feat: add 6 production-ready plugins

All plugins are stdlib-only Python with no external dependencies.

- auto-summarizer: extractive conversation summary for long context
  compression, persists per-agent summaries to disk
- conversation-logger: JSONL audit logs per agent with ISO 8601
  timestamps, auto-creates log directory tree
- guardrails: safety filter detecting PII (email, phone, SSN, CC),
  prompt injection patterns, and credential exposure via regex
- keyword-memory: extracts entities (emails, URLs, dates, technical
  terms like camelCase/snake_case/dotted identifiers) as memories
- sentiment-tracker: keyword-based sentiment scoring with intensifiers
  and negation handling, only injects context for non-neutral sentiment
- todo-tracker: detects action items via 7 task patterns, tracks
  completion, deduplicates, persists per-agent with 20-item FIFO limit

plugins/sentiment-tracker/README.md

@@ -0,0 +1,44 @@
+# sentiment-tracker
+
+Analyzes user message sentiment using keyword-based scoring and injects emotional context so agents can respond with appropriate tone. No external ML libraries required (stdlib only).
+
+## Scoring Method
+
+- **Positive words** (~30): great, love, excellent, awesome, helpful, appreciate, etc. (+1 each)
+- **Negative words** (~30): bad, terrible, frustrated, broken, bug, error, crash, etc. (-1 each)
+- **Intensifiers**: very, extremely, really, absolutely, totally (multiply next sentiment word by 1.5x)
+- **Negators**: not, no, never, don't, doesn't, isn't, can't, won't (flip next word's polarity)
+
+The raw score is normalized by message length and clamped to [-1.0, 1.0].
+
+## Classification
+
+| Score Range | Label | Action |
+|-------------|-------|--------|
+| > 0.3 | positive | Inject positive context memory |
+| < -0.3 | negative | Inject frustration-aware memory |
+| -0.3 to 0.3 | neutral | No memory injected (avoid context clutter) |
+
+## Hooks
+
+| Hook | Script | Description |
+|------|--------|-------------|
+| ingest | `hooks/ingest.py` | Analyzes message sentiment and returns emotional context as a memory fragment |
+
+## Example Output
+
+Negative sentiment:
+```json
+{"type": "ingest_result", "memories": [{"content": "[sentiment] User appears frustrated (score: -0.6). Consider acknowledging the issue."}]}
+```
+
+Positive sentiment:
+```json
+{"type": "ingest_result", "memories": [{"content": "[sentiment] User seems satisfied (score: 0.7). Positive interaction."}]}
+```
+
+Neutral sentiment returns an empty memories list.
+
+## Usage
+
+Installed automatically when enabled in agent configuration. No external dependencies required (stdlib only).