indexed knowledge base, and obtain AI-generated answers that cite specific page locations and OCR evidence.
+
+## Steps
+
+1. Import documents into a project; files are queued for asynchronous processing.
+2. Convert non-PDF formats (DOCX, PPTX, XLSX) to PDF using native Office or LibreOffice fallback.
+3. Run OCR (PaddleOCR) on PDF pages/images to extract text, page numbers, polygons, and confidence scores.
+4. Chunk text and generate embeddings; publish to LanceDB hybrid index (dense vector + full-text) only when fully processed.
+5. User starts a research session or branch; Leader agent analyzes query.
+6. Hybrid RAG retrieves candidate chunks from ready documents; dynamic material scope ensures no half-indexed docs.
+7. Leader delegates tasks to sub-agents (researcher, reviewer, writer) via constrained task capability; each delegation logs start, completion, duration, and evidence.
+8. Generate answer that references only actually used evidence; citations include document ID, page, coordinates.
+9. User clicks citation to open original document and view highlighted OCR location.
+
+## Inputs
+
+- Local document files (PDF, DOCX, PPTX, XLSX, images, TXT)
+- Configured LLM API endpoint and keys (via .env or settings)
+- Optional web search service config if enabled
+- Local OCR model cache (downloaded on first use)
+
+## Outputs
+
+- Project with indexed document library (SQLite metadata + LanceDB vectors)
+- AI answers with verifiable citations to source pages
+- Evidence preview with page image and OCR highlight polygons
+- Persistent session history, branches, and long-term memory
+
+## Failure Modes
+
+- OCR quality low for scanned images leading to poor extraction
+- Missing Office/LibreOffice causes conversion failure for Office docs
+- Interrupted indexing leaves documents unpublished and excluded from retrieval
+- LLM API outage or misconfiguration yields no answer
+- Citation coordinates mismatch due to chunk drift
+- Sub-agent recursion if constraints not enforced
+
+## Source
+
+Extracted from: [https://github.com/0verL1nk/PaperSage.git](https://github.com/0verL1nk/PaperSage.git)
+Confidence: 0.85

skills/local-document-research-with-traceable-citations/commands.md

@@ -0,0 +1,6 @@
+# Commands: local-document-research-with-traceable-citations
+
+## Available Commands
+
+- `/skill local-document-research-with-traceable-citations` — Load this skill
+- `/run local-document-research-with-traceable-citations` — Execute workflow

skills/local-document-research-with-traceable-citations/examples.md

@@ -0,0 +1,10 @@
+# Examples: local-document-research-with-traceable-citations
+
+## Usage Example
+
+```python
+# How to use this skill
+# Inputs: Local document files (PDF, DOCX, PPTX, XLSX, images, TXT), Configured LLM API endpoint and keys (via .env or settings), Optional web search service config if enabled, Local OCR model cache (downloaded on first use)
+# Process: Import documents into a project; files are queued for asynchronous processing. → Convert non-PDF formats (DOCX, PPTX, XLSX) to PDF using native Office or LibreOffice fallback. → Run OCR (PaddleOCR) on PDF pages/images to extract text, page numbers, polygons, and confidence scores.
+# Outputs: Project with indexed document library (SQLite metadata + LanceDB vectors), AI answers with verifiable citations to source pages, Evidence preview with page image and OCR highlight polygons, Persistent session history, branches, and long-term memory
+```

skills/local-document-research-with-traceable-citations/metadata.json

@@ -0,0 +1,40 @@
+{
+  "name": "local-document-research-with-traceable-citations",
+  "version": "1.0.0",
+  "goal": "Enable users to import local documents, asynchronously process them into an indexed knowledge base, and obtain AI-generated answers that cite specific page locations and OCR evidence.",
+  "inputs": [
+    "Local document files (PDF, DOCX, PPTX, XLSX, images, TXT)",
+    "Configured LLM API endpoint and keys (via .env or settings)",
+    "Optional web search service config if enabled",
+    "Local OCR model cache (downloaded on first use)"
+  ],
+  "steps": [
+    "Import documents into a project; files are queued for asynchronous processing.",
+    "Convert non-PDF formats (DOCX, PPTX, XLSX) to PDF using native Office or LibreOffice fallback.",
+    "Run OCR (PaddleOCR) on PDF pages/images to extract text, page numbers, polygons, and confidence scores.",
+    "Chunk text and generate embeddings; publish to LanceDB hybrid index (dense vector + full-text) only when fully processed.",
+    "User starts a research session or branch; Leader agent analyzes query.",
+    "Hybrid RAG retrieves candidate chunks from ready documents; dynamic material scope ensures no half-indexed docs.",
+    "Leader delegates tasks to sub-agents (researcher, reviewer, writer) via constrained task capability; each delegation logs start, completion, duration, and evidence.",
+    "Generate answer that references only actually used evidence; citations include document ID, page, coordinates.",
+    "User clicks citation to open original document and view highlighted OCR location."
+  ],
+  "outputs": [
+    "Project with indexed document library (SQLite metadata + LanceDB vectors)",
+    "AI answers with verifiable citations to source pages",
+    "Evidence preview with page image and OCR highlight polygons",
+    "Persistent session history, branches, and long-term memory"
+  ],
+  "failure_modes": [
+    "OCR quality low for scanned images leading to poor extraction",
+    "Missing Office/LibreOffice causes conversion failure for Office docs",
+    "Interrupted indexing leaves documents unpublished and excluded from retrieval",
+    "LLM API outage or misconfiguration yields no answer",
+    "Citation coordinates mismatch due to chunk drift",
+    "Sub-agent recursion if constraints not enforced"
+  ],
+  "confidence": 0.85,
+  "explanation": "The README outlines a clear pipeline from document import to cited answer with evidence location, which is a reusable pattern for local-first RAG applications requiring traceability.",
+  "source_repo": "https://github.com/0verL1nk/PaperSage.git",
+  "score": 1.0
+}

skills/local-document-research-with-traceable-citations/tests.md

@@ -1,4 +1,4 @@
-# Tests: agent-builder-workflow
+# Tests: local-document-research-with-traceable-citations
 
 ## Test Checklist