feat(project-tool): multi-repo single-owner ownership + takeover (§7 M3)

Ports the M3 milestone from the superseded feat/hierarchies branch onto main's
evolved apply engine. A group node is authoritatively managed by at most one
project-root; `pic apply --dir` claims, conflicts, takes over, and (with
--prune) structurally reaps owned nodes.

- Migration 0066: `projects(id, key UNIQUE)` + promote the inert
  `groups.owner_project` (0047) to a real FK (ON DELETE SET NULL) + index.
- CLI mints a stable, gitignored project key in `.picloud/project.json`
  (`pic init`, or lazily on first tree plan/apply) and presents it on every
  tree request; `pic apply --dir --takeover` flag; `pic plan --dir` surfaces
  ownership conflicts + structural-prune candidates read-only.
- Server: prepare_tree resolves ownership read-only (token folds each group's
  owner key, so a claim/takeover by another repo between plan and apply trips
  StateMoved). apply_tree upserts the project in-tx, claims unclaimed declared
  groups, and takes over owned ones under `--takeover`. --prune reaps
  owned-but-undeclared groups leaf-first (delete_group_tx RESTRICT — never
  another repo's or a UI-owned node).
- Authz (§7.4, ownership ⟂ RBAC): takeover requires GroupAdmin per contested
  node — enforced in authz_tree (pre-tx) AND re-verified in-tx at the ownership
  decision, so --force (which waives the staleness token) can't open a
  takeover-without-admin window. The attacker-supplied project key is
  length/charset-validated server-side.

Adaptation to main (which lacks the superseded branch's M2 declarative group
create/reparent): groups pre-exist, so ownership stamps existing declared
nodes rather than claim-on-create; the declarative attach-point is deferred.

Review fixes (adversarial pass, 3 findings, all closed):
- HIGH: an empty `[group]` node was claimed capability-free — require a
  baseline GroupScriptsWrite (editor) on every group apply node, so claiming
  ownership needs write authority (Ownership ⟂ RBAC).
- MEDIUM: structural prune would silently CASCADE a group's §11.6 shared
  collections + secrets (which postdate M3's original design). Guard: a prune
  candidate holding shared data/secrets is KEPT with a warning (plus its
  candidate ancestors, so a preserved child never aborts a parent delete).
- LOW: a corrupt `.picloud/project.json` silently re-minted a new key
  (orphaning ownership) — now fails loudly if the file exists but won't parse.

Tests: tests/ownership.rs (claim → conflict → takeover → flip; non-admin
takeover → 403; prune-owned-only; prune-refuses-to-cascade-shared-data);
format_conflicts + validate_project_key unit tests; schema golden reblessed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

crates/manager-core/migrations/0066_owner_project.sql

@@ -0,0 +1,39 @@
+-- Phase 5 / M3 (v1.2 Hierarchies): multi-repo single-owner ownership (§7).
+--
+-- A group node is authoritatively MANAGED by at most one project-root (the
+-- repo whose manifest declares it as something it applies, not merely
+-- references). `groups.owner_project` has carried this seam since 0047 as an
+-- inert, FK-less UUID. M3 makes it a real reference: a `projects` table keyed
+-- by a stable, gitignored project key the CLI mints in `.picloud/`, and a
+-- foreign key from `groups.owner_project` to it.
+--
+-- Ownership is recorded at GROUP granularity; apps inherit their owning project
+-- from their group (apps are never claimed directly). A NULL `owner_project`
+-- means the node is UI/API-owned — no manifest fights it (§7.5). First apply
+-- claims; a second project's apply to an owned node is refused unless
+-- `--takeover` (group-admin gated). Ownership ⟂ RBAC: owning the manifest does
+-- NOT grant write — the actor still needs the usual group capabilities (§7.4).
+--
+-- ON DELETE SET NULL: deleting a project row (not something the platform does
+-- today) reverts its nodes to UI-owned rather than cascading away real groups.
+
+CREATE TABLE projects (
+    id         UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    -- Stable, opaque key minted by `pic init` and persisted (gitignored) in the
+    -- repo's `.picloud/`. The CLI presents it on every tree apply; the server
+    -- maps it to this row (upsert-by-key), so the same repo keeps the same
+    -- project identity across clones/CI without committing any server id.
+    key        TEXT NOT NULL UNIQUE,
+    created_at TIMESTAMPTZ NOT NULL DEFAULT NOW()
+);
+
+-- Promote the inert `groups.owner_project` (0047) to a real FK now that the
+-- referent exists. Existing rows are all NULL (no projects yet), so this adds
+-- no validation burden.
+ALTER TABLE groups
+    ADD CONSTRAINT groups_owner_project_fkey
+    FOREIGN KEY (owner_project) REFERENCES projects(id) ON DELETE SET NULL;
+
+-- The ownership reconcile reads "which groups does project P own?" (to find
+-- owned-but-undeclared nodes for structural prune) and "who owns group G?".
+CREATE INDEX groups_owner_project_idx ON groups (owner_project);

crates/manager-core/src/apply_api.rs

@@ -280,6 +280,16 @@ async fn group_apply_handler(
 // must hold the relevant read/write caps for EVERY node touched.
 // ----------------------------------------------------------------------------
 
+#[derive(Deserialize)]
+struct TreePlanRequest {
+    bundle: TreeBundle,
+    /// Stable project key from the repo's `.picloud/` link state (§7, M3).
+    /// Lets `plan` surface ownership conflicts and prune candidates. Optional
+    /// for legacy callers (then no ownership diagnosis is reported).
+    #[serde(default)]
+    project_key: Option<String>,
+}
+
 #[derive(Deserialize)]
 struct TreeApplyRequest {
     bundle: TreeBundle,
@@ -296,15 +306,25 @@ struct TreeApplyRequest {
     /// unless it appears here; `--yes` alone does NOT add it.
     #[serde(default)]
     approved_envs: Vec<String>,
+    /// Stable project key (§7, M3): the apply claims unclaimed declared groups
+    /// for this project and refuses ones another project owns.
+    #[serde(default)]
+    project_key: Option<String>,
+    /// Take over declared groups owned by another project (group-admin gated).
+    #[serde(default)]
+    allow_takeover: bool,
 }
 
 async fn tree_plan_handler(
     State(svc): State<ApplyService>,
     Extension(principal): Extension<Principal>,
-    Json(bundle): Json<TreeBundle>,
+    Json(req): Json<TreePlanRequest>,
 ) -> Result<Json<TreePlanResult>, ApplyError> {
-    authz_tree(&svc, &principal, &bundle, None).await?;
+    authz_tree(&svc, &principal, &req.bundle, None, false).await?;
-    Ok(Json(svc.plan_tree(&bundle).await?))
+    Ok(Json(
+        svc.plan_tree(&req.bundle, req.project_key.as_deref())
+            .await?,
+    ))
 }
 
 async fn tree_apply_handler(
@@ -312,7 +332,14 @@ async fn tree_apply_handler(
     Extension(principal): Extension<Principal>,
     Json(req): Json<TreeApplyRequest>,
 ) -> Result<Json<ApplyReport>, ApplyError> {
-    authz_tree(&svc, &principal, &req.bundle, Some(req.prune)).await?;
+    authz_tree(
+        &svc,
+        &principal,
+        &req.bundle,
+        Some(req.prune),
+        req.allow_takeover,
+    )
+    .await?;
     enforce_env_approval(
         &svc,
         &principal,
@@ -325,8 +352,10 @@ async fn tree_apply_handler(
         .apply_tree(
             &req.bundle,
             req.prune,
-            principal.user_id,
+            &principal,
             req.expected_token.as_deref(),
+            req.project_key.as_deref(),
+            req.allow_takeover,
         )
         .await?;
     Ok(Json(report))
@@ -396,6 +425,7 @@ async fn authz_tree(
     principal: &Principal,
     bundle: &TreeBundle,
     apply_prune: Option<bool>,
+    allow_takeover: bool,
 ) -> Result<(), ApplyError> {
     for node in &bundle.nodes {
         match node.kind {
@@ -419,6 +449,28 @@ async fn authz_tree(
                     Some(prune) => {
                         require_group_node_writes(svc, principal, group_id, &node.bundle, prune)
                             .await?;
+                        // Takeover gate (§7.4): seizing a group another project
+                        // owns needs GroupAdmin specifically — a second, independent
+                        // gate on top of the write caps. We gate it for any
+                        // already-claimed node under `--takeover` (a superset of
+                        // genuinely-contested, never laxer); claiming an unclaimed
+                        // node, or applying without `--takeover`, does not require
+                        // admin. The authoritative owner rewrite happens in-tx in
+                        // `apply_tree` (which re-checks under the tx lock).
+                        if allow_takeover
+                            && crate::group_repo::get_group_owner(&svc.pool, group_id)
+                                .await
+                                .map_err(|e| ApplyError::Backend(e.to_string()))?
+                                .is_some()
+                        {
+                            require(
+                                svc.authz.as_ref(),
+                                principal,
+                                Capability::GroupAdmin(group_id),
+                            )
+                            .await
+                            .map_err(map_authz)?;
+                        }
                     }
                     None => {
                         require(
@@ -509,15 +561,19 @@ async fn require_group_node_writes(
     bundle: &Bundle,
     prune: bool,
 ) -> Result<(), ApplyError> {
-    if prune || !bundle.scripts.is_empty() {
+    // Baseline (§7.4, M3): applying a group node CLAIMS its ownership
-        require(
+    // (`owner_project`) in-tx — a management write — even when the bundle has no
-            svc.authz.as_ref(),
+    // scripts/vars to reconcile. Require an editor-level group write cap floor so
-            principal,
+    // an EMPTY `[group]` node can't be claimed capability-free (Ownership ⟂ RBAC:
-            Capability::GroupScriptsWrite(group_id),
+    // owning the manifest never grants write, and claiming still needs write).
-        )
+    // Subsumes the per-resource GroupScriptsWrite below.
-        .await
+    require(
-        .map_err(map_authz)?;
+        svc.authz.as_ref(),
-    }
+        principal,
+        Capability::GroupScriptsWrite(group_id),
+    )
+    .await
+    .map_err(map_authz)?;
     if prune || !bundle.vars.is_empty() {
         require(
             svc.authz.as_ref(),
@@ -576,7 +632,7 @@ impl IntoResponse for ApplyError {
                 StatusCode::UNPROCESSABLE_ENTITY,
                 json!({ "error": self.to_string() }),
             ),
-            Self::StateMoved | Self::ApprovalRequired(_) => {
+            Self::StateMoved | Self::ApprovalRequired(_) | Self::OwnershipConflict(_) => {
                 (StatusCode::CONFLICT, json!({ "error": self.to_string() }))
             }
             Self::Forbidden => (StatusCode::FORBIDDEN, json!({ "error": self.to_string() })),

crates/manager-core/src/apply_service.rs

@@ -32,15 +32,16 @@ use std::sync::Arc;
 use picloud_orchestrator_core::routing::{pattern, RouteTable};
 use picloud_shared::{
     AdminUserId, AppId, DispatchMode, DocsEventOp, FilesEventOp, GroupId, HostKind, KvEventOp,
-    MasterKey, PathKind, Route, Script, ScriptId, ScriptKind, ScriptOwner, ScriptSandbox,
+    MasterKey, PathKind, Principal, Route, Script, ScriptId, ScriptKind, ScriptOwner,
-    ScriptValidator, TriggerId,
+    ScriptSandbox, ScriptValidator, TriggerId,
 };
 use serde::{Deserialize, Serialize};
 use sqlx::PgPool;
+use uuid::Uuid;
 
 use crate::app_domain_repo::AppDomainRepository;
 use crate::app_repo::AppRepository;
-use crate::authz::AuthzRepo;
+use crate::authz::{require, AuthzDenied, AuthzRepo, Capability};
 use crate::config_resolver::CHAIN_LEVELS_CTE;
 use crate::repo::{
     delete_script_tx, insert_script_tx, update_script_tx, NewScript, ScriptPatch, ScriptRepository,
@@ -568,6 +569,23 @@ pub struct TreePlanResult {
     /// the gate at `plan` time, before an `apply` refuses.
     #[serde(default, skip_serializing_if = "Vec::is_empty")]
     pub approvals_required: Vec<String>,
+    /// Group nodes this manifest declares that another project already owns
+    /// (§7, M3). A read-only `plan` surfaces them so CI sees the conflict before
+    /// an `apply` refuses; `apply` proceeds only with `--takeover`.
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub conflicts: Vec<OwnershipConflict>,
+    /// Slugs of groups this project owns that are absent from the manifest —
+    /// what `apply --prune` would delete (leaf-first, RESTRICT-guarded).
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub structural_prunes: Vec<String>,
+}
+
+/// One ownership conflict: a declared group already claimed by another project.
+#[derive(Debug, Clone, Serialize)]
+pub struct OwnershipConflict {
+    pub slug: String,
+    /// The owning project's key (so the operator knows whose node it is).
+    pub owner_key: String,
 }
 
 // ----------------------------------------------------------------------------
@@ -677,6 +695,11 @@ pub enum ApplyError {
          re-run with `--approve {0}` (a blanket `--yes` does not cover it)"
     )]
     ApprovalRequired(String),
+    #[error(
+        "one or more declared groups are owned by another project; \
+         re-run with `--takeover` to claim them (group-admin required): {0}"
+    )]
+    OwnershipConflict(String),
     #[error("authorization repo error: {0}")]
     AuthzRepo(String),
     #[error("backend: {0}")]
@@ -1503,14 +1526,19 @@ impl ApplyService {
     ///
     /// # Errors
     /// `Invalid` for a malformed/unknown bundle; `Backend` for repo failures.
-    pub async fn plan_tree(&self, bundle: &TreeBundle) -> Result<TreePlanResult, ApplyError> {
+    pub async fn plan_tree(
-        let (prepared, token) = self.prepare_tree(bundle).await?;
+        &self,
-        let nodes = prepared
+        bundle: &TreeBundle,
-            .into_iter()
+        project_key: Option<&str>,
+    ) -> Result<TreePlanResult, ApplyError> {
+        let pt = self.prepare_tree(bundle, project_key).await?;
+        let nodes = pt
+            .prepared
+            .iter()
             .map(|p| NodePlan {
                 kind: p.node.kind,
                 slug: p.node.slug.clone(),
-                plan: p.plan,
+                plan: p.plan.clone(),
             })
             .collect();
         let approvals_required = bundle
@@ -1520,8 +1548,10 @@ impl ApplyService {
             .unwrap_or_default();
         Ok(TreePlanResult {
             nodes,
-            state_token: token,
+            state_token: pt.token,
             approvals_required,
+            conflicts: pt.conflicts,
+            structural_prunes: pt.prune_candidates.into_iter().map(|(_, s)| s).collect(),
         })
     }
 
@@ -1534,19 +1564,37 @@ impl ApplyService {
     /// # Errors
     /// `Invalid` for a bad bundle; `StateMoved` if the bound token is stale;
     /// `Backend` for repo/transaction failures.
+    #[allow(clippy::too_many_arguments, clippy::too_many_lines)]
     pub async fn apply_tree(
         &self,
         bundle: &TreeBundle,
         prune: bool,
-        actor: AdminUserId,
+        principal: &Principal,
         expected_token: Option<&str>,
+        project_key: Option<&str>,
+        allow_takeover: bool,
     ) -> Result<ApplyReport, ApplyError> {
-        let (prepared, token) = self.prepare_tree(bundle).await?;
+        let actor = principal.user_id;
+        let pt = self.prepare_tree(bundle, project_key).await?;
         if let Some(expected) = expected_token {
-            if token != expected {
+            if pt.token != expected {
                 return Err(ApplyError::StateMoved);
             }
         }
+        // Ownership pre-check (§7, M3): refuse, before opening the tx, if any
+        // declared group is owned by another project and `--takeover` was not
+        // given. The in-tx re-read below is the authoritative gate against a
+        // concurrent claim; this is the fast, clear path for the common case.
+        if !allow_takeover && !pt.conflicts.is_empty() {
+            return Err(ApplyError::OwnershipConflict(format_conflicts(
+                &pt.conflicts,
+            )));
+        }
+        let PreparedTree {
+            prepared,
+            existing_group_nodes,
+            ..
+        } = pt;
 
         let mut tx = self
             .pool
@@ -1569,6 +1617,59 @@ impl ApplyService {
         let mut report = ApplyReport::default();
         let mut group_script_index: HashMap<GroupId, HashMap<String, ScriptId>> = HashMap::new();
 
+        // Resolve (upsert) this repo's project row inside the tx, so declared
+        // group nodes can be stamped with a stable owner id (§7, M3). A tree
+        // apply with no project key (legacy callers) skips all ownership.
+        let our_project_id = match project_key {
+            Some(k) => Some(
+                crate::group_repo::upsert_project_tx(&mut tx, k)
+                    .await
+                    .map_err(map_group_err)?,
+            ),
+            None => None,
+        };
+
+        // Phase 0.5 — ownership reconcile (§7, M3): for each declared group,
+        // re-read the owner under this tx's lock (authoritative) and claim it
+        // (NULL owner) or take it over (`--takeover`). A conflict that appeared
+        // after `plan` and wasn't authorized aborts the whole apply.
+        if let Some(pid) = our_project_id {
+            for (gid, slug) in &existing_group_nodes {
+                let owner = crate::group_repo::get_group_owner_tx(&mut tx, *gid)
+                    .await
+                    .map_err(map_group_err)?;
+                match owner {
+                    Some((oid, _)) if oid == pid => {} // already ours — no write
+                    Some((_, okey)) => {
+                        if !allow_takeover {
+                            return Err(ApplyError::OwnershipConflict(format!(
+                                "{slug} (owned by project {okey})"
+                            )));
+                        }
+                        // Authoritative takeover gate (§7.4): GroupAdmin is
+                        // re-verified HERE, at the in-tx ownership decision, not
+                        // only in the pre-tx `authz_tree` pool read — otherwise
+                        // `--force` (which waives the staleness token) would let
+                        // a non-admin seize a group that another project claimed
+                        // after the pre-tx check saw it unowned. Ownership ⟂ RBAC.
+                        require(self.authz.as_ref(), principal, Capability::GroupAdmin(*gid))
+                            .await
+                            .map_err(map_authz_denied)?;
+                        crate::group_repo::set_group_owner_tx(&mut tx, *gid, pid)
+                            .await
+                            .map_err(map_group_err)?;
+                        report.groups_taken_over += 1;
+                    }
+                    None => {
+                        crate::group_repo::set_group_owner_tx(&mut tx, *gid, pid)
+                            .await
+                            .map_err(map_group_err)?;
+                        report.groups_claimed += 1;
+                    }
+                }
+            }
+        }
+
         // Phase A — groups first: reconcile each and record its post-create
         // `name -> id` so descendant app nodes can resolve inherited bindings.
         for p in &prepared {
@@ -1623,6 +1724,17 @@ impl ApplyService {
             }
         }
 
+        // Phase C — structural prune (§7, M3): with `--prune`, delete groups
+        // THIS project owns that the manifest no longer declares, leaf-first
+        // (delete_group_tx is RESTRICT, so a group still holding apps/child-groups
+        // aborts the apply with a clear error rather than orphaning data).
+        if prune {
+            if let Some(pid) = our_project_id {
+                self.structural_prune_tx(&mut tx, pid, &existing_group_nodes, &mut report)
+                    .await?;
+            }
+        }
+
         tx.commit()
             .await
             .map_err(|e| ApplyError::Backend(e.to_string()))?;
@@ -1641,6 +1753,107 @@ impl ApplyService {
         Ok(report)
     }
 
+    /// Phase C of a tree apply (§7, M3): delete groups THIS project owns that the
+    /// manifest no longer declares, leaf-first. `delete_group_tx` is RESTRICT, so
+    /// a candidate still holding apps or non-pruned child groups aborts the whole
+    /// apply (no silent orphaning). Scoped to project-owned nodes: another repo's
+    /// groups and UI-owned (unclaimed) groups are never touched.
+    async fn structural_prune_tx(
+        &self,
+        tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+        project_id: Uuid,
+        existing_group_nodes: &[(GroupId, String)],
+        report: &mut ApplyReport,
+    ) -> Result<(), ApplyError> {
+        // Declared groups are kept; every other group this project owns is a
+        // prune candidate.
+        let declared: HashSet<String> = existing_group_nodes
+            .iter()
+            .map(|(_, s)| s.to_lowercase())
+            .collect();
+
+        let owned = crate::group_repo::list_owned_groups_tx(tx, project_id)
+            .await
+            .map_err(map_group_err)?;
+        let mut candidates: Vec<(GroupId, String, Option<GroupId>)> = owned
+            .into_iter()
+            .filter(|(_, slug, _)| !declared.contains(&slug.to_lowercase()))
+            .collect();
+        if candidates.is_empty() {
+            return Ok(());
+        }
+
+        // Data-loss guard (§7 M3): the apps/child-group FKs are RESTRICT (a
+        // non-empty group aborts loudly via delete_group_tx), but the §11.6
+        // shared surfaces (group_collections + secrets) CASCADE. Refuse to
+        // silently vaporize a tenant's shared data/secrets just because a
+        // manifest node was dropped — mark such a group KEEP (with a warning),
+        // AND keep every candidate ancestor of it (RESTRICT would abort the
+        // apply if we tried to delete a parent whose protected child we're
+        // preserving). Also re-confirm ownership under the tx here — a
+        // concurrent `--takeover` could have flipped a candidate (undeclared, so
+        // the bound-plan token doesn't cover it) since the candidate SELECT.
+        let parent_of: HashMap<GroupId, Option<GroupId>> =
+            candidates.iter().map(|(g, _, p)| (*g, *p)).collect();
+        let mut keep: HashSet<GroupId> = HashSet::new();
+        for (gid, slug, _) in &candidates {
+            let still_owned = crate::group_repo::get_group_owner_tx(tx, *gid)
+                .await
+                .map_err(map_group_err)?
+                .map(|(oid, _)| oid)
+                == Some(project_id);
+            let protected = crate::group_repo::group_has_protected_data_tx(tx, *gid)
+                .await
+                .map_err(map_group_err)?;
+            if !still_owned || protected {
+                if protected {
+                    report.warnings.push(format!(
+                        "group `{slug}` owns shared collections or secrets; \
+                         not pruned (delete it explicitly with `pic groups rm`)"
+                    ));
+                }
+                // Walk up the candidate chain, keeping this node + every
+                // candidate ancestor (a kept descendant RESTRICT-pins them).
+                let mut cur = Some(*gid);
+                while let Some(g) = cur {
+                    if !keep.insert(g) {
+                        break; // already walked from here
+                    }
+                    cur = parent_of.get(&g).copied().flatten();
+                }
+            }
+        }
+        candidates.retain(|(g, _, _)| !keep.contains(g));
+        if candidates.is_empty() {
+            return Ok(());
+        }
+
+        // Delete leaf-first: defer any candidate that is still the parent of
+        // another candidate until that child is gone.
+        while !candidates.is_empty() {
+            let parents: HashSet<GroupId> = candidates.iter().filter_map(|(_, _, p)| *p).collect();
+            let before = candidates.len();
+            let mut remaining: Vec<(GroupId, String, Option<GroupId>)> = Vec::new();
+            for (gid, slug, parent) in std::mem::take(&mut candidates) {
+                if parents.contains(&gid) {
+                    remaining.push((gid, slug, parent));
+                    continue;
+                }
+                crate::group_repo::delete_group_tx(tx, gid)
+                    .await
+                    .map_err(map_group_err)?;
+                report.groups_pruned += 1;
+            }
+            if remaining.len() == before {
+                return Err(ApplyError::Invalid(
+                    "structural prune could not order owned groups for deletion (cycle?)".into(),
+                ));
+            }
+            candidates = remaining;
+        }
+        Ok(())
+    }
+
     /// Resolve nodes to owners, validate each (apps' route/trigger targets may
     /// bind to in-tree ancestor group scripts), compute each node's plan, and
     /// fold a combined bound-plan token. Shared by `plan_tree` / `apply_tree`.
@@ -1648,10 +1861,14 @@ impl ApplyService {
     async fn prepare_tree<'a>(
         &self,
         bundle: &'a TreeBundle,
-    ) -> Result<(Vec<PreparedNode<'a>>, String), ApplyError> {
+        project_key: Option<&str>,
+    ) -> Result<PreparedTree<'a>, ApplyError> {
         if bundle.nodes.is_empty() {
             return Err(ApplyError::Invalid("project tree has no nodes".into()));
         }
+        if let Some(k) = project_key {
+            validate_project_key(k)?;
+        }
         // Register in-tree group nodes first: their (resolved) ids and endpoint
         // script names, which an app node's binding validation may reference.
         let mut group_id_by_slug: HashMap<String, GroupId> = HashMap::new();
@@ -1787,8 +2004,70 @@ impl ApplyService {
             envs.sort_unstable();
             token_parts.push(format!("proj|{}", envs.join(",")));
         }
+
+        // Ownership (§7, M3): resolve this repo's project, then classify each
+        // declared group node as ours / unclaimed / owned-by-another. Folding
+        // the owner key into the token means a claim or takeover by a different
+        // repo between plan and apply trips StateMoved. (Every group node on the
+        // tree-apply path pre-exists — `resolve_group` errors otherwise — so
+        // there is no to-create node to skip, unlike a create-capable engine.)
+        let our_project_id = match project_key {
+            Some(k) => crate::group_repo::get_project_id_by_key(&self.pool, k)
+                .await
+                .map_err(|e| ApplyError::Backend(e.to_string()))?,
+            None => None,
+        };
+        let mut existing_group_nodes: Vec<(GroupId, String)> = Vec::new();
+        let mut declared_group_slugs: HashSet<String> = HashSet::new();
+        let mut conflicts: Vec<OwnershipConflict> = Vec::new();
+        for n in &bundle.nodes {
+            if n.kind != NodeKind::Group {
+                continue;
+            }
+            let lc = n.slug.to_lowercase();
+            declared_group_slugs.insert(lc.clone());
+            let gid = group_id_by_slug[&lc];
+            existing_group_nodes.push((gid, n.slug.clone()));
+            let owner = crate::group_repo::get_group_owner(&self.pool, gid)
+                .await
+                .map_err(|e| ApplyError::Backend(e.to_string()))?;
+            match &owner {
+                Some((_, key)) => token_parts.push(format!("own|{lc}|{key}")),
+                None => token_parts.push(format!("own|{lc}|none")),
+            }
+            if let Some((oid, okey)) = owner {
+                if Some(oid) != our_project_id {
+                    conflicts.push(OwnershipConflict {
+                        slug: n.slug.clone(),
+                        owner_key: okey,
+                    });
+                }
+            }
+        }
+
+        // Structural-prune candidates: groups THIS project owns that the
+        // manifest omits. Computed read-only here (so `plan` can list them);
+        // `apply --prune` re-derives the set in-tx and deletes leaf-first.
+        let mut prune_candidates: Vec<(GroupId, String)> = Vec::new();
+        if let Some(pid) = our_project_id {
+            for (gid, slug) in crate::group_repo::list_owned_groups(&self.pool, pid)
+                .await
+                .map_err(|e| ApplyError::Backend(e.to_string()))?
+            {
+                if !declared_group_slugs.contains(&slug.to_lowercase()) {
+                    prune_candidates.push((gid, slug));
+                }
+            }
+        }
+
         token_parts.sort_unstable();
-        Ok((prepared, combine_tokens(&token_parts)))
+        Ok(PreparedTree {
+            prepared,
+            token: combine_tokens(&token_parts),
+            existing_group_nodes,
+            conflicts,
+            prune_candidates,
+        })
     }
 
     /// Resolve an app node's inherited route/trigger targets to script ids,
@@ -3723,6 +4002,15 @@ pub struct ApplyReport {
     pub suppressions_created: u32,
     #[serde(default)]
     pub suppressions_deleted: u32,
+    /// Group nodes this apply claimed (NULL owner → this project), §7 M3.
+    #[serde(default)]
+    pub groups_claimed: u32,
+    /// Group nodes this apply took over from another project (`--takeover`).
+    #[serde(default)]
+    pub groups_taken_over: u32,
+    /// Owned-but-undeclared groups deleted by structural prune (`--prune`).
+    #[serde(default)]
+    pub groups_pruned: u32,
     #[serde(skip_serializing_if = "Vec::is_empty")]
     pub warnings: Vec<String>,
 }
@@ -3745,6 +4033,21 @@ struct PreparedNode<'a> {
     app_chain: Vec<GroupId>,
 }
 
+/// The read-only result of `prepare_tree`: per-node plans + one combined token,
+/// plus the ownership classification (§7, M3) the plan surfaces and the apply
+/// acts on.
+struct PreparedTree<'a> {
+    prepared: Vec<PreparedNode<'a>>,
+    token: String,
+    /// Declared group nodes in the bundle: `(id, slug)` — the apply path
+    /// re-reads each owner in-tx (authoritative) to claim / detect takeover.
+    existing_group_nodes: Vec<(GroupId, String)>,
+    /// Declared groups already owned by ANOTHER project (read-only diagnosis).
+    conflicts: Vec<OwnershipConflict>,
+    /// Owned-but-undeclared groups: `(id, slug)` — structural-prune candidates.
+    prune_candidates: Vec<(GroupId, String)>,
+}
+
 /// FNV-1a over a set of already-sorted per-resource token parts — the same
 /// hashing `state_token` uses, lifted to combine many nodes into one tree
 /// token. Order-independent because the caller sorts `parts`.
@@ -3863,6 +4166,57 @@ fn map_trig(e: TriggerRepoError) -> ApplyError {
     }
 }
 
+/// Map a group-repo error to the apply surface (§7, M3). A `Conflict` (e.g. a
+/// structural-prune candidate still holding descendants) is caller-actionable →
+/// `Invalid`; a `NotFound` mid-apply is a lost race; `Db` is backend.
+fn map_group_err(e: crate::group_repo::GroupRepositoryError) -> ApplyError {
+    use crate::group_repo::GroupRepositoryError as E;
+    match e {
+        E::Conflict(m) => ApplyError::Invalid(m),
+        E::NotFound(_) => ApplyError::Invalid("a referenced group was not found".into()),
+        E::Db(e) => ApplyError::Backend(e.to_string()),
+    }
+}
+
+/// Map an in-service authz denial to the apply error surface (403 vs 500).
+fn map_authz_denied(e: AuthzDenied) -> ApplyError {
+    match e {
+        AuthzDenied::Denied => ApplyError::Forbidden,
+        AuthzDenied::Repo(r) => ApplyError::AuthzRepo(r.to_string()),
+    }
+}
+
+/// Validate an attacker-supplied project key (§7, M3). The key is a bearer
+/// handle for ownership, so it must be bounded and opaque before it's reflected
+/// into error messages or upserted. The CLI mints a v4 UUID; we accept any
+/// short, printable, delimiter-free token (ASCII alphanumeric plus `-`/`_`).
+fn validate_project_key(key: &str) -> Result<(), ApplyError> {
+    if key.is_empty() || key.len() > 128 {
+        return Err(ApplyError::Invalid(
+            "project key must be 1–128 characters".into(),
+        ));
+    }
+    if !key
+        .bytes()
+        .all(|b| b.is_ascii_alphanumeric() || b == b'-' || b == b'_')
+    {
+        return Err(ApplyError::Invalid(
+            "project key may contain only ASCII letters, digits, '-' and '_'".into(),
+        ));
+    }
+    Ok(())
+}
+
+/// Render an ownership-conflict list as `slug (owned by project key), …` for
+/// the `OwnershipConflict` error message.
+fn format_conflicts(conflicts: &[OwnershipConflict]) -> String {
+    conflicts
+        .iter()
+        .map(|c| format!("{} (owned by project {})", c.slug, c.owner_key))
+        .collect::<Vec<_>>()
+        .join(", ")
+}
+
 /// A stable fingerprint of an app's live state, covering exactly what the
 /// reconcile diff keys on: script identity + write-counter (`version` bumps on
 /// any script edit), route identity + binding/attrs, trigger semantic identity
@@ -4820,4 +5174,34 @@ mod tests {
         assert!(!policy.confirm_required("unknown"));
         assert_eq!(policy.gated_envs(), vec!["production".to_string()]);
     }
+
+    #[test]
+    fn format_conflicts_lists_each_node_and_owner() {
+        // The conflict message drives the CLI error a developer sees; pin it.
+        let c = vec![
+            OwnershipConflict {
+                slug: "team-a".into(),
+                owner_key: "key-x".into(),
+            },
+            OwnershipConflict {
+                slug: "team-b".into(),
+                owner_key: "key-y".into(),
+            },
+        ];
+        assert_eq!(
+            format_conflicts(&c),
+            "team-a (owned by project key-x), team-b (owned by project key-y)"
+        );
+        assert_eq!(format_conflicts(&[]), "");
+    }
+
+    #[test]
+    fn validate_project_key_bounds_and_charset() {
+        assert!(validate_project_key("a1b2-c3_d4").is_ok());
+        assert!(validate_project_key(&uuid::Uuid::nil().to_string()).is_ok());
+        assert!(validate_project_key("").is_err());
+        assert!(validate_project_key(&"x".repeat(129)).is_err());
+        assert!(validate_project_key("has space").is_err());
+        assert!(validate_project_key("semi;colon").is_err());
+    }
 }

crates/manager-core/src/group_repo.rs

@@ -339,6 +339,184 @@ impl GroupRepository for PostgresGroupRepository {
     }
 }
 
+// ----------------------------------------------------------------------------
+// Project ownership (§7, M3). A `projects` row is keyed by the stable, opaque
+// key the CLI mints in `.picloud/`; `groups.owner_project` references it. These
+// helpers are free functions (pool + tx variants) so the apply path can claim
+// ownership inside the single apply transaction (first-commit-wins), while the
+// read-only plan path can surface conflicts without writing.
+// ----------------------------------------------------------------------------
+
+/// Resolve a project key to its id, if the project has ever been persisted.
+/// `None` means no apply has claimed under this key yet (so `plan` treats every
+/// node as claimable). Read-only — used by the plan path.
+pub(crate) async fn get_project_id_by_key(
+    pool: &PgPool,
+    key: &str,
+) -> Result<Option<Uuid>, GroupRepositoryError> {
+    let row: Option<(Uuid,)> = sqlx::query_as("SELECT id FROM projects WHERE key = $1")
+        .bind(key)
+        .fetch_optional(pool)
+        .await?;
+    Ok(row.map(|r| r.0))
+}
+
+/// Upsert the project keyed by `key` inside the apply tx and return its id.
+/// Idempotent: re-applying the same repo reuses the same project identity.
+pub(crate) async fn upsert_project_tx(
+    tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+    key: &str,
+) -> Result<Uuid, GroupRepositoryError> {
+    let row: (Uuid,) = sqlx::query_as(
+        "INSERT INTO projects (key) VALUES ($1) \
+         ON CONFLICT (key) DO UPDATE SET key = EXCLUDED.key RETURNING id",
+    )
+    .bind(key)
+    .fetch_one(&mut **tx)
+    .await?;
+    Ok(row.0)
+}
+
+/// The project that owns `group_id`, as `(project_id, project_key)`. `None` when
+/// the node is UI/API-owned (no claim). Read-only — used by plan + the takeover
+/// pre-gate. Joins through `projects` so the caller gets the human-facing key
+/// for a conflict message.
+pub(crate) async fn get_group_owner(
+    pool: &PgPool,
+    group_id: GroupId,
+) -> Result<Option<(Uuid, String)>, GroupRepositoryError> {
+    let row: Option<(Uuid, String)> = sqlx::query_as(
+        "SELECT p.id, p.key FROM groups g \
+         JOIN projects p ON p.id = g.owner_project WHERE g.id = $1",
+    )
+    .bind(group_id.into_inner())
+    .fetch_optional(pool)
+    .await?;
+    Ok(row)
+}
+
+/// The same owner lookup, but inside the apply tx — the authoritative read that
+/// the claim/conflict decision keys on (so a concurrent claim that committed
+/// after `plan` is seen under this tx's per-node advisory lock).
+pub(crate) async fn get_group_owner_tx(
+    tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+    group_id: GroupId,
+) -> Result<Option<(Uuid, String)>, GroupRepositoryError> {
+    let row: Option<(Uuid, String)> = sqlx::query_as(
+        "SELECT p.id, p.key FROM groups g \
+         JOIN projects p ON p.id = g.owner_project WHERE g.id = $1",
+    )
+    .bind(group_id.into_inner())
+    .fetch_optional(&mut **tx)
+    .await?;
+    Ok(row)
+}
+
+/// Stamp (claim or take over) `group_id`'s owning project inside the apply tx.
+/// Bumps `structure_version` so an ownership flip trips a bound plan token.
+pub(crate) async fn set_group_owner_tx(
+    tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+    group_id: GroupId,
+    project_id: Uuid,
+) -> Result<(), GroupRepositoryError> {
+    let res = sqlx::query(
+        "UPDATE groups SET owner_project = $2, \
+             structure_version = structure_version + 1, updated_at = NOW() WHERE id = $1",
+    )
+    .bind(group_id.into_inner())
+    .bind(project_id)
+    .execute(&mut **tx)
+    .await?;
+    if res.rows_affected() == 0 {
+        return Err(GroupRepositoryError::NotFound(group_id));
+    }
+    Ok(())
+}
+
+/// Every group `(id, slug)` owned by `project_id` — the candidate set for
+/// structural prune (those absent from the manifest are the ones to delete).
+/// Read-only; the apply path filters and deletes leaf-first under `delete_group_tx`.
+pub(crate) async fn list_owned_groups(
+    pool: &PgPool,
+    project_id: Uuid,
+) -> Result<Vec<(GroupId, String)>, GroupRepositoryError> {
+    let rows: Vec<(Uuid, String)> =
+        sqlx::query_as("SELECT id, slug FROM groups WHERE owner_project = $1")
+            .bind(project_id)
+            .fetch_all(pool)
+            .await?;
+    Ok(rows
+        .into_iter()
+        .map(|(id, slug)| (id.into(), slug))
+        .collect())
+}
+
+/// The same owned-group enumeration inside the apply tx, with each node's
+/// `parent_id` so the prune can delete leaf-first (children before parents).
+pub(crate) async fn list_owned_groups_tx(
+    tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+    project_id: Uuid,
+) -> Result<Vec<(GroupId, String, Option<GroupId>)>, GroupRepositoryError> {
+    let rows: Vec<(Uuid, String, Option<Uuid>)> =
+        sqlx::query_as("SELECT id, slug, parent_id FROM groups WHERE owner_project = $1")
+            .bind(project_id)
+            .fetch_all(&mut **tx)
+            .await?;
+    Ok(rows
+        .into_iter()
+        .map(|(id, slug, parent)| (id.into(), slug, parent.map(GroupId::from)))
+        .collect())
+}
+
+/// Does this group hold data a structural prune would SILENTLY cascade away
+/// (§7 M3 safety guard)? The `apps`/child-group FKs are RESTRICT (delete aborts
+/// loudly), but the §11.6 group-owned surfaces are `ON DELETE CASCADE`: shared
+/// collections (`group_collections`, covering kv/docs/files/topic/queue) and
+/// group secrets (owner-polymorphic `secrets.group_id`). Rather than let
+/// `apply --prune` vaporize a tenant's shared data + secrets because a manifest
+/// node was dropped, the prune skips such a group with a warning — the operator
+/// must delete it explicitly. Returns true if the group has any shared-collection
+/// marker or any secret.
+pub(crate) async fn group_has_protected_data_tx(
+    tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+    group_id: GroupId,
+) -> Result<bool, GroupRepositoryError> {
+    let row: (bool,) = sqlx::query_as(
+        "SELECT EXISTS (SELECT 1 FROM group_collections WHERE group_id = $1) \
+              OR EXISTS (SELECT 1 FROM secrets           WHERE group_id = $1)",
+    )
+    .bind(group_id.into_inner())
+    .fetch_one(&mut **tx)
+    .await?;
+    Ok(row.0)
+}
+
+/// Delete a group inside the apply tx (structural prune, §7 M3). RESTRICT is the
+/// real guard: the FK from apps/child-groups aborts the whole apply (mapped to a
+/// clean `Conflict`) rather than orphaning data, so the caller must delete
+/// leaf-first and never reaches a group that still holds live descendants.
+pub(crate) async fn delete_group_tx(
+    tx: &mut sqlx::Transaction<'_, sqlx::Postgres>,
+    group_id: GroupId,
+) -> Result<(), GroupRepositoryError> {
+    let res = sqlx::query("DELETE FROM groups WHERE id = $1")
+        .bind(group_id.into_inner())
+        .execute(&mut **tx)
+        .await;
+    match res {
+        Ok(r) if r.rows_affected() == 0 => Err(GroupRepositoryError::NotFound(group_id)),
+        Ok(_) => Ok(()),
+        Err(sqlx::Error::Database(e)) if e.is_foreign_key_violation() => {
+            Err(GroupRepositoryError::Conflict(
+                "group still has descendants (apps or subgroups); \
+                 structural prune refuses to orphan them"
+                    .into(),
+            ))
+        }
+        Err(e) => Err(e.into()),
+    }
+}
+
 #[derive(sqlx::FromRow)]
 struct GroupRow {
     id: Uuid,

crates/manager-core/tests/expected_schema.txt

@@ -318,6 +318,11 @@ table: outbox
   claimed_by: text NULL
   created_at: timestamp with time zone NOT NULL default=now()
 
+table: projects
+  id: uuid NOT NULL default=gen_random_uuid()
+  key: text NOT NULL
+  created_at: timestamp with time zone NOT NULL default=now()
+
 table: pubsub_trigger_details
   trigger_id: uuid NOT NULL
   topic_pattern: text NOT NULL
@@ -579,6 +584,7 @@ indexes on group_queue_messages:
   idx_group_queue_messages_group_collection: public.group_queue_messages USING btree (group_id, collection)
 
 indexes on groups:
+  groups_owner_project_idx: public.groups USING btree (owner_project)
   groups_parent_id_idx: public.groups USING btree (parent_id)
   groups_pkey: public.groups USING btree (id)
   groups_slug_key: public.groups USING btree (slug)
@@ -595,6 +601,10 @@ indexes on outbox:
   idx_outbox_due: public.outbox USING btree (next_attempt_at) WHERE (claimed_at IS NULL)
   outbox_pkey: public.outbox USING btree (id)
 
+indexes on projects:
+  projects_key_key: public.projects USING btree (key)
+  projects_pkey: public.projects USING btree (id)
+
 indexes on pubsub_trigger_details:
   pubsub_trigger_details_pkey: public.pubsub_trigger_details USING btree (trigger_id)
 
@@ -817,6 +827,7 @@ constraints on group_queue_messages:
   [PRIMARY KEY] group_queue_messages_pkey: PRIMARY KEY (id)
 
 constraints on groups:
+  [FOREIGN KEY] groups_owner_project_fkey: FOREIGN KEY (owner_project) REFERENCES projects(id) ON DELETE SET NULL
   [FOREIGN KEY] groups_parent_id_fkey: FOREIGN KEY (parent_id) REFERENCES groups(id) ON DELETE RESTRICT
   [PRIMARY KEY] groups_pkey: PRIMARY KEY (id)
   [UNIQUE] groups_slug_key: UNIQUE (slug)
@@ -834,6 +845,10 @@ constraints on outbox:
   [FOREIGN KEY] outbox_app_id_fkey: FOREIGN KEY (app_id) REFERENCES apps(id) ON DELETE CASCADE
   [PRIMARY KEY] outbox_pkey: PRIMARY KEY (id)
 
+constraints on projects:
+  [PRIMARY KEY] projects_pkey: PRIMARY KEY (id)
+  [UNIQUE] projects_key_key: UNIQUE (key)
+
 constraints on pubsub_trigger_details:
   [FOREIGN KEY] pubsub_trigger_details_trigger_id_fkey: FOREIGN KEY (trigger_id) REFERENCES triggers(id) ON DELETE CASCADE
   [PRIMARY KEY] pubsub_trigger_details_pkey: PRIMARY KEY (trigger_id)
@@ -973,3 +988,4 @@ constraints on vars:
   0063: materialized unique
   0064: shared topic queue kinds
   0065: group queues
+  0066: owner project

crates/picloud-cli/Cargo.toml

@@ -34,6 +34,8 @@ toml = "0.8"
 directories = "5"
 rpassword = "7"
 anyhow = "1"
+# §7 M3: minting the stable project key in `.picloud/project.json`.
+uuid = { version = "1", features = ["v4"] }
 
 [dev-dependencies]
 assert_cmd = "2"

crates/picloud-cli/src/client.rs

@@ -1262,17 +1262,29 @@ impl Client {
     }
 
     /// `POST /api/v1/admin/tree/plan` — diff a whole project tree (Phase 5).
-    pub async fn plan_tree(&self, bundle: &serde_json::Value) -> Result<TreePlanDto> {
+    /// `project_key` (§7, M3) lets the plan surface ownership conflicts + prune
+    /// candidates for this repo.
+    pub async fn plan_tree(
+        &self,
+        bundle: &serde_json::Value,
+        project_key: Option<&str>,
+    ) -> Result<TreePlanDto> {
+        let body = serde_json::json!({
+            "bundle": bundle,
+            "project_key": project_key,
+        });
         let resp = self
             .request(Method::POST, "/api/v1/admin/tree/plan")
-            .json(bundle)
+            .json(&body)
             .send()
             .await?;
         decode(resp).await
     }
 
     /// `POST /api/v1/admin/tree/apply` — reconcile a whole project tree in one
-    /// transaction (Phase 5).
+    /// transaction (Phase 5). `project_key` + `allow_takeover` drive the §7 M3
+    /// ownership claim / takeover.
+    #[allow(clippy::too_many_arguments)]
     pub async fn apply_tree(
         &self,
         bundle: &serde_json::Value,
@@ -1280,6 +1292,8 @@ impl Client {
         expected_token: Option<&str>,
         env: Option<&str>,
         approved_envs: &[String],
+        project_key: Option<&str>,
+        allow_takeover: bool,
     ) -> Result<ApplyReportDto> {
         let body = serde_json::json!({
             "bundle": bundle,
@@ -1287,6 +1301,8 @@ impl Client {
             "expected_token": expected_token,
             "env": env,
             "approved_envs": approved_envs,
+            "project_key": project_key,
+            "allow_takeover": allow_takeover,
         });
         let resp = self
             .request(Method::POST, "/api/v1/admin/tree/apply")
@@ -1296,6 +1312,12 @@ impl Client {
         if resp.status() == reqwest::StatusCode::CONFLICT {
             let body = resp.text().await.unwrap_or_default();
             let msg = parse_error_body(&body).unwrap_or(body);
+            // An ownership conflict (§7) is self-explanatory ("re-run with
+            // --takeover"); only a staleness (token) conflict warrants the
+            // re-plan hint. Distinguish on the server's message.
+            if msg.contains("owned by another project") {
+                return Err(anyhow!("{msg}"));
+            }
             return Err(anyhow!(
                 "{msg}\nThe project tree changed since your last `pic plan`. Re-run \
                  `pic plan --dir` to review, then `pic apply --dir` — or add `--force`."
@@ -1538,6 +1560,19 @@ pub struct TreePlanDto {
     /// Environments the project marks confirm-required (§4.2, M5).
     #[serde(default)]
     pub approvals_required: Vec<String>,
+    /// Declared groups another project already owns (§7, M3).
+    #[serde(default)]
+    pub conflicts: Vec<OwnershipConflictDto>,
+    /// Owned-but-undeclared group slugs `apply --prune` would delete (§7, M3).
+    #[serde(default)]
+    pub structural_prunes: Vec<String>,
+}
+
+/// A group declared by this manifest that another project owns (§7, M3).
+#[derive(Debug, Deserialize)]
+pub struct OwnershipConflictDto {
+    pub slug: String,
+    pub owner_key: String,
 }
 
 #[derive(Debug, Deserialize)]
@@ -1599,6 +1634,15 @@ pub struct ApplyReportDto {
     pub suppressions_created: u32,
     #[serde(default)]
     pub suppressions_deleted: u32,
+    /// Group nodes claimed this apply (§7, M3).
+    #[serde(default)]
+    pub groups_claimed: u32,
+    /// Group nodes taken over from another project this apply (§7, M3).
+    #[serde(default)]
+    pub groups_taken_over: u32,
+    /// Owned-but-undeclared groups deleted by structural prune (§7, M3).
+    #[serde(default)]
+    pub groups_pruned: u32,
     #[serde(default)]
     pub warnings: Vec<String>,
 }

crates/picloud-cli/src/cmds/apply.rs

@@ -143,6 +143,7 @@ pub async fn run_tree(
     force: bool,
     env: Option<&str>,
     approve: &[String],
+    takeover: bool,
     mode: OutputMode,
 ) -> Result<()> {
     let creds = config::resolve()?;
@@ -159,6 +160,11 @@ pub async fn run_tree(
     // local check just fails fast with a clear message.
     let approved = resolve_approvals(env, &envs, approve)?;
 
+    // Mint/read this repo's stable project key (§7, M3): the apply claims the
+    // declared groups for this project and refuses ones another project owns
+    // (unless `--takeover`, group-admin gated).
+    let project_key = crate::linkstate::ensure_project_key(dir)?;
+
     let token_key = crate::cmds::plan::TREE_TOKEN_KEY;
     let expected_token = if force {
         None
@@ -169,13 +175,28 @@ pub async fn run_tree(
     };
 
     let report = client
-        .apply_tree(&bundle, prune, expected_token.as_deref(), env, &approved)
+        .apply_tree(
+            &bundle,
+            prune,
+            expected_token.as_deref(),
+            env,
+            &approved,
+            Some(&project_key),
+            takeover,
+        )
         .await?;
     crate::linkstate::clear_plan(dir, token_key);
 
     let mut block = KvBlock::new();
     block
         .field("nodes", node_count.to_string())
+        .field(
+            "groups",
+            format!(
+                "claimed {} taken-over {} pruned {}",
+                report.groups_claimed, report.groups_taken_over, report.groups_pruned
+            ),
+        )
         .field(
             "scripts",
             format!(

crates/picloud-cli/src/cmds/init.rs

@@ -87,6 +87,9 @@ pub fn run(
     }
     fs::write(&manifest_path, body).with_context(|| format!("writing {MANIFEST_FILE}"))?;
     ensure_gitignored(dir)?;
+    // Mint this repo's stable, gitignored project identity (§7, M3) up front so
+    // the first tree apply already has an ownership handle to claim under.
+    crate::linkstate::ensure_project_key(dir)?;
 
     let mut block = KvBlock::new();
     block

crates/picloud-cli/src/cmds/plan.rs

@@ -50,7 +50,10 @@ pub async fn run_tree(dir: &Path, env: Option<&str>, mode: OutputMode) -> Result
     let creds = config::resolve()?;
     let client = Client::from_creds(&creds)?;
     let (bundle, _count, _envs) = crate::discover::build_tree(dir, env)?;
-    let plan = client.plan_tree(&bundle).await?;
+    // Mint/read this repo's stable project key (§7, M3) so the plan can surface
+    // ownership conflicts + structural-prune candidates for us.
+    let project_key = crate::linkstate::ensure_project_key(dir)?;
+    let plan = client.plan_tree(&bundle, Some(&project_key)).await?;
     if !plan.state_token.is_empty() {
         if let Err(e) = crate::linkstate::write_plan(dir, TREE_TOKEN_KEY, &plan.state_token) {
             eprintln!("warning: could not record plan state for `pic apply`: {e}");
@@ -96,6 +99,24 @@ fn render_tree(plan: &TreePlanDto, mode: OutputMode) {
             eprintln!("  - {e}");
         }
     }
+
+    // Ownership (§7, M3): surface groups another project owns (apply refuses
+    // without `--takeover`) and owned-but-undeclared groups `--prune` would
+    // delete, so both are visible before an apply acts on them.
+    if mode != OutputMode::Json {
+        if !plan.conflicts.is_empty() {
+            eprintln!("\ngroups owned by another project (apply with --takeover to claim):");
+            for c in &plan.conflicts {
+                eprintln!("  - {} (owned by project {})", c.slug, c.owner_key);
+            }
+        }
+        if !plan.structural_prunes.is_empty() {
+            eprintln!("\ngroups this project owns but no longer declares (--prune deletes):");
+            for s in &plan.structural_prunes {
+                eprintln!("  - {s}");
+            }
+        }
+    }
 }
 
 /// Assemble the wire bundle: scripts carry inlined source (read from

crates/picloud-cli/src/linkstate.rs

@@ -13,6 +13,71 @@ use serde::{Deserialize, Serialize};
 
 const DIR: &str = ".picloud";
 const PLAN_FILE: &str = "plan.json";
+const PROJECT_FILE: &str = "project.json";
+
+/// The repo's stable project identity (§7, M3): a random, opaque key minted on
+/// first need and persisted (gitignored) in `.picloud/`. It is the server-side
+/// ownership handle — the same repo keeps the same project across clones/CI
+/// because the key travels in the working tree, never in a committed file.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ProjectLink {
+    pub key: String,
+}
+
+fn project_path(base: &Path) -> PathBuf {
+    base.join(DIR).join(PROJECT_FILE)
+}
+
+/// Read the project key, if one has been minted under `base/.picloud/`.
+#[must_use]
+pub fn read_project_key(base: &Path) -> Option<String> {
+    let body = fs::read(project_path(base)).ok()?;
+    serde_json::from_slice::<ProjectLink>(&body)
+        .ok()
+        .map(|l| l.key)
+}
+
+/// Read the project key, minting and persisting a fresh one if absent. Used by
+/// the tree `plan`/`apply` paths so a repo that never ran `pic init` still gets
+/// a stable ownership identity on first use. The new key is a random v4 UUID.
+pub fn ensure_project_key(base: &Path) -> Result<String> {
+    if let Some(k) = read_project_key(base) {
+        return Ok(k);
+    }
+    // A present-but-unreadable/corrupt `project.json` must NOT be silently
+    // re-minted: a fresh key would orphan every group this repo already claimed
+    // (the old key is lost, so the next apply hits an ownership conflict). Fail
+    // loudly and let the operator restore or intentionally re-key it.
+    let path = project_path(base);
+    if path.exists() {
+        anyhow::bail!(
+            "{} exists but could not be parsed as a project key; refusing to mint a new one \
+             (a new key would orphan this repo's group ownership). Restore or delete it.",
+            path.display()
+        );
+    }
+    let key = uuid::Uuid::new_v4().to_string();
+    write_project_key(base, &key)?;
+    Ok(key)
+}
+
+/// Persist `key` as the project identity, creating `.picloud/` (self-ignored)
+/// if needed. Idempotent overwrite — callers usually go through
+/// [`ensure_project_key`].
+pub fn write_project_key(base: &Path, key: &str) -> Result<()> {
+    let dir = base.join(DIR);
+    fs::create_dir_all(&dir).with_context(|| format!("creating {}", dir.display()))?;
+    let ignore = dir.join(".gitignore");
+    if !ignore.exists() {
+        fs::write(&ignore, "*\n").context("writing .picloud/.gitignore")?;
+    }
+    let body = serde_json::to_vec_pretty(&ProjectLink {
+        key: key.to_string(),
+    })
+    .context("encoding .picloud/project.json")?;
+    fs::write(project_path(base), body).context("writing .picloud/project.json")?;
+    Ok(())
+}
 
 /// The recorded result of the last `pic plan`, scoped to the app it was for.
 #[derive(Debug, Clone, Serialize, Deserialize)]

crates/picloud-cli/src/main.rs

@@ -255,6 +255,11 @@ struct ApplyArgs {
     /// policy, §4.2). Repeatable. A blanket `--yes` does NOT cover it.
     #[arg(long = "approve", requires = "dir")]
     approve: Vec<String>,
+    /// Tree apply only (`--dir`): claim declared groups another project owns
+    /// (§7, M3). Group-admin gated. Without it, an owned-elsewhere group is a
+    /// hard conflict.
+    #[arg(long, requires = "dir")]
+    takeover: bool,
 }
 
 #[derive(Args)]
@@ -1428,6 +1433,7 @@ async fn main() -> ExitCode {
                     args.force,
                     args.env.as_deref(),
                     &args.approve,
+                    args.takeover,
                     mode,
                 )
                 .await

crates/picloud-cli/tests/cli.rs

@@ -36,6 +36,7 @@ mod init;
 mod invoke;
 mod logs;
 mod output;
+mod ownership;
 mod plan;
 mod prune;
 mod pull;

crates/picloud-cli/tests/ownership.rs

@@ -0,0 +1,288 @@
+//! M3 multi-repo single-owner ownership (§7) via `pic apply --dir`.
+//!
+//! On this build groups pre-exist (created via `pic groups create`; the tree
+//! apply reconciles a node's CONTENT and CLAIMS its ownership, it does not
+//! create the group). So each test first creates the group(s), then:
+//!   * the first repo to `apply` a group CLAIMS it (its `.picloud/` project key
+//!     becomes the group's `owner_project`),
+//!   * a SECOND repo (a distinct dir → a distinct project key) applying the
+//!     same group is REJECTED with an ownership conflict,
+//!   * `--takeover` lets the second repo seize it — admin-gated, so the fixture
+//!     instance-owner succeeds, flipping ownership (proven by the first repo now
+//!     being the one rejected), but a group `editor` is 403'd (ownership ⟂ RBAC,
+//!     §7.4),
+//!   * `apply --prune` deletes an owned, now-undeclared, empty group; a group
+//!     this project never claimed is never touched.
+//!
+//! Each project lives in its own `TempDir`, so `pic` mints a distinct project
+//! key per repo (`.picloud/project.json`) — the two-repo topology §7 governs.
+
+use std::fs;
+use std::path::Path;
+
+use tempfile::TempDir;
+
+use crate::common;
+use crate::common::cleanup::GroupGuard;
+use crate::common::member;
+
+/// A single-group project dir declaring the (pre-existing) group `slug`. No
+/// scripts, so a structural prune can delete it (a group holding scripts/apps
+/// is RESTRICT-pinned). `extra` is appended to the `[group]` table verbatim.
+fn group_dir(slug: &str, extra: &str) -> TempDir {
+    let dir = TempDir::new().expect("tempdir");
+    fs::write(
+        dir.path().join("picloud.toml"),
+        format!("[group]\nslug = \"{slug}\"\nname = \"Owned Group\"\n{extra}"),
+    )
+    .unwrap();
+    dir
+}
+
+fn apply(env: &common::TestEnv, dir: &Path, extra: &[&str]) -> std::process::Output {
+    let mut cmd = common::pic_as(env);
+    cmd.args(["apply", "--dir"]).arg(dir).args(extra);
+    cmd.output().expect("apply --dir")
+}
+
+fn create_group(env: &common::TestEnv, slug: &str, parent: Option<&str>) {
+    let mut cmd = common::pic_as(env);
+    cmd.args(["groups", "create", slug]);
+    if let Some(p) = parent {
+        cmd.args(["--parent", p]);
+    }
+    let out = cmd.output().expect("groups create");
+    assert!(
+        out.status.success(),
+        "groups create {slug} failed: {}",
+        String::from_utf8_lossy(&out.stderr)
+    );
+}
+
+fn ls_groups(env: &common::TestEnv) -> String {
+    String::from_utf8(
+        common::pic_as(env)
+            .args(["groups", "ls"])
+            .output()
+            .expect("groups ls")
+            .stdout,
+    )
+    .unwrap()
+}
+
+#[ignore = "needs DATABASE_URL pointing at a running Postgres"]
+#[test]
+fn first_repo_claims_second_is_rejected_then_takeover_flips_ownership() {
+    let Some(fx) = common::fixture_or_skip() else {
+        return;
+    };
+    let env = common::admin_env(fx);
+    let slug = common::unique_slug("own-g");
+    let _g = GroupGuard::new(&env.url, &env.token, &slug);
+    create_group(&env, &slug, None);
+
+    // Repo A claims the group on first apply (owner NULL → project A).
+    let repo_a = group_dir(&slug, "");
+    let a1 = apply(&env, repo_a.path(), &[]);
+    assert!(
+        a1.status.success(),
+        "repo A claim apply failed: {}",
+        String::from_utf8_lossy(&a1.stderr)
+    );
+
+    // Repo B (a different dir → a different project key) is rejected: the group
+    // is owned by project A.
+    let repo_b = group_dir(&slug, "");
+    let b1 = apply(&env, repo_b.path(), &[]);
+    assert!(
+        !b1.status.success(),
+        "repo B apply must be rejected (group owned by A)"
+    );
+    let b1_err = String::from_utf8_lossy(&b1.stderr).to_lowercase();
+    assert!(
+        b1_err.contains("owned by another project") || b1_err.contains("takeover"),
+        "conflict message should name the ownership clash:\n{b1_err}"
+    );
+
+    // Repo B with --takeover succeeds (the fixture token is instance owner, so
+    // it holds GroupAdmin on every node) and flips ownership to project B.
+    let b2 = apply(&env, repo_b.path(), &["--takeover"]);
+    assert!(
+        b2.status.success(),
+        "repo B --takeover should succeed: {}",
+        String::from_utf8_lossy(&b2.stderr)
+    );
+
+    // Proof the flip took: repo A — formerly the owner — is now the one
+    // rejected without --takeover.
+    let a2 = apply(&env, repo_a.path(), &[]);
+    assert!(
+        !a2.status.success(),
+        "after takeover, repo A must now be rejected (B owns the group)"
+    );
+}
+
+#[ignore = "needs DATABASE_URL pointing at a running Postgres"]
+#[test]
+fn takeover_without_group_admin_is_forbidden() {
+    // §7.4 — ownership ⟂ RBAC: `--takeover` needs GroupAdmin specifically, a
+    // second gate beyond the write caps. A group `editor` (who CAN write group
+    // vars) still cannot seize the group for their repo.
+    let Some(fx) = common::fixture_or_skip() else {
+        return;
+    };
+    let env = common::admin_env(fx);
+    let slug = common::unique_slug("own-ta");
+    let _g = GroupGuard::new(&env.url, &env.token, &slug);
+    create_group(&env, &slug, None);
+
+    // Admin (repo A) claims the group.
+    let repo_a = group_dir(&slug, "");
+    assert!(
+        apply(&env, repo_a.path(), &[]).status.success(),
+        "admin claim apply should succeed"
+    );
+
+    // A member granted `editor` (not app_admin/group_admin) on the group.
+    let m = member::member_user(fx, &common::unique_username("ta"));
+    common::pic_as(&env)
+        .args(["groups", "members", "add", &slug, &m.id, "--role", "editor"])
+        .output()
+        .expect("add group member");
+    let member_env = common::custom_env(&fx.url, &m.token);
+    common::seed_credentials(&member_env, &m.username);
+
+    // The member's repo B (a distinct project key) declares a group var (so its
+    // write cap is satisfied) and attempts a takeover → 403 at the GroupAdmin
+    // gate, before any write lands.
+    let repo_b = group_dir(&slug, "\n[vars]\nregion = \"eu\"\n");
+    let out = apply(&member_env, repo_b.path(), &["--takeover"]);
+    assert!(
+        !out.status.success(),
+        "non-admin --takeover must be forbidden"
+    );
+    let err = String::from_utf8_lossy(&out.stderr).to_lowercase();
+    assert!(
+        err.contains("forbidden") || err.contains("403"),
+        "takeover denial should be an authz error:\n{err}"
+    );
+}
+
+#[ignore = "needs DATABASE_URL pointing at a running Postgres"]
+#[test]
+fn prune_deletes_owned_undeclared_group_only() {
+    let Some(fx) = common::fixture_or_skip() else {
+        return;
+    };
+    let env = common::admin_env(fx);
+    let parent = common::unique_slug("own-par");
+    let child = common::unique_slug("own-child");
+    // Drop order: child (registered last → dropped first), then parent.
+    let _gp = GroupGuard::new(&env.url, &env.token, &parent);
+    let _gc = GroupGuard::new(&env.url, &env.token, &child);
+    create_group(&env, &parent, None);
+    create_group(&env, &child, Some(&parent));
+
+    // Repo declares parent (root manifest) + child (a nested dir); apply claims
+    // both for this project.
+    let dir = TempDir::new().expect("tempdir");
+    fs::write(
+        dir.path().join("picloud.toml"),
+        format!("[group]\nslug = \"{parent}\"\nname = \"Parent\"\n"),
+    )
+    .unwrap();
+    fs::create_dir_all(dir.path().join("sub")).unwrap();
+    fs::write(
+        dir.path().join("sub/picloud.toml"),
+        format!("[group]\nslug = \"{child}\"\nname = \"Child\"\n"),
+    )
+    .unwrap();
+    let out = apply(&env, dir.path(), &[]);
+    assert!(
+        out.status.success(),
+        "initial claim apply failed: {}",
+        String::from_utf8_lossy(&out.stderr)
+    );
+    let groups = ls_groups(&env);
+    assert!(
+        groups.contains(&parent) && groups.contains(&child),
+        "both groups should exist after claim"
+    );
+
+    // Drop the child manifest from the SAME repo (keeping its project key) so
+    // the child becomes owned-but-undeclared, then apply --prune: the empty
+    // child is deleted; the parent (still declared) is kept.
+    fs::remove_dir_all(dir.path().join("sub")).unwrap();
+    let out2 = apply(&env, dir.path(), &["--prune", "--yes"]);
+    assert!(
+        out2.status.success(),
+        "prune apply failed: {}",
+        String::from_utf8_lossy(&out2.stderr)
+    );
+    let groups = ls_groups(&env);
+    assert!(groups.contains(&parent), "declared parent must be kept");
+    assert!(
+        !groups.contains(&child),
+        "owned-but-undeclared child must be pruned"
+    );
+}
+
+#[ignore = "needs DATABASE_URL pointing at a running Postgres"]
+#[test]
+fn prune_refuses_to_cascade_a_groups_shared_data() {
+    // §7 M3 data-loss guard: an owned-but-undeclared group is NOT structurally
+    // pruned if it holds §11.6 shared collections/secrets (those CASCADE) — it
+    // is kept with a warning instead of being silently vaporized.
+    let Some(fx) = common::fixture_or_skip() else {
+        return;
+    };
+    let env = common::admin_env(fx);
+    let parent = common::unique_slug("own-dpar");
+    let child = common::unique_slug("own-dchild");
+    let _gp = GroupGuard::new(&env.url, &env.token, &parent);
+    let _gc = GroupGuard::new(&env.url, &env.token, &child);
+    create_group(&env, &parent, None);
+    create_group(&env, &child, Some(&parent));
+
+    // Declare parent + child; the child declares a shared KV collection, so the
+    // first apply claims both AND creates a group-owned collection marker.
+    let dir = TempDir::new().expect("tempdir");
+    fs::write(
+        dir.path().join("picloud.toml"),
+        format!("[group]\nslug = \"{parent}\"\nname = \"Parent\"\n"),
+    )
+    .unwrap();
+    fs::create_dir_all(dir.path().join("sub")).unwrap();
+    fs::write(
+        dir.path().join("sub/picloud.toml"),
+        format!("[group]\nslug = \"{child}\"\nname = \"Child\"\n\ncollections = [\"catalog\"]\n"),
+    )
+    .unwrap();
+    assert!(
+        apply(&env, dir.path(), &[]).status.success(),
+        "initial claim+collection apply should succeed"
+    );
+
+    // Drop the child node, then prune: the child owns a shared collection, so it
+    // must be KEPT (with a warning), not cascaded away.
+    fs::remove_dir_all(dir.path().join("sub")).unwrap();
+    let out = apply(&env, dir.path(), &["--prune", "--yes"]);
+    assert!(
+        out.status.success(),
+        "prune apply should succeed (skipping the protected group): {}",
+        String::from_utf8_lossy(&out.stderr)
+    );
+    let combined = format!(
+        "{}{}",
+        String::from_utf8_lossy(&out.stdout),
+        String::from_utf8_lossy(&out.stderr)
+    );
+    assert!(
+        combined.contains("shared collections"),
+        "prune should warn it skipped a group owning shared data:\n{combined}"
+    );
+    assert!(
+        ls_groups(&env).contains(&child),
+        "a group owning shared collections must NOT be pruned"
+    );
+}

docs/design/groups-and-project-tool.md

@@ -1120,11 +1120,25 @@ Resolved items now live inline next to their topic. What genuinely remains:
 5. **Project tool maps onto groups.** Nested manifests, attach point, single-owner, server-computed
    tree plan, per-env approval gating.
 
-   > **Status (Phase 5): ✅ shipped — single-repo nested tree apply, atomic.** A directory tree of
+   > **Status (Phase 5): ✅ shipped — nested tree apply, atomic, with multi-repo single-owner
-   > `picloud.toml` manifests (each declaring an `[app]` or `[group]` node) applies as ONE
+   > ownership.** A directory tree of `picloud.toml` manifests (each declaring an `[app]` or `[group]`
-   > server-computed plan in ONE Postgres transaction. Per the §11.1 review, the **multi-repo
+   > node) applies as ONE server-computed plan in ONE Postgres transaction. **Multi-repo single-owner
-   > single-owner / attach-point / takeover** layer (§7) is **deferred** for the solo-dev / single-repo
+   > ownership + takeover shipped (§7, M3, 2026-07-05)** — each repo mints a stable, gitignored project
-   > start. **Per-env approval-policy gating shipped (`[project.environments]`, M5, 2026-07-05)** — the
+   > key in `.picloud/project.json`; the first `pic apply --dir` to touch a group **claims** it
+   > (`groups.owner_project` FK → the new `projects` table, migration 0066 promoting the inert 0047
+   > column). A second repo (a distinct key) applying an owned group is **refused** (`owned by another
+   > project; use --takeover`) unless it passes `--takeover`, which is **GroupAdmin-gated per contested
+   > node** — re-verified **in-tx** at the ownership decision (not only in the pre-tx pool read), so
+   > `--force` (which waives the staleness token) can't open a takeover-without-admin window. Ownership ⟂
+   > RBAC: owning the manifest never grants write — the actor still needs the usual group caps. The owner
+   > key folds into the bound-plan token (a claim/takeover by another repo between plan and apply trips
+   > `StateMoved`), and the attacker-supplied key is length/charset-validated server-side. `apply --prune`
+   > **structurally reaps** groups THIS project owns that the manifest no longer declares, leaf-first
+   > (`delete_group_tx` RESTRICT — a group still holding apps/subgroups aborts the apply, never orphans),
+   > and never touches another repo's or an unclaimed (UI-owned) group. `pic plan --dir` surfaces
+   > conflicts + prune candidates read-only. **Deferred:** the declarative **attach-point** / a single
+   > repo *creating* the group tree (groups still pre-exist — a manifest owns each node's *content*, not
+   > the tree *shape*). **Per-env approval-policy gating shipped (`[project.environments]`, M5, 2026-07-05)** — the
    > root manifest declares which environments are confirm-required; `pic apply --dir --env <e>` to such
    > an env needs an explicit `--approve <e>` (a blanket `--yes` does NOT cover it), the approval is
    > admin-gated (the actor needs admin on every declared node, not just write) + audited, and the policy
@@ -1249,10 +1263,12 @@ to take, not yet taken:
   - *Trigger/route templates (§4.5):* bundled into Phase 6 as "much later," but the per-app binding
     tax bites in Phase 5 at high tenant cardinality (100 tenants × 5 triggers = 500 declarations).
     Decide against **actual** target tenant counts before defaulting it late.
-  - *Multi-repo ownership (§7):* the single-owner / attach-point / takeover machinery is substantial
+  - *Multi-repo ownership (§7):* the single-owner / takeover machinery is substantial and only earns
-    and only earns its keep with a **second** managing repo. For a solo-dev / single-repo start, a
+    its keep with a **second** managing repo. For a solo-dev / single-repo start, a "one repo owns the
-    "one repo owns the whole subtree" model covers the near term; confirm the multi-repo need exists
+    whole subtree" model covers the near term. **Resolved: shipped as M3 (2026-07-05)** — single-owner
-    before building the ownership layer.
+    claim / conflict / `--takeover` (GroupAdmin-gated) + structural prune, over a `projects` table and
+    the promoted `groups.owner_project` FK. The declarative **attach-point** (a repo *creating* the
+    group tree, not just claiming pre-existing groups) remains deferred.
 
 ---