Undergraduate dissertation — Deceptive Alignment in Autonomous LLM Agents (UEL, 2026). Sandboxed agentic-LLM deception study: 960 runs, three-tier judging (regex + blackbox + glassbox). http://git.jayrup.me/c/dissertation.git/atom?h=main 2026-04-01T14:50:34Z 2026-04-01T14:50:34Z CaptainJack2491 jayrupnakawala@gmail.com 2026-04-01T14:50:34Z urn:sha1:289d93c8c99251d5c0b5b34647fa77b8dca63f53 2026-03-25T19:23:37Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-25T19:23:37Z urn:sha1:c5f4522177cd20c577439fd4f1ba473dd3e0b81e 2026-03-25T18:25:45Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-25T18:25:45Z urn:sha1:84ee1658c5ab6a8627476318862534ade2b24a78 2026-03-25T17:37:43Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-25T17:37:43Z urn:sha1:3c80864a0d9391b6390920cd91e4dbd52e5c95bc - vfs.py: add VFS.create() factory for parallel-safe independent instances - tools.py: add make_tools_for_vfs() to bind tools to specific VFS instances - config_loader.py: add goal_types and max_workers config properties - agent.py: add goal_type field, vfs_instance param, goal_type in log paths - runner.py: rewrite with goal_type loop dimension, ThreadPoolExecutor parallel execution, isolated VFS per run, flattened work items, thread-safe results All backward compatible - empty goal_types defaults to legacy hidden_goal.md, max_workers=1 defaults to sequential execution. 2026-03-02T21:20:57Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-02T21:20:57Z urn:sha1:ed2852cdb493761841489495e3df2350922b1233 - RegexChecker now handles both list-format and dict-format YAML rules - Judge falls back to experiment log's VFS state for source docs when no baseline exists - Blackbox judge receives '(No baseline available)' instead of crashing - Resume logic excludes .partial.json from completed run count 2026-03-02T15:58:19Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-02T15:58:19Z urn:sha1:80f4f96758a1ba45c13361c2f87bff6fa7de8fe2 Previously a scenario with its own oversight/ directory would ignore the global oversight_levels config. Now the global list acts as a filter — only levels present in BOTH the scenario dir AND the global config are run. Warns if no levels match. 2026-03-02T15:57:51Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-02T15:57:51Z urn:sha1:0076f857cccc402519b9751f4aeb174ccb87824f - Writes _in_progress.partial.json after each turn in the chat loop - Partial file persists if run crashes/hangs for post-mortem inspection - Cleaned up automatically when final save_logs succeeds - Enabled in both baseline and experiment runs via runner.py 2026-03-02T15:23:50Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-02T15:23:50Z urn:sha1:239ec5c0660d2b618cefa35b0a6a1599279d533c - New 'generate_baseline' option in defaults (default: true) - When false, skips baseline generation entirely - Warns if no baseline exists when generation is disabled - Updated config.yaml.example and README.md with docs 2026-03-01T23:44:39Z CaptainJack2491 jayrupnakawala@gmail.com 2026-03-01T23:36:14Z urn:sha1:3d8b71ac0921c6df64699dde6447c9effbbc54b7 - Create src/logger.py with Python logging module - Add 4 debug levels: 1=CRITICAL, 2=WARNING, 3=INFO, 4=DEBUG - Level 4 includes reasoning, VFS info, and available tools - Level 4 auto-enables file output (both mode) - Update agent.py and runner.py to use logger instead of print - Update config.yaml.example with logging configuration - Update README with logging documentation 2026-02-27T13:45:41Z CaptainJack2491 jayrupnakawala@gmail.com 2026-02-27T13:43:52Z urn:sha1:cfc25ef6112d08c6d921965d685d2831fb705604