What causes "You no longer have access to subagent ${agentId}." in danny-avila/LibreChat?

Verify the user still has view access to the subagent and re-initiate if restored.

danny-avila/LibreChat · error · Error

You no longer have access to subagent ${agentId}.

Error message

You no longer have access to subagent ${agentId}.

What it means

Thrown during lazy subagent initialization when hasSubagentViewAccess returns false. The user had access when the graph was planned but lost it before the subagent was initialized (permission/role/ACL change mid-request).

Source

Thrown at api/server/services/Endpoints/agents/initialize.js:839

      return null;
    }
  };

  /**
   * Resolves the selected descriptor inside the foreground request. The
   * legacy initializer requires request/response objects for tool and MCP
   * setup, so this intentionally remains request-scoped until AI-1597 gives
   * child execution a durable runtime context.
   */
  const initializeLazySubagent = async ({ agentId, configId, context, lazyChildren }) => {
    throwIfAborted(context.signal);
    const agent = await waitForAbort(db.getAgentWithVersionCount({ id: agentId }), context.signal);
    throwIfAborted(context.signal);
    if (!agent || getLazySubagentConfigId(agent) !== configId) {
      throw new Error(`Subagent ${agentId} changed before it could be initialized.`);
    }
    if (!(await hasSubagentViewAccess(agent, agentId, context.signal))) {
      throw new Error(`You no longer have access to subagent ${agentId}.`);
    }
    const validation = await waitForAbort(
      validateAgentModel({ req, res, agent, modelsConfig, logViolation }),
      context.signal,
    );
    throwIfAborted(context.signal);
    if (!validation.isValid) {
      throw new Error(validation.error?.message ?? `Subagent ${agentId} failed model validation.`);
    }
    const scopedSkillIds = resolveAgentScopedSkillIds({
      agent,
      accessibleSkillIds,
      skillsCapabilityEnabled,
      ephemeralSkillsToggle,
    });
    const scopedEditableSkillIds = resolveAgentScopedSkillIds({
      agent,
      accessibleSkillIds: editableSkillIds,

View on GitHub (pinned to 5ff282f900)

Solutions

Verify the user still has view access to the subagent and re-initiate if restored.
Avoid revoking access to agents that have in-flight requests.
Adjust the subagent graph to only reference agents the user durably can access.
Surface the access loss to the end user with a re-auth/re-request prompt.

Defensive patterns

Strategy: try-catch

Validate before calling

if (!(await hasSubagentViewAccess(agent, agentId, signal))) {
  throw new Error(`Access to subagent ${agentId} revoked`);
}

Try / catch

try { await initializeLazySubagent({ agentId, configId, context }); }
catch (e) { if (/no longer have access/.test(e.message)) return res.status(403).json({ error: 'Access revoked' }); throw e; }

Prevention

Avoid revoking access to agents with in-flight requests.
Restrict graphs to agents the user durably owns/can view.
Re-check access early and surface revocation clearly to the user.

When it happens

Trigger: A permission or role change revokes the user's access to the subagent between graph planning and lazy initialization; the subagent was moved to a scope the user cannot view.

Common situations: An admin revoked a role or project membership during an in-flight request; the agent's access scope was tightened; sharing revoked concurrently.

Related errors

AI-assisted analysis of danny-avila/LibreChat@5ff282f900 (2026-08-12). Data as JSON: /api/errors/a6ef248287d784fc. Report an issue: GitHub.