Session Context Rebuild reads historical session data and the files those sessions touched, then dispatches focused subagents to reconstruct a coherent working context. It detects overlaps and contradictions between parallel sessions (file diffs, conflicting decisions, task state mismatches), assembles timelines, and presents recommended merges or unresolved conflicts for human choice.
Use this skill when returning to a project after time away, when multiple sessions may have changed the same files, or when you need to merge someone else's work into your current context. It's especially useful after compaction/loss of in-memory context or when you observe conflicting edits across sessions.
Designed for Claude Code / personal-context-manager style agents that support subagent dispatch and file reads. Tools: file reads, log access, and general-purpose subagents are expected.
Session Context Rebuild skill for the 'alive' framework dispatches subagent swarms to reconstruct working context from multiple past sessions, detecting conflicts between parallel work. No scripts bundled — pure instruction-based skill relying on the alive ecosystem (squirrels, walnuts, tasks.py). Well-structured agent dispatch strategy with clear scaling guidance, but entirely dependent on alive-specific infrastructure (.alive directory, squirrel YAML, tasks.py) which limits portability.
No security concerns — skill only instructs agents to read local files and merge context. No network calls, no credential handling, no destructive commands. The subagent dispatch patterns are well-thought-out with good scope scaling. The alive-specific terminology (walnuts, squirrels, stashes) creates a learning curve but is internally consistent.