algomath-extract 1.0.0 → 1.0.2

package/commands/algo-extract.md

@@ -0,0 +1,99 @@
+---
+description: "Extract algorithm from PDF or text file"
+argument-hint: "[file-path] [--auto] [--step] [--name <name>]"
+tools:
+  read: true
+  write: true
+---
+
+<objective>
+Extract algorithm from PDF or text file using LLM-powered parsing.
+Supports automatic extraction or step-by-step with review points.
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/src/extraction/pdf_processor.py
+@/home/milgraph/Projects/algo_framework/src/extraction/llm_extractor.py
+</execution_context>
+
+<process>
+Execute extraction workflow:
+
+1. **Parse Arguments**
+   - Check if file path provided
+   - Determine mode: auto, step-by-step, or interactive
+
+2. **Extract Text from File**
+   - Auto-detect file type (PDF, .txt, .md)
+   - Use pdfplumber for PDFs
+   - Handle encoding issues
+   - Validate extraction success
+
+3. **Parse Algorithm Structure**
+   - Use opencode LLM to parse text
+   - Identify algorithm name
+   - Extract inputs and outputs
+   - Parse step-by-step procedure
+   - Handle mathematical notation
+
+4. **Review Point (if step-by-step)**
+   - Display extracted text
+   - Ask user to confirm or edit
+   - Apply user corrections
+
+5. **Generate Structured Steps**
+   - Convert parsed text to Algorithm object
+   - Validate step structure
+   - Check for completeness
+
+6. **Review Point (if step-by-step)**
+   - Display structured steps
+   - Show inputs, outputs, step list
+   - Allow editing
+
+7. **Save Algorithm**
+   - Create directory: .algomath/algorithms/{name}/
+   - Save source.txt (original text)
+   - Save steps.json (structured)
+   - Save metadata.json (info)
+
+8. **Display Summary**
+   - Show algorithm name
+   - Show step count
+   - Display next steps
+   - Suggest: /algo-generate
+</process>
+
+<examples>
+
+**Extract from PDF (interactive):**
+/algo-extract research_paper.pdf
+
+**Auto-extract (no prompts):**
+/algo-extract --auto research_paper.pdf
+
+**Extract from text file:**
+/algo-extract algorithm.txt
+
+**Specify algorithm name:**
+/algo-extract research.pdf --name "Dijkstra Shortest Path"
+
+</examples>
+
+<options>
+
+--auto
+Skip review points, extract automatically
+
+--step
+Enable step-by-step mode with review at each stage
+
+--name <name>
+Specify algorithm name (optional)
+
+</options>
+
+<next_steps>
+/algo-generate - Generate Python code from extracted steps
+/algo-status - Check current algorithm state
+</next_steps>

package/commands/algo-generate.md

@@ -0,0 +1,97 @@
+---
+description: "Generate Python code from extracted algorithm steps"
+argument-hint: "[--review] [--template-only] [--llm-only]"
+tools:
+  read: true
+  write: true
+---
+
+<objective>
+Generate executable Python code from structured algorithm steps.
+Supports template-based generation (fast, reliable) or LLM-enhanced (for complex expressions).
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/src/generation/code_generator.py
+@/home/milgraph/Projects/algo_framework/src/generation/llm_generator.py
+</execution_context>
+
+<process>
+Execute code generation workflow:
+
+1. **Load Algorithm**
+   - Read steps.json from current algorithm
+   - Validate step structure
+   - Check for completeness
+
+2. **Select Generation Mode**
+   - Default: Hybrid (template + LLM for complex parts)
+   - --template-only: Rule-based only (faster)
+   - --llm-only: Full LLM generation (slower, more accurate)
+
+3. **Generate Function Signature**
+   - Extract inputs and outputs
+   - Add type hints
+   - Generate docstring
+
+4. **Generate Step Code**
+   - Convert each step to Python
+   - Handle loops, conditionals, assignments
+   - Preserve variable names
+   - Add comments explaining logic
+
+5. **Review Point (if --review)**
+   - Display generated code
+   - Syntax highlighting
+   - Ask for approval
+   - Allow editing
+
+6. **Validate Code**
+   - Syntax check with Python parser
+   - Check for common errors
+   - Ensure all variables defined
+
+7. **Save Code**
+   - Write to .algomath/algorithms/{name}/generated.py
+   - Update metadata.json
+
+8. **Display Summary**
+   - Show generated file path
+   - Line count
+   - Next steps: /algo-run
+</process>
+
+<examples>
+
+**Generate with default hybrid mode:**
+/algo-generate
+
+**Generate with review:**
+/algo-generate --review
+
+**Template-based only (faster):**
+/algo-generate --template-only
+
+**LLM-enhanced only (better for complex):**
+/algo-generate --llm-only
+
+</examples>
+
+<options>
+
+--review
+Show generated code for review before saving
+
+--template-only
+Use rule-based generation only (no LLM)
+
+--llm-only
+Use LLM for all generation (slower but more accurate)
+
+</options>
+
+<next_steps>
+/algo-run - Execute the generated code
+/algo-verify - Verify execution results
+/algo-status - Check algorithm state
+</next_steps>

package/commands/algo-help.md

@@ -0,0 +1,62 @@
+---
+description: "Show help and available commands"
+argument-hint: "[command]"
+tools:
+  read: true
+  write: false
+---
+
+<objective>
+Display help information, available commands, and usage examples.
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/README.md
+</execution_context>
+
+<process>
+Display help:
+
+1. **Show Welcome Banner**
+   - AlgoMath logo/name
+   - Version
+   - Brief description
+
+2. **List Commands**
+   - /algo-extract - Extract from PDF/text
+   - /algo-generate - Generate Python code
+   - /algo-run - Execute code
+   - /algo-verify - Verify results
+   - /algo-status - Check current state
+   - /algo-list - List saved algorithms
+   - /algo-help - This help
+
+3. **Show Workflow**
+   ```
+   PDF/Text → Extract → Generate → Run → Verify
+   ```
+
+4. **Show Examples**
+   - Basic usage
+   - Common flags
+   - Tips
+
+5. **Show Documentation Links**
+   - User guide
+   - Examples
+   - Troubleshooting
+</process>
+
+<examples>
+
+**Show general help:**
+/algo-help
+
+**Show help for specific command:**
+/algo-help extract
+
+</examples>
+
+<next_steps>
+/algo-extract - Start with a new algorithm
+</next_steps>

package/commands/algo-list.md

@@ -0,0 +1,60 @@
+---
+description: "List all saved algorithms"
+argument-hint: "[--detailed]"
+tools:
+  read: true
+  write: false
+---
+
+<objective>
+Display list of all saved algorithms with their status and basic info.
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/src/cli/commands.py
+</execution_context>
+
+<process>
+List algorithms:
+
+1. **Scan Algorithms Directory**
+   - Read .algomath/algorithms/
+   - Find all algorithm folders
+
+2. **Load Metadata**
+   - Read metadata.json from each
+   - Extract name, state, timestamps
+
+3. **Display List**
+   - Algorithm name
+   - Current state
+   - Step count
+   - Last modified date
+   - Status indicators
+
+4. **Detailed View (if --detailed)**
+   - Show full path
+   - Show all files present
+   - Show execution results
+   - Show verification status
+
+5. **Sort and Format**
+   - Sort by last modified
+   - Format as table or list
+   - Color code by status
+</process>
+
+<examples>
+
+**List algorithms:**
+/algo-list
+
+**Detailed list:**
+/algo-list --detailed
+
+</examples>
+
+<next_steps>
+/algo-extract - Create new algorithm
+/algo-status - Check specific algorithm
+</next_steps>

package/commands/algo-run.md

@@ -0,0 +1,81 @@
+---
+description: "Execute generated Python code"
+argument-hint: "[--skip] [--timeout <seconds>]"
+tools:
+  read: true
+  write: true
+---
+
+<objective>
+Execute generated Python code in a safe sandboxed environment.
+Captures stdout, stderr, and execution results.
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/src/execution/sandbox.py
+</execution_context>
+
+<process>
+Execute code in sandbox:
+
+1. **Load Generated Code**
+   - Read generated.py from current algorithm
+   - Validate code exists
+   - Check syntax
+
+2. **Create Sandbox**
+   - Create isolated temp directory
+   - Copy code to sandbox
+   - Set resource limits
+
+3. **Execute Code**
+   - Run in subprocess with timeout (default 30s)
+   - Capture stdout and stderr
+   - Monitor resource usage
+   - Handle exceptions
+
+4. **Process Results**
+   - Capture output
+   - Record execution time
+   - Check for errors
+   - Format results
+
+5. **Save Results**
+   - Write to execution.log
+   - Update metadata.json
+   - Update state to EXECUTION_COMPLETE
+
+6. **Display Results**
+   - Show output
+   - Show execution time
+   - Show status (success/error)
+   - Suggest next steps
+</process>
+
+<examples>
+
+**Execute code:**
+/algo-run
+
+**Skip execution:**
+/algo-run --skip
+
+**Custom timeout:**
+/algo-run --timeout 60
+
+</examples>
+
+<options>
+
+--skip
+Skip execution and proceed directly to verification
+
+--timeout <seconds>
+Maximum execution time (default: 30)
+
+</options>
+
+<next_steps>
+/algo-verify - Verify execution results
+/algo-status - Check algorithm state
+</next_steps>

package/commands/algo-status.md

@@ -0,0 +1,71 @@
+---
+description: "Show current algorithm state and progress"
+argument-hint: "[--verbose]"
+tools:
+  read: true
+  write: false
+---
+
+<objective>
+Display current algorithm state, progress, and available next steps.
+Shows workflow state, completion status, and suggestions.
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/src/cli/commands.py
+</execution_context>
+
+<process>
+Display status:
+
+1. **Load Current State**
+   - Read context manager state
+   - Get current algorithm name
+   - Get workflow state
+
+2. **Display Progress**
+   - State: IDLE → EXTRACTED → STEPS → CODE → EXECUTED → VERIFIED
+   - Show completion percentage
+   - Visual progress bar
+
+3. **Show Algorithm Info**
+   - Name
+   - Step count (if extracted)
+   - Code generated (yes/no)
+   - Executed (yes/no)
+   - Verified (yes/no)
+
+4. **Show Data Status**
+   - Source text present
+   - Steps structured
+   - Code generated
+   - Results captured
+
+5. **Suggest Next Steps**
+   Based on current state:
+   - IDLE: /algo-extract
+   - EXTRACTED: /algo-generate
+   - CODE_GENERATED: /algo-run
+   - EXECUTION_COMPLETE: /algo-verify
+   - VERIFIED: Done! /algo-extract for new
+
+6. **Display in Verbose Mode (if --verbose)**
+   - Full algorithm details
+   - Step list
+   - File paths
+   - Timestamps
+</process>
+
+<examples>
+
+**Show status:**
+/algo-status
+
+**Verbose status:**
+/algo-status --verbose
+
+</examples>
+
+<next_steps>
+Depends on current state - displayed dynamically
+</next_steps>

package/commands/algo-verify.md

@@ -0,0 +1,104 @@
+---
+description: "Verify execution results and explain algorithm behavior"
+argument-hint: "[--step <N>] [--detailed] [--diagnostic] [--expected <value>]"
+tools:
+  read: true
+  write: true
+---
+
+<objective>
+Verify execution results, compare to expected values, explain algorithm behavior,
+and identify potential edge cases.
+</objective>
+
+<execution_context>
+@/home/milgraph/Projects/algo_framework/src/verification/checker.py
+@/home/milgraph/Projects/algo_framework/src/verification/explainer.py
+@/home/milgraph/Projects/algo_framework/src/verification/edge_cases.py
+</execution_context>
+
+<process>
+Execute verification workflow:
+
+1. **Load Execution Results**
+   - Read execution.log
+   - Load algorithm steps
+   - Get generated code
+
+2. **Verify Execution**
+   - Check if execution succeeded
+   - Verify no errors
+   - Validate output present
+
+3. **Compare Results (if --expected)**
+   - Compare actual vs expected output
+   - Support numeric tolerance for floats
+   - Show differences
+
+4. **Generate Explanation**
+   - Brief mode: 1-2 sentence summary
+   - Detailed mode: Step-by-step walkthrough
+   - Step mode: Explain specific step N
+
+5. **Detect Edge Cases**
+   - Empty inputs
+   - Boundary values
+   - Division by zero
+   - Array bounds
+   - Infinite loops
+
+6. **Diagnostic Mode (if failed)**
+   - Analyze error
+   - Suggest fixes
+   - Explain failure point
+
+7. **Save Report**
+   - Write to verification.log
+   - Update state to VERIFIED
+
+8. **Display Summary**
+   - Execution status
+   - Explanation
+   - Edge cases found
+   - Next steps
+</process>
+
+<examples>
+
+**Verify results:**
+/algo-verify
+
+**Explain specific step:**
+/algo-verify --step 3
+
+**Detailed explanation:**
+/algo-verify --detailed
+
+**Compare with expected:**
+/algo-verify --expected "42"
+
+**Diagnostic for failed runs:**
+/algo-verify --diagnostic
+
+</examples>
+
+<options>
+
+--step <N>
+Explain step number N in detail
+
+--detailed
+Show detailed step-by-step explanation
+
+--diagnostic
+Run diagnostic mode for failed executions
+
+--expected <value>
+Expected output for comparison
+
+</options>
+
+<next_steps>
+/algo-extract - Start with new algorithm
+/algo-status - Check current state
+</next_steps>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "algomath-extract",
-  "version": "1.0.0",
+  "version": "1.0.2",
   "description": "AlgoMath Framework - Transform AI assistants into reliable mathematical problem-solving environments",
   "main": "index.js",
   "bin": {
@@ -33,6 +33,7 @@
   "files": [
     "bin/",
     "scripts/",
+    "commands/",
     "src/",
     "requirements.txt",
     "README.md"

package/scripts/postinstall.js

@@ -0,0 +1,20 @@
+#!/usr/bin/env node
+
+/**
+ * AlgoMath Post-Install Hook
+ * 
+ * Runs automatically after npm install
+ */
+
+console.log('\n╔════════════════════════════════════════════════════╗');
+console.log('║ AlgoMath Post-Install ║');
+console.log('╚════════════════════════════════════════════════════╝\n');
+
+console.log('AlgoMath has been installed successfully!\n');
+console.log('To install the opencode commands, run:');
+console.log('  npx algomath-extract --opencode\n');
+console.log('Or manually:');
+console.log('  node node_modules/algomath-extract/bin/install.js --opencode\n');
+
+console.log('Then try:');
+console.log('  /algo-help\n');