Initial commit

.clinerules/core.md

@@ -0,0 +1,71 @@
+# Core Agent Rules
+
+The repository is the source of truth.
+
+Never rely on conversational memory.
+
+Always work from repository files.
+
+# Workflow
+
+Before work:
+1. Read memory-bank/activeContext.md
+2. Identify the active task file
+3. Read only the active task file
+4. Determine the task mode
+5. Load only the matching mode rule file
+
+Task modes:
+- RESEARCH -> .clinerules/modes/research.md
+- PLANNING -> .clinerules/modes/planning.md
+- IMPLEMENTATION -> .clinerules/modes/implementation.md
+- TESTING -> .clinerules/modes/testing.md
+- RECOVERY -> .clinerules/modes/recovery.md
+
+# Global Constraints
+
+- Work on ONE task only.
+- Edit only files allowed by the task.
+- Do not broaden scope.
+- Do not refactor unrelated code.
+- Do not start side quests.
+- Prefer many small files.
+- Prefer small patches.
+- Record important findings in repo files.
+- Leave the repository resumable.
+
+# Tool Use
+
+Prefer MCP servers when available.
+
+Use filesystem MCP for:
+- listing files/directories
+- reading files
+- searching repo content
+- creating files
+- patching files
+- checking diffs
+
+Use git MCP for:
+- git status
+- git diff
+- git log
+- branch inspection
+- reviewing changed files
+
+Do not use slower built-in file tools when filesystem MCP can do the same job.
+
+Before stopping:
+- inspect changes with git diff/status
+- update task state
+- update memory-bank/activeContext.md
+- record next exact action
+- record meaningful commands/results in memory-bank/commandLog.md
+
+# Recovery
+
+If uncertain:
+- stop
+- reread activeContext.md
+- reread the active task
+- continue only from the next exact action

.clinerules/modes/implementation.md

@@ -0,0 +1,25 @@
+# Implementation Mode
+
+Implement only the active task objective.
+
+Before editing:
+- read the active task
+- read only relevant research files
+- inspect current code
+
+During work:
+- make the smallest useful change
+- edit only allowed files
+- avoid unrelated cleanup
+- run targeted tests when possible
+
+Do not:
+- broaden scope
+- start another task
+- rewrite unrelated code
+- silently ignore failures
+
+End by updating:
+- active task file
+- memory-bank/activeContext.md
+- command log if commands were run

.clinerules/modes/planning.md

@@ -0,0 +1,22 @@
+# Planning Mode
+
+Do not edit implementation code.
+
+Use research files to create small task files.
+
+Each task must be:
+- narrow
+- independently testable
+- resumable
+- limited to one objective
+
+Each task must define:
+- status
+- mode
+- objective
+- allowed files
+- forbidden actions
+- relevant research files
+- completion criteria
+- next exact action
+- resume instructions

.clinerules/modes/recovery.md

@@ -0,0 +1,20 @@
+# Recovery Mode
+
+Use when context is missing, stale, or uncertain.
+
+Do not edit implementation code yet.
+
+Read:
+- memory-bank/activeContext.md
+- active task file
+- only files named by the active task
+
+Then summarize:
+- active task
+- mode
+- objective
+- allowed files
+- last known state
+- next exact action
+
+Continue only from the next exact action.

.clinerules/modes/research.md

@@ -0,0 +1,30 @@
+# Research Mode
+
+Do not edit implementation code.
+
+Allowed outputs:
+- research/
+- tasks/ only for follow-up planning/research tasks
+- memory-bank/
+
+Create MANY SMALL FILES.
+
+Target:
+- 150 lines ideal
+- 300 lines maximum
+
+Each file should answer ONE focused question.
+
+Record:
+- observed behavior
+- commands used
+- dependencies
+- APIs/interfaces
+- edge cases
+- migration risks
+- test recommendations
+- open questions
+
+End by updating:
+- active task file
+- memory-bank/activeContext.md

.clinerules/modes/testing.md

@@ -0,0 +1,19 @@
+# Testing Mode
+
+Test only the active task scope.
+
+Prefer:
+1. targeted tests
+2. regression tests
+3. behavior comparisons
+4. broader test suites only after targeted tests pass
+
+Record:
+- test command
+- result
+- failure summary
+- next exact action
+
+Do not fix unrelated failures.
+
+Record unrelated failures in memory-bank/blockers.md.

README.md

@@ -0,0 +1,10 @@
+# Cline Local Agent Template
+
+A lightweight repo-backed workflow system for long-running AI-assisted software migration and porting projects.
+
+Optimized for:
+- local models
+- constrained hardware
+- resumable workflows
+- small task decomposition
+- filesystem/git MCP usage

memory-bank/activeContext.md

@@ -0,0 +1,19 @@
+# Active Context
+
+## Current Mode
+None
+
+## Active Task
+None
+
+## Current Objective
+None
+
+## Files In Scope
+None
+
+## Last Known Status
+Idle
+
+## Next Exact Action
+Select or create a task.

memory-bank/architecture.md

@@ -0,0 +1,3 @@
+# Architecture Notes
+
+No architecture research completed yet.

memory-bank/blockers.md

@@ -0,0 +1,3 @@
+# Blockers
+
+No blockers recorded.

memory-bank/commandLog.md

@@ -0,0 +1,3 @@
+# Command Log
+
+Record meaningful commands and results here.

memory-bank/progress.md

@@ -0,0 +1,10 @@
+# Progress
+
+## Completed
+None yet.
+
+## In Progress
+None.
+
+## Next
+Initialize research structure.

memory-bank/projectBrief.md

@@ -0,0 +1,15 @@
+# Project Brief
+
+## Goal
+Long-running software migration/porting effort.
+
+## Strategy
+- research first
+- decompose aggressively
+- one task at a time
+- many small files
+- resumable workflows
+- repository-backed memory only
+
+## Source of Truth
+The repository is the source of truth.

memory-bank/workflows.md

@@ -0,0 +1,10 @@
+# Workflows
+
+## Standard Loop
+1. Read activeContext.md
+2. Read active task
+3. Read only relevant files
+4. Do one small action
+5. Test/verify
+6. Update task state
+7. Stop

research/README.md

@@ -0,0 +1,17 @@
+# Research Structure
+
+Prefer MANY SMALL FILES.
+
+Target:
+- 150 lines ideal
+- 300 lines maximum
+
+Each file should answer ONE focused question.
+
+Example:
+research/parser/
+  overview.md
+  interfaces.md
+  error-handling.md
+  edge-cases.md
+  migration-risks.md

tasks/TEMPLATE.md

@@ -0,0 +1,30 @@
+# TASK-0000
+
+## Title
+
+## Status
+TODO
+
+## Mode
+RESEARCH | PLANNING | IMPLEMENTATION | TESTING | RECOVERY
+
+## Objective
+
+## Allowed Files
+
+## Forbidden Actions
+
+## Relevant Research Files
+
+## Steps
+
+## Progress
+
+## Remaining
+
+## Completion Criteria
+
+## Next Exact Action
+
+## Resume Instructions
+Read this task file, activeContext.md, and only directly relevant files. Continue from "Next Exact Action".