feat(tealtiger): add TealTiger governance memory integration

[nagasatish007]

Summary

Adds a TealTiger integration that stores AI agent governance decisions in Hindsight with importance-weighted retention. Critical security events (DENY: PII detected, secrets blocked) persist for compliance; routine approvals naturally decay.

Closes #2284

What it does

• Stores governance decisions via retain() with importance scores:
  • DENY → 0.90 (months retention)
  • MONITOR → 0.70 (weeks retention)
  • ALLOW → 0.55 (natural decay)
• Recalls contextually relevant past decisions for the same agent/tool
• Reflects on governance patterns (denial trends, anomalies)

Design principle

Storage = evidence, NOT authority. A stored ALLOW cannot authorize a future action. The policy engine is deterministic; memory provides context only.

Files

hindsight-integrations/tealtiger/
├── hindsight_tealtiger/
│   ├── __init__.py
│   └── memory.py
├── tests/
│   ├── __init__.py
│   └── test_memory.py
├── README.md
├── pyproject.toml
└── .gitignore

Testing

cd hindsight-integrations/tealtiger
pip install -e ".[dev]"
pytest tests/ -v

15 tests covering store, recall, reflect, importance mapping, tagging, and content formatting.

Related

• Issue: Importance-weighted governance decision storage for AI agent compliance #2284
• TealTiger: https://github.com/agentguard-ai/tealtiger
• PyPI (standalone): https://pypi.org/project/tealtiger-hindsight/

[nicoloboschi]

please provide a community integration on your own repo and add open up a pr here to being listed in https://hindsight.vectorize.io/integrations