v0.2.2 - Local MCP server and client registration

Patch release focused on exposing Emotion Engine through a local stdio MCP server and making client registration harder to misconfigure.

Highlights

• Added a zero-dependency local stdio MCP server for Emotion Engine runtime/protocol tools, including status, compact summaries, record policy, advisory appraisal, session lifecycle, turn recording, trust settlement, and recent log inspection.
• Bundled the MCP server into the Codex skill package while keeping Agent Harness responsibilities out of the MCP boundary: install refresh, doctor, repair, manifest checks, and sidecar drift remain Agent Harness work.
• Added MCP client registration examples for Codex CLI, Claude Code, Claude Desktop, and project-level .mcp.json.
• Added scripts/register_mcp_client.py so users can register the local stdio server without hand-writing MCP config.
• Clarified state path rules: generic projects use .emotion-engine/emotion-state.json, while Codex and Agent Harness targets use .emotion-engine/codex-state.json so wrappers and MCP clients do not split state.

Verification

• PYTHONDONTWRITEBYTECODE=1 python3 tests/test_mcp_registration.py -> 6 tests OK
• PYTHONDONTWRITEBYTECODE=1 python3 tests/test_mcp_server.py -> 5 tests OK
• PYTHONDONTWRITEBYTECODE=1 python3 tests/test_demo_static.py -> 6 tests OK
• python3 -m py_compile scripts/emotion_engine_mcp.py scripts/register_mcp_client.py tests/test_mcp_server.py tests/test_mcp_registration.py -> OK
• PYTHONDONTWRITEBYTECODE=1 python3 tests/test_codex_integration.py -> 7 tests OK
• PYTHONDONTWRITEBYTECODE=1 python3 -m unittest discover -s tests -> 82 tests OK
• git diff --check -> OK

v0.2.1 - Persistence hardening and agent install guide

Patch release focused on making v0.2 safer and easier to try.

Highlights

• Hardened local state persistence with state-file locking, atomic temp-file writes, fsync/replace, backup recovery for corrupt JSON reads, and locked CLI transactions for mutating commands.
• Added INSTALL_WITH_AGENT.md, a copy-ready guide for asking Codex, Claude Code, Cursor, or similar coding agents to install Emotion Engine as a local state sidecar.
• Added a web demo v0.2 Compare block that explains Mood only as an internal ablation and the default integrated mood + affective_pulse state package as the real runtime path.
• Improved Codex policy handling for playful, intimacy, and relationship-calibration turns; repeated generic praise now habituates instead of creating low-value records.
• Expanded tests for demo/install assets, low-value log compaction, atomic writes, backup recovery, and concurrent state updates.

Verification

• PYTHONDONTWRITEBYTECODE=1 python3 -m unittest discover -s tests -> 70 tests OK
• git diff --check -> OK

v0.2.0 - Affective pulse expressiveness layer

Emotion Engine v0.2.0 - Affective pulse expressiveness layer

This release separates stable continuity from visible per-turn movement.

What changed

• Added affective_pulse, a short-lived event movement layer that complements slow PAD mood.
• Added volatility_profile with steady, expressive, and dramatic_test envelopes.
• Updated protocol/schema, prompt preview, status output, and the minimal agent demo to expose pulse separately from mood.
• Improved style onboarding for warm, intimate, playful, and assertive profiles.
• Preserved existing state baselines during shape upgrades; hosts can still recalibrate character baselines explicitly.
• Fixed strict schema semantics so PAD deltas and affective pulse movement can carry negative A/D values.

Why it matters

In v0.1.x, PAD mood intentionally moved slowly. That was useful for continuity, but it could make long-running agents feel too flat when a specific turn was playful, tense, corrective, or warm.

v0.2.0 keeps PAD mood stable and adds a separate pulse layer for visible short-term movement. This lets hosts show turn-level affect without making durable mood or trust swing wildly.

Validation

• python3 -m unittest discover -s tests -> 61 tests OK
• python3 examples/minimal-agent/run_demo.py --state /tmp/emotion-engine-v020-demo-check-2.json
• python3 scripts/prompt_preview.py --message "..."
• git diff --check
• JSON syntax check for root and bundled schema/templates

v0.1.2 - Conservative trust settlement

v0.1.2 - Conservative trust settlement

This release strengthens Emotion Engine's session-end lifecycle and trust-settlement behavior.

Included:

• Added settle_trust, a conservative host-side trust settlement command.
• Made trust settlement idempotent for the same session trajectory, returning already_settled on repeat execution.
• Kept trust_history as a numeric ledger while writing settlement evidence and reasons to emotion_log.
• Added trust tier progress fields to status without removing existing status fields.
• Updated Codex, Hermes, Claude, OpenClaw, GPT/API, protocol, and concept docs to use settle_trust as the default session-end trust path.
• Expanded tests for trust settlement policy, idempotency, evidence placement, and minimal-agent lifecycle behavior.

Verification:

• 57 tests passed with python3 -m unittest discover -s tests.

Live demo:
https://pioneerjeff-labs.github.io/emotion-engine/demo/

Repository:
https://github.com/pioneerjeff-labs/emotion-engine

v0.1.1 - Protocol, Codex, and GPT integration guide

This release adds the Codex integration and clarifies Emotion Engine's protocol and GPT/API integration boundary.

Included:

• Codex user-level Skill package
• OpenAI GPT / API host-side integration guide
• emotion-engine-state/v2 protocol and schema docs
• Clearer trust v1 semantics: agent-to-user only
• trust_history as numeric ledger, with explanations kept in emotion_log
• GitHub Pages root redirect to the live demo

Verification:

• 42 tests passed

v0.1.0 - Emotional continuity for LLM agents

Initial public release of Emotion Engine.

Emotion Engine is a lightweight state layer for emotionally continuous LLM agents. It persists PAD state, trust, decay, boundaries, and compact emotional memories while leaving contextual judgment and reply generation to the LLM.

Included in this release:

• Side-by-side web demo with English / Chinese toggle
• Core Python state lifecycle tools
• Prompt preview utility
• OpenClaw integration
• Claude Skill / Claude Code integration
• Hermes Agent integration
• HermesHub preparation script
• English and Chinese documentation

Live demo:
https://pioneerjeff-labs.github.io/emotion-engine/demo/