avc timeline

Renders the branch’s snapshots grouped by the agent session that produced them, each with a one-line change summary, interleaved with the restores, merges, and undos from the operations log. This is the “what did my agents do while I was away” report.

Usage

avc timeline                     # active branch, all sessions
avc timeline --session sess-42   # one session's story
avc timeline --branch main       # a specific branch
avc timeline --limit 100 --json

Flags

Flag	Description
--session <id>	Show only this session
--branch <name>	Branch to show (default: active branch)
--limit <n>	Max snapshots to include (default 50)
--json	JSON output

Where sessions come from

Sessions are the session_id / task attribution on snapshots. Pass them when snapshotting:

avc snapshot "before auth refactor" --session sess-42 --task "add auth endpoints"

or via the matching MCP avc_snapshot arguments. Agents are instructed to pass a stable session ID for the whole conversation and a one-line task description. Unattributed snapshots appear under (no session).

JSON output

{
  "branch": "main",
  "sessions": [
    {
      "session_id": "sess-42",
      "task": "add auth endpoints",
      "agents": ["claude"],
      "started_at": 1712275200,
      "ended_at": 1712278800,
      "events": [
        { "kind": "snapshot", "timestamp": 1712275200, "snapshot_id": "snap-abc",
          "label": "auto: before auth refactor", "agent_name": "claude",
          "file_count": 42, "summary": "1 file: modified auth.go (+40 -12)" },
        { "kind": "operation", "timestamp": 1712278800, "op_kind": "restore",
          "details": "restored snapshot snap-abc" }
      ]
    }
  ]
}

Summaries missing from older snapshots are computed (and cached) lazily. The same data is served by the web UI at /api/timeline.