Validate prompts before submission
Intercept user prompts before they reach the model — block, rewrite, redact, annotate, or inject context
Tool-Specific Hooks Hooks & Lifecycle
Supported
Partial
No
Unknown
2.1.78 - 2.1.143
3.82.0 - 3.83.0
0.116.0 - 0.128.0
2026.03 - 2026.05
2.5 - 3.2
1.13.3 - 2.1.29
2.1.10 - 2.1.78
3.82.0
0.92.0 - 0.115.0
2026.01 - 2026.03
2.4 - 2.5
1.13.3
2.1.10
3.82.0
0.88.0
2025.10
2.4
1.13.3
2.0.30 - 2.1.10
3.82.0
0.53.0 - 0.88.0
2025.10
2.0
1.12.31 - 1.13.3
1.0.54 - 2.0.30
3.82.0
0.2.0 - 0.53.0
2025.07 - 2025.10
1.2 - 2.0
1.2.1 - 1.12.28
0.2.54 - 1.0.48
3.36.0 - 3.82.0
0.2.0
2025.06 - 2025.07
0.45 - 1.2
1.2.1
0.2.31 - 0.2.54
1.5.0 - 3.33.0
—
—
0.45
1.0.2 - 1.2.1
Supported by all 6 agents.
HoverTap any cell for version details.
Agent event mapping
Sub-capabilities
- Block prompts that violate policies (PII, secrets, off-topic)
- Rewrite or augment prompts with additional context
- Claude Code: receives CWD, transcript ID, prompt text; return value can mutate or block
- Cursor: returns continue boolean + modified user_message