feat(daemon): agent-driven repo checkout with bare clone cache

Agents now decide which repo to use based on issue context and check out
repos on demand via `multica repo checkout <url>`. Workspace repos are
cached locally as bare clones for fast worktree creation.

Key changes:
- Add repocache package for bare clone management (clone, fetch, worktree)
- Add `multica repo checkout` CLI command that talks to local daemon
- Add POST /repo/checkout endpoint on daemon health server
- Pass workspace repos metadata through register + task claim responses
- Remove pre-created worktrees from execenv (workdir starts empty)
- Update CLAUDE.md template to instruct agents to use `multica repo checkout`
- Pass MULTICA_DAEMON_PORT, WORKSPACE_ID, AGENT_NAME, TASK_ID env vars to agent

server/internal/handler/agent.go

@@ -81,6 +81,13 @@ func agentToResponse(a db.Agent) AgentResponse {
 	}
 }
 
+// RepoData holds repository information included in claim responses so the
+// daemon can set up worktrees for each workspace repo.
+type RepoData struct {
+	URL         string `json:"url"`
+	Description string `json:"description"`
+}
+
 type AgentTaskResponse struct {
 	ID             string         `json:"id"`
 	AgentID        string         `json:"agent_id"`
@@ -94,6 +101,7 @@ type AgentTaskResponse struct {
 	Result         any            `json:"result"`
 	Error          *string        `json:"error"`
 	Agent          *TaskAgentData `json:"agent,omitempty"`
+	Repos          []RepoData     `json:"repos,omitempty"`
 	CreatedAt      string         `json:"created_at"`
 	PriorSessionID string         `json:"prior_session_id,omitempty"` // session ID from a previous task on same issue
 }

server/internal/handler/daemon.go

@@ -51,7 +51,8 @@ func (h *Handler) DaemonRegister(w http.ResponseWriter, r *http.Request) {
 		writeError(w, http.StatusBadRequest, "at least one runtime is required")
 		return
 	}
-	if _, err := h.Queries.GetWorkspace(r.Context(), parseUUID(req.WorkspaceID)); err != nil {
+	ws, err := h.Queries.GetWorkspace(r.Context(), parseUUID(req.WorkspaceID))
+	if err != nil {
 		writeError(w, http.StatusNotFound, "workspace not found")
 		return
 	}
@@ -106,7 +107,16 @@ func (h *Handler) DaemonRegister(w http.ResponseWriter, r *http.Request) {
 		"runtimes": resp,
 	})
 
-	writeJSON(w, http.StatusOK, map[string]any{"runtimes": resp})
+	// Include workspace repos so the daemon can cache them locally.
+	var repos []RepoData
+	if ws.Repos != nil {
+		json.Unmarshal(ws.Repos, &repos)
+	}
+	if repos == nil {
+		repos = []RepoData{}
+	}
+
+	writeJSON(w, http.StatusOK, map[string]any{"runtimes": resp, "repos": repos})
 }
 
 // DaemonDeregister marks runtimes as offline when the daemon shuts down.
@@ -217,6 +227,16 @@ func (h *Handler) ClaimTaskByRuntime(w http.ResponseWriter, r *http.Request) {
 		}
 	}
 
+	// Include workspace repos so the daemon can set up worktrees.
+	if issue, err := h.Queries.GetIssue(r.Context(), task.IssueID); err == nil {
+		if ws, err := h.Queries.GetWorkspace(r.Context(), issue.WorkspaceID); err == nil && ws.Repos != nil {
+			var repos []RepoData
+			if json.Unmarshal(ws.Repos, &repos) == nil && len(repos) > 0 {
+				resp.Repos = repos
+			}
+		}
+	}
+
 	// Look up the prior session for this (agent, issue) pair so the daemon
 	// can resume the Claude Code conversation context.
 	if prior, err := h.Queries.GetLastTaskSession(r.Context(), db.GetLastTaskSessionParams{