code-context-control 2.57.1__tar.gz → 2.59.0__tar.gz

{code_context_control-2.57.1 → code_context_control-2.59.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: code-context-control
-Version: 2.57.1
+Version: 2.59.0
 Summary: Local code-intelligence layer for AI coding tools (Claude Code, Codex, Copilot, Cursor, Antigravity). Retrieve less, read less, edit safer — version the configs that shape your agent (CLAUDE.md, skills, hooks, MCP), and manage multi-project + sub-project hierarchies from a local hub.
 Author-email: Dimitri Tselenchuk <dtselenc@gmail.com>
 License-Expression: Apache-2.0
@@ -283,6 +283,7 @@ C3 exposes 18 tools as a native MCP server. Your IDE calls them directly:
 | `c3_edits` | Edit-ledger queries + version diffs + restore points + per-branch filter |
 | `c3_bitbucket` | Bitbucket Data Center integration — PRs, branches, builds, repo admin (v2.30.0) |
 | `c3_jira` | Jira integration — Cloud + Data Center: JQL search, issues, transitions, My Work board (v2.56.0) |
+| `c3_credentials` | Credential vault — named secrets (global + per-project), injection-first: agents use them by name, values never enter model context (v2.58.0) |
 | `c3_project` | Cross-project — discover & operate on other c3-installed projects; guarded writes (v2.31.0) |
 | `c3_task` | Durable per-project PM — tasks with dependencies & subtasks, milestones, decision notes, event history, health reports, and auto+manual time tracking (v2.53.0) |
 | `c3_artifacts` | Agent-config tracking — version history, diff & restore for CLAUDE.md, settings/hooks, MCP configs, skills (v2.46.0) |
@@ -291,7 +292,7 @@ On Windows, `c3_shell` uses Git Bash when available. Git Bash does not bundle
 optional utilities such as `jq`; use `python -m json.tool` for portable JSON
 formatting, or install `jq` separately when filter expressions are required.
 
-Every tool is **read-only safe in plan mode** (except `c3_edit`, `c3_shell`, `c3_artifacts(action='restore')`, and write actions on `c3_bitbucket` / `c3_jira` / `c3_project` / `c3_task`).
+Every tool is **read-only safe in plan mode** (except `c3_edit`, `c3_shell`, `c3_artifacts(action='restore')`, and write actions on `c3_bitbucket` / `c3_jira` / `c3_credentials` / `c3_project` / `c3_task`).
 
 ### Bitbucket Data Center / Server (v2.30.0)
 
@@ -319,6 +320,52 @@ it has no active account C3 falls back to the global `~/.c3/config.json`. So a
 single `login --global` (or any login done from your home directory) is reusable
 across every C3 project — the PAT always lives in the OS keyring, never on disk.
 
+### Credential vault (v2.58.0)
+
+`c3_credentials` gives agents a protected, user-managed place for API keys,
+tokens, and `.env`-style values — **global** (`~/.c3`, every project) or
+**per-project** (`.c3`, shadows the global name). Values live in the **OS
+keyring** (large values in a Fernet-encrypted `.c3/secrets.enc` whose master
+key lives in the keyring) — never in config files, and *never in the model's
+context*: the agent addresses secrets by name and C3 decodes them only at the
+subprocess boundary.
+
+```bash
+# Store a secret for this project (value prompted, masked)
+c3 creds set OPENAI_KEY --desc "OpenAI billing key"
+
+# ...or globally for every C3 project
+c3 creds set NPM_TOKEN --global
+
+# Bulk-import an existing .env; list what the agent can see
+c3 creds import .env
+c3 creds list
+```
+
+The agent then runs commands *with* the secret but without ever seeing it:
+
+```
+c3_shell(cmd='npm publish', env_creds='NPM_TOKEN')       # injected as env var
+c3_shell(cmd='curl -H "Authorization: Bearer {{cred:OPENAI_KEY}}" …')  # expanded server-side
+```
+
+Echoed values are auto-redacted from output (`env` dumps come back as
+`[cred:NAME]`), every use is ledger-logged by name, and `reveal` — the only
+action that returns a value — is disabled per entry until you flip
+`agent_readable` in the **Credentials UI tab** or via
+`c3 creds set NAME --agent-readable`. A hostile repo config can't siphon your
+global secrets (realm-atomic resolution, tested), cross-project shells run
+with credentials disabled, and the vault is hard-excluded from the Oracle
+Discovery API.
+
+Since **v2.59.0** the Hub has a top-level **Credentials** view: manage the
+global vault (`~/.c3`) and every registered project's entries from one place —
+create / edit / delete / `.env` import / flag toggles, with shadowing shown
+both ways ("shadows global" on project entries, "shadowed in N projects" on
+globals). The write-only wire contract extends to the hub: values are
+submitted inbound-only and **no hub route ever returns a stored value** (there
+is no `reveal` on the hub at all).
+
 ### Jira — Cloud + Data Center (v2.56.0)
 
 `c3_jira` connects to Jira Cloud (REST v3, email + API token) or self-hosted

{code_context_control-2.57.1 → code_context_control-2.59.0}/README.md

@@ -220,6 +220,7 @@ C3 exposes 18 tools as a native MCP server. Your IDE calls them directly:
 | `c3_edits` | Edit-ledger queries + version diffs + restore points + per-branch filter |
 | `c3_bitbucket` | Bitbucket Data Center integration — PRs, branches, builds, repo admin (v2.30.0) |
 | `c3_jira` | Jira integration — Cloud + Data Center: JQL search, issues, transitions, My Work board (v2.56.0) |
+| `c3_credentials` | Credential vault — named secrets (global + per-project), injection-first: agents use them by name, values never enter model context (v2.58.0) |
 | `c3_project` | Cross-project — discover & operate on other c3-installed projects; guarded writes (v2.31.0) |
 | `c3_task` | Durable per-project PM — tasks with dependencies & subtasks, milestones, decision notes, event history, health reports, and auto+manual time tracking (v2.53.0) |
 | `c3_artifacts` | Agent-config tracking — version history, diff & restore for CLAUDE.md, settings/hooks, MCP configs, skills (v2.46.0) |
@@ -228,7 +229,7 @@ On Windows, `c3_shell` uses Git Bash when available. Git Bash does not bundle
 optional utilities such as `jq`; use `python -m json.tool` for portable JSON
 formatting, or install `jq` separately when filter expressions are required.
 
-Every tool is **read-only safe in plan mode** (except `c3_edit`, `c3_shell`, `c3_artifacts(action='restore')`, and write actions on `c3_bitbucket` / `c3_jira` / `c3_project` / `c3_task`).
+Every tool is **read-only safe in plan mode** (except `c3_edit`, `c3_shell`, `c3_artifacts(action='restore')`, and write actions on `c3_bitbucket` / `c3_jira` / `c3_credentials` / `c3_project` / `c3_task`).
 
 ### Bitbucket Data Center / Server (v2.30.0)
 
@@ -256,6 +257,52 @@ it has no active account C3 falls back to the global `~/.c3/config.json`. So a
 single `login --global` (or any login done from your home directory) is reusable
 across every C3 project — the PAT always lives in the OS keyring, never on disk.
 
+### Credential vault (v2.58.0)
+
+`c3_credentials` gives agents a protected, user-managed place for API keys,
+tokens, and `.env`-style values — **global** (`~/.c3`, every project) or
+**per-project** (`.c3`, shadows the global name). Values live in the **OS
+keyring** (large values in a Fernet-encrypted `.c3/secrets.enc` whose master
+key lives in the keyring) — never in config files, and *never in the model's
+context*: the agent addresses secrets by name and C3 decodes them only at the
+subprocess boundary.
+
+```bash
+# Store a secret for this project (value prompted, masked)
+c3 creds set OPENAI_KEY --desc "OpenAI billing key"
+
+# ...or globally for every C3 project
+c3 creds set NPM_TOKEN --global
+
+# Bulk-import an existing .env; list what the agent can see
+c3 creds import .env
+c3 creds list
+```
+
+The agent then runs commands *with* the secret but without ever seeing it:
+
+```
+c3_shell(cmd='npm publish', env_creds='NPM_TOKEN')       # injected as env var
+c3_shell(cmd='curl -H "Authorization: Bearer {{cred:OPENAI_KEY}}" …')  # expanded server-side
+```
+
+Echoed values are auto-redacted from output (`env` dumps come back as
+`[cred:NAME]`), every use is ledger-logged by name, and `reveal` — the only
+action that returns a value — is disabled per entry until you flip
+`agent_readable` in the **Credentials UI tab** or via
+`c3 creds set NAME --agent-readable`. A hostile repo config can't siphon your
+global secrets (realm-atomic resolution, tested), cross-project shells run
+with credentials disabled, and the vault is hard-excluded from the Oracle
+Discovery API.
+
+Since **v2.59.0** the Hub has a top-level **Credentials** view: manage the
+global vault (`~/.c3`) and every registered project's entries from one place —
+create / edit / delete / `.env` import / flag toggles, with shadowing shown
+both ways ("shadows global" on project entries, "shadowed in N projects" on
+globals). The write-only wire contract extends to the hub: values are
+submitted inbound-only and **no hub route ever returns a stored value** (there
+is no `reveal` on the hub at all).
+
 ### Jira — Cloud + Data Center (v2.56.0)
 
 `c3_jira` connects to Jira Cloud (REST v3, email + API token) or self-hosted

{code_context_control-2.57.1 → code_context_control-2.59.0}/cli/c3.py

@@ -92,7 +92,7 @@ console = Console() if HAS_RICH else None
 # Config
 CONFIG_DIR = ".c3"
 CONFIG_FILE = ".c3/config.json"
-__version__ = "2.57.1"
+__version__ = "2.59.0"
 
 
 def _command_deps() -> CommandDeps:
@@ -248,7 +248,7 @@ _C3_MCP_ALLOW = [
     "mcp__c3__c3_memory", "mcp__c3__c3_validate", "mcp__c3__c3_edit",
     "mcp__c3__c3_agent", "mcp__c3__c3_delegate", "mcp__c3__c3_edits",
     "mcp__c3__c3_impact", "mcp__c3__c3_shell", "mcp__c3__c3_bitbucket",
-    "mcp__c3__c3_jira",
+    "mcp__c3__c3_jira", "mcp__c3__c3_credentials",
     "mcp__c3__c3_project", "mcp__c3__c3_task", "mcp__c3__c3_artifacts",
 ]
 
@@ -5738,6 +5738,150 @@ def _bb_cmd_set_default(args, project_path: str) -> None:
     print(f"[OK] Default repo: {args.project}/{args.repo}")
 
 
+def cmd_creds(args):
+    """Credential vault management (global + per-project scopes)."""
+    sub = getattr(args, "creds_cmd", None)
+    if not sub:
+        print("Usage: c3 creds {set,get,list,rm,import} [args]")
+        return
+
+    project_path = getattr(args, "project_path", ".") or "."
+
+    if sub == "set":
+        _creds_cmd_set(args, project_path)
+    elif sub == "get":
+        _creds_cmd_get(args, project_path)
+    elif sub == "list":
+        _creds_cmd_list(args, project_path)
+    elif sub == "rm":
+        _creds_cmd_rm(args, project_path)
+    elif sub == "import":
+        _creds_cmd_import(args, project_path)
+    else:
+        print(f"Unknown creds subcommand: {sub}")
+
+
+def _creds_scope(args) -> str:
+    return "global" if getattr(args, "use_global", False) else "project"
+
+
+def _creds_entry_line(name: str, entry: dict) -> str:
+    flags = [f for f in ("inject", "agent_readable") if entry.get(f)]
+    parts = [
+        f"{name:<24} {entry.get('scope', '?'):<8} {entry.get('type', 'token'):<10}"
+        f" len={entry.get('value_len', '?')}",
+    ]
+    if entry.get("env_var"):
+        parts.append(f"env_var={entry['env_var']}")
+    if flags:
+        parts.append("[" + ",".join(flags) + "]")
+    if entry.get("description"):
+        parts.append(f"-- {entry['description']}")
+    return "  ".join(parts)
+
+
+def _creds_cmd_set(args, project_path: str) -> None:
+    import getpass
+
+    from services import credential_store as cred_store
+
+    value = getattr(args, "value", "") or ""
+    if getattr(args, "stdin", False):
+        value = sys.stdin.read().rstrip("\n")
+    if not value:
+        value = getpass.getpass(f"Value for {args.name}: ")
+    if not value:
+        print("Cancelled -- value required.")
+        return
+    scope = _creds_scope(args)
+    try:
+        entry = cred_store.set_credential(
+            args.name, value,
+            scope=scope, project_path=project_path,
+            description=getattr(args, "desc", "") or "",
+            ctype=getattr(args, "ctype", "token") or "token",
+            env_var=getattr(args, "env_var", "") or "",
+            agent_readable=bool(getattr(args, "agent_readable", False)),
+            inject=bool(getattr(args, "inject", False)),
+        )
+    except (cred_store.CredentialError, RuntimeError) as exc:
+        print(f"[error] {exc}")
+        return
+    print(f"[OK] Stored credential '{args.name}' "
+          f"(scope={scope}, storage={entry['storage']}, len={entry['value_len']})")
+    if entry["agent_readable"]:
+        print("[warn] agent_readable=true -- the agent can read this value "
+              "into its context and transcripts.")
+
+
+def _creds_cmd_get(args, project_path: str) -> None:
+    from services import credential_store as cred_store
+
+    entry = cred_store.get_entry(args.name, project_path=project_path)
+    if not entry:
+        print(f"[error] no credential named '{args.name}'")
+        return
+    print(_creds_entry_line(args.name, entry))
+    print(f"storage={entry.get('storage', 'keyring')}  "
+          f"created={entry.get('created', '?')}  updated={entry.get('updated', '?')}")
+    fp = cred_store.fingerprint(args.name, project_path=project_path)
+    print(f"fingerprint={fp or 'unresolvable'}")
+    if getattr(args, "show", False):
+        value = cred_store.get_value(args.name, project_path=project_path)
+        print(value if value is not None else "[error] value missing from store")
+
+
+def _creds_cmd_list(args, project_path: str) -> None:
+    from services import credential_store as cred_store
+
+    entries = cred_store.list_entries(project_path)
+    if not entries:
+        print("No credentials registered. Use `c3 creds set NAME` "
+              "(add --global for all projects).")
+        return
+    print(f"{len(entries)} credential(s) — project scope shadows global:")
+    for name, entry in entries.items():
+        print("  " + _creds_entry_line(name, entry))
+
+
+def _creds_cmd_rm(args, project_path: str) -> None:
+    from services import credential_store as cred_store
+
+    scope = _creds_scope(args)
+    entry = cred_store.get_entry(args.name, project_path=project_path)
+    if entry and entry["scope"] == "global" and scope == "project":
+        print(f"[error] '{args.name}' is a global credential -- "
+              "re-run with --global to delete it.")
+        return
+    if cred_store.delete_credential(args.name, scope=scope, project_path=project_path):
+        print(f"[OK] Removed credential '{args.name}' (scope={scope})")
+    else:
+        print(f"[error] no credential named '{args.name}' in {scope} scope")
+
+
+def _creds_cmd_import(args, project_path: str) -> None:
+    from services import credential_store as cred_store
+
+    env_path = Path(getattr(args, "env_file", ""))
+    if not env_path.exists():
+        print(f"[error] file not found: {env_path}")
+        return
+    try:
+        text = env_path.read_text(encoding="utf-8")
+        result = cred_store.import_env(
+            text, scope=_creds_scope(args), project_path=project_path,
+            overwrite=bool(getattr(args, "overwrite", False)),
+        )
+    except (cred_store.CredentialError, RuntimeError) as exc:
+        print(f"[error] {exc}")
+        return
+    print(f"[OK] Imported {len(result['created'])} credential(s): "
+          f"{', '.join(result['created']) or '-'}")
+    if result["skipped"]:
+        print(f"Skipped {len(result['skipped'])}: {', '.join(result['skipped'])} "
+              "(use --overwrite to replace)")
+
+
 def cmd_jira(args):
     """Jira Cloud / Data Center credential + workspace management."""
     sub = getattr(args, "jira_cmd", None)
@@ -7036,6 +7180,7 @@ def main():
         "hub": cmd_hub,
         "bitbucket": cmd_bitbucket,
         "jira": cmd_jira,
+        "creds": cmd_creds,
         "oracle": cmd_oracle,
         "upgrade": cmd_upgrade,
     }

{code_context_control-2.57.1 → code_context_control-2.59.0}/cli/commands/parser.py

@@ -401,6 +401,48 @@ def build_parser(version: str, parse_cli_ide_arg):
     jr_default.add_argument("--name", default="", help="Account name (defaults to the default account)")
     jr_default.add_argument("project_path", nargs="?", default=".")
 
+    # ── Credential vault (v2.58.0) ──────────────────────────────────────
+    p_creds = subparsers.add_parser(
+        "creds",
+        help="Credential vault — named secrets for agents (global + per-project)",
+    )
+    creds_subs = p_creds.add_subparsers(dest="creds_cmd")
+
+    cr_set = creds_subs.add_parser(
+        "set", help="Create or update a credential (value via hidden prompt)"
+    )
+    cr_set.add_argument("name", help="Entry name (env-var safe: [A-Za-z_][A-Za-z0-9_]*)")
+    cr_set.add_argument("--value", default="", help="Secret value (prompted via getpass if omitted — preferred)")
+    cr_set.add_argument("--stdin", action="store_true", help="Read the value from stdin (piped/multiline values)")
+    cr_set.add_argument("--type", dest="ctype", choices=["token", "env", "multiline"], default="token", help="Entry type")
+    cr_set.add_argument("--desc", default="", help="Human description shown in list/UI")
+    cr_set.add_argument("--env-var", default="", help="Env var name used at injection (default: entry name)")
+    cr_set.add_argument("--agent-readable", action="store_true", help="Allow the agent to reveal the decoded value into its context (default: injection-only)")
+    cr_set.add_argument("--inject", action="store_true", help="Auto-inject into every c3_shell run")
+    cr_set.add_argument("--global", dest="use_global", action="store_true", help="Store in the global scope (~/.c3) so every C3 project can use it")
+    # --path (not a trailing positional): a second positional after options
+    # breaks argparse on py<3.12 ("unrecognized arguments").
+    cr_set.add_argument("--path", dest="project_path", default=".", help="Project directory (default: current)")
+
+    cr_get = creds_subs.add_parser("get", help="Show entry metadata (masked; --show prints the value)")
+    cr_get.add_argument("name")
+    cr_get.add_argument("--show", action="store_true", help="Print the decoded value to the terminal")
+    cr_get.add_argument("--path", dest="project_path", default=".", help="Project directory (default: current)")
+
+    cr_list = creds_subs.add_parser("list", help="List credentials visible to this project")
+    cr_list.add_argument("--path", dest="project_path", default=".", help="Project directory (default: current)")
+
+    cr_rm = creds_subs.add_parser("rm", help="Delete a credential (value + registry entry)")
+    cr_rm.add_argument("name")
+    cr_rm.add_argument("--global", dest="use_global", action="store_true", help="Delete from the global scope")
+    cr_rm.add_argument("--path", dest="project_path", default=".", help="Project directory (default: current)")
+
+    cr_import = creds_subs.add_parser("import", help="Import KEY=VALUE lines from a .env file")
+    cr_import.add_argument("env_file", help="Path to the .env file")
+    cr_import.add_argument("--global", dest="use_global", action="store_true", help="Import into the global scope")
+    cr_import.add_argument("--overwrite", action="store_true", help="Replace entries already registered in the target scope")
+    cr_import.add_argument("--path", dest="project_path", default=".", help="Project directory (default: current)")
+
     # ── Oracle Discovery API (v2.32.0) ──────────────────────────────────
     p_oracle = subparsers.add_parser(
         "oracle",

{code_context_control-2.57.1 → code_context_control-2.59.0}/cli/guide/tools.html

@@ -125,6 +125,7 @@
       <div class="sidebar-label">SCM</div>
       <a href="#c3_task" class="sidebar-link"><span class="icon">✅</span> c3_task</a>
       <a href="#c3_bitbucket" class="sidebar-link"><span class="icon">🪣</span> c3_bitbucket</a>
+      <a href="#c3_credentials" class="sidebar-link"><span class="icon">🔐</span> c3_credentials</a>
     </div>
     <div class="sidebar-section">
       <div class="sidebar-label">Cross-project</div>
@@ -161,6 +162,7 @@
         <a href="#c3_task" class="toc-item">c3_task <span class="cat">PM</span></a>
         <a href="#c3_artifacts" class="toc-item">c3_artifacts <span class="cat">Config</span></a>
         <a href="#c3_bitbucket" class="toc-item">c3_bitbucket <span class="cat">SCM</span></a>
+        <a href="#c3_credentials" class="toc-item">c3_credentials <span class="cat">Secrets</span></a>
         <a href="#c3_project"  class="toc-item">c3_project  <span class="cat">Multi-project</span></a>
       </div>
     </div>
@@ -1239,6 +1241,45 @@ c3_bitbucket(action=<span class="str">'merge_pr'</span>, pr_id=<span class="num"
           <p class="tool-desc"><strong>Audit trail:</strong> mutating actions (<code>merge_pr</code>, <code>create_branch</code>, <code>delete_branch</code>, webhook writes, etc.) are appended to the C3 edit ledger so the local audit log covers platform-side state changes too.</p>
         </div>
       </div>
+
+      <!-- c3_credentials -->
+      <div class="tool-card" id="c3_credentials">
+        <div class="tool-card-header">
+          <span class="tool-name">c3_credentials</span>
+          <div class="tag-row">
+            <span class="badge badge-purple">secrets</span>
+            <span class="badge">v2.58.0</span>
+          </div>
+          <span class="tool-tagline">Credential vault — the agent uses secrets by name, never by value</span>
+        </div>
+        <div class="tool-card-body">
+          <p class="tool-desc">A user-managed secret store, <strong>global</strong> (<code>~/.c3</code>, visible in every project) or <strong>per-project</strong> (<code>.c3</code>, shadows the same-named global). Values live in the OS keyring (service <code>c3-creds</code>; values over ~1KB in a Fernet-encrypted <code>.c3/secrets.enc</code> whose random master key is itself a keyring entry) — never in config files, and <em>never in the model's context</em>. <strong>Injection-first:</strong> the agent addresses secrets by name and C3 decodes them only at the subprocess boundary — <code>c3_shell(env_creds='A,B')</code> injects env vars, <code>{{cred:NAME}}</code> inside <code>cmd</code> expands server-side, and <code>inject</code>-flagged entries auto-inject into every shell run; echoed values are redacted back to <code>[cred:NAME]</code>. Resolution is <strong>realm-atomic</strong>: a project-registered name resolves in the project realm or not at all, so a hostile repo's committed config can never siphon your global values (tested).</p>
+
+          <h4>Actions</h4>
+          <table class="params-table">
+            <thead><tr><th>Action</th><th>Group</th><th>Description</th></tr></thead>
+            <tbody>
+              <tr><td class="param-name">list</td><td>Read</td><td class="param-desc">Merged registry (project shadows global) — names, scope, type, flags, usage. Never values</td></tr>
+              <tr><td class="param-name">describe</td><td>Read</td><td class="param-desc">Metadata + storage + live fingerprint (first 8 hex of SHA-256, computed on demand, never persisted)</td></tr>
+              <tr><td class="param-name">check</td><td>Read</td><td class="param-desc">Resolvability probe — does the stored value decode?</td></tr>
+              <tr><td class="param-name">set</td><td>Write</td><td class="param-desc">Create/replace an entry. <code>agent_readable</code> may be set at creation only — raising it on an existing entry is user-only</td></tr>
+              <tr><td class="param-name">delete</td><td>Write</td><td class="param-desc">Remove value + registry entry (owning scope inferred when omitted)</td></tr>
+              <tr><td class="param-name">reveal</td><td>Gated</td><td class="param-desc">The only value-returning action — refused unless the user enabled <code>agent_readable</code> on that entry</td></tr>
+            </tbody>
+          </table>
+
+          <h4>Examples</h4>
+          <pre><code><span class="com"># What can I use? (names and metadata only)</span>
+c3_credentials(action=<span class="str">'list'</span>)
+c3_credentials(action=<span class="str">'check'</span>, name=<span class="str">'NPM_TOKEN'</span>)
+
+<span class="com"># USE a secret without ever seeing it</span>
+c3_shell(cmd=<span class="str">'npm publish'</span>, env_creds=<span class="str">'NPM_TOKEN'</span>)
+c3_shell(cmd=<span class="str">'curl -H "Authorization: Bearer {{cred:API_KEY}}" https://api.example.com'</span>)</code></pre>
+
+          <p class="tool-desc"><strong>Surfaces:</strong> the <code>c3 creds</code> CLI (<code>set</code> / <code>get</code> / <code>list</code> / <code>rm</code> / <code>import .env</code>, <code>--global</code> for the shared scope), a per-project dashboard <strong>Credentials</strong> tab, and — since v2.59.0 — a top-level <strong>Credentials</strong> view in the Hub: a Global-vault manager plus a per-project browser with shadowing shown both ways ("shadows global" on project entries, "shadowed in N projects" on globals); the Hub's per-project drill panel manages too. No HTTP route ever returns a stored value (write-only wire contract, endpoint-sweep tested), every mutation is ledger-logged by name, the vault is hard-excluded from the Oracle Discovery API, and cross-project (<code>c3_project</code>) shells run with credentials disabled.</p>
+        </div>
+      </div>
     </div>
 
     <!-- ═══════════════════════════════════════ CROSS-PROJECT ═════ -->