Skip to content

CHANGELOG.md

Latest commit

 

History

History
63 lines (57 loc) · 3.06 KB

CHANGELOG.md

File metadata and controls

63 lines (57 loc) · 3.06 KB

Changelog

v0.1.0-beta - Current Clean Release

  • Restored a clean public release boundary: private integration setup notes are not part of this repository.
  • Added runtime premium gating in ToolRegistry.execute() so premium tools are blocked no matter whether they are called from CLI, HTTP, or MCP.
  • Enabled MCP authentication by default in main.py; set ANA_MCP_KEY or MCP_API_KEY before calling tool endpoints.
  • Added release-safe backing modules for window_manager and clipboard_manager so AI Core adapters are no longer phantom tools.
  • Added edge_tts_voice as a release-safe voice tool. It registers cleanly and reports disabled if local TTS dependencies are unavailable.
  • Updated agent rules and project map to require fact-based edits, no private workspace paths, and no documentation for missing tools.
  • Added a release sync rule so code, docs, .env.example, tests, and release counts stay aligned when behavior changes.
  • Added regression tests for runtime premium blocking and adapter backing modules.
  • Documented MCP_API_KEY in .env.example.
  • Rewrote the WorkGraph architecture note as ASCII-only guidance focused on observation-first agents, tool quality over tool count, and task-based Frida usage.
  • Added workspace_situational_awareness, an observation-only WorkGraph tool that returns compact workspace JSON before an agent acts.
  • Hardened workspace_situational_awareness after bug hunt: file paths now resolve to their parent repo, loose AI parameters fall back safely, and local absolute repo paths are not exposed in the snapshot.
  • Repaired index.html for the public website: removed mojibake/emoji, aligned tool counts with the release baseline, documented MCP auth, and added it to release hygiene tests.
  • Added a first-page demo section to index.html so users can understand the workflow before reading the architecture notes.
  • Repositioned the public website and README around ANA MAX as a privacy-first hybrid situational-awareness runtime: observe, instrument when needed, act, verify, and learn.
  • Added docs/ANA_MAX_WOW_DEMO.md, a 90-second public demo blueprint that explains how to show ANA MAX as a situational-awareness runtime for agents.
  • Reframed the public README and website around the local QA lab vision: offline agents, private workstations, Ollama-style local models, voice feedback, and tool-assisted observe-act-verify work.
  • Added docs/LOCAL_QA_LAB_VISION.md to explain why ANA tools matter when internet access is blocked or data cannot leave the machine.
  • Removed large demo videos from git tracking and documented the rule that public videos should be hosted externally, then linked from the site.
  • Linked the current public YouTube demo from README, the website, and the video map.

Verification Baseline

python -m compileall -q main.py core tools vscode_extension
python main.py --test
python main.py --list-tools
python -m unittest discover -s tests -v

Expected:

  • Quick test: 3 PASS / 0 FAIL
  • Tool list: 64 loaded tools
  • Unit tests: all tests passing