[FEATURE] PostCompaction hook stage for long-running skills

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

When a conversation is compacted (context truncated), skill instructions from earlier in the conversation are lost. This causes Claude to deviate from skill protocols after compaction.

Proposed Solution

Add a PostCompaction hook stage that fires after context compaction completes.

Use cases:

1. Re-inject active skill instructions into the conversation
2. Log/alert that compaction occurred during a long-running operation
3. Restore state markers needed for skill continuity

Example hook configuration:
{
"hooks": {
"PostCompaction": [
{
"hooks": [
{
"type": "command",
"command": "check-active-skill.sh",
"timeout": 5
}
]
}
]
}
}

Alternative Solutions

You need to monitor when compaction occurs and then re-instruct the team lead to re-read the skill to ensure they have the details.

Priority

High - Significant impact on productivity

Feature Category

Interactive mode (TUI)

Use Case Example

Example scenario:

• User invokes /my-team-review skill that orchestrates 6 agents over multiple QA rounds
• Skill instructions define a strict protocol (e.g., "never commit without permission", "always run cross-review before consolidation")
• After 10+ rounds, context compaction occurs
• Post-compaction, Claude only has a summary - the full skill instructions are gone
• Claude proceeds to violate the protocol (e.g., commits without permission, skips required steps)

Additional Context

_No response_