logly/mureo

v0.10.24 Security

This release includes 2 security fixes for security teams reviewing exposed deployments.

Published 13d AI Agents & Assistants

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

This release patches 2 known CVEs

Topics

advertising agentic-ai ai-agents claude-code cli codex

+11 more

cursor facebook-ads gemini-cli google-ads marketing marketing-automation mcp meta-ads model-context-protocol python search-console

Affected surfaces

auth rbac breaking_upgrade

ReleasePort's take

Moderate signal

editorial:auto 12d

Scope `account_id` / `customer_id` to the active workspace on multi‑account backends, preventing cross‑client data access. Guardrails now reject non‑finite or oversized budget values, closing bypasses that allowed unlimited budgets.

Why it matters: These changes block unauthorized data exposure across clients and eliminate a critical path for unbounded budget allocation; operators should verify workspace scoping and budget guardrail configurations immediately after upgrade.

Summary

AI summary

Scoped account/customer IDs prevent cross‑client data access and budget guardrails now reject non‑finite or oversized values.

Changes in this release

Type	Severity	Summary	CVE
Security	Critical	Scope `account_id` / `customer_id` to active workspace on multi-account backends, preventing cross‑client data access. Scope `account_id` / `customer_id` to active workspace on multi-account backends, preventing cross‑client data access. Source: llm_adapter@2026-07-14 Confidence: high	—
Security	High	Guardrails reject non‑finite or oversized budget values, closing bypasses that allowed unlimited budgets. Guardrails reject non‑finite or oversized budget values, closing bypasses that allowed unlimited budgets. Source: llm_adapter@2026-07-14 Confidence: high	—
Feature	Medium	Adds Creative Studio gallery tab for browsing generated visuals per client on multi‑account backends. Adds Creative Studio gallery tab for browsing generated visuals per client on multi‑account backends. Source: llm_adapter@2026-07-14 Confidence: high	—
Feature	Medium	Adds plugin budget‑key declarations in MCP metadata for StrategyPolicyGate enforcement. Adds plugin budget‑key declarations in MCP metadata for StrategyPolicyGate enforcement. Source: llm_adapter@2026-07-14 Confidence: high	—
Bugfix	Medium	Fixes race condition in Configure UI causing credential status to appear unset after reload. Fixes race condition in Configure UI causing credential status to appear unset after reload. Source: llm_adapter@2026-07-14 Confidence: high	—

Full changelog

Security

Meta account_id / Google Ads customer_id are scoped to the active workspace on multi-account backends (#411). These free caller arguments were used with operator-shared credentials without validation, so on an agency backend one client's conversation could read — or with write tools mutate — a sibling client's account by passing its id. New allow-list resolvers enforce the effective id (argument and credentials default alike) before any client is built, fail closed when a multi-account backend declares nothing, and filter google_ads_accounts_list. Standalone installs unaffected.
Budget guardrails can no longer be bypassed with non-finite or oversized values (#419). A NaN (reachable over the wire) made every cap comparison abstain, and an oversized integer raised an OverflowError the gate swallowed into an allow. Both now fail closed across the built-in scan, the total/increase caps, and the plugin declaration path.

Added

Creative Studio gallery (#409) — a read-only dashboard tab browsing generated visuals and composed banners per run, browsable per client on multi-account backends, with strict path containment on image serving.
Plugin budget-key declarations (#414) — a plugin declares its budget argument keys in MCP metadata so the one built-in StrategyPolicyGate enforces STRATEGY.md Guardrails on its platform, no per-plugin gate.

Fixed

Configure UI credential status/writes no longer race the credentials path (#406) — set_host now publishes the fully resolved HostPaths in a single atomic assignment, fixing saved credentials showing as unset after a reload.

Docs

Japanese trigger phrases on operational skill descriptions (#396); docs/plugin-authoring.md refreshed for #324/#327 and the new budget seam (#414); SECURITY.md rotation/backup note (#394).

Existing installs: run mureo upgrade (or the configure UI's Reinstall button) if you have not since 0.10.23, to refresh the credential-guard hooks.

Full Changelog: https://github.com/logly/mureo/compare/v0.10.23...v0.10.24

Security Fixes

Meta `account_id` / Google Ads `customer_id` now scoped to active workspace, preventing cross‑client data access (CVE not specified)
Budget guardrails reject non‑finite and oversized values, closing bypass vulnerability (CVE not specified)

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track logly/mureo

Get notified when new releases ship.

About logly/mureo

Framework for AI agents (Claude Code, Cursor, Codex, Gemini) to operate Google Ads, Meta Ads, and Search Console. Grounded in a local STRATEGY.md — not metric-chasing. Defense-in-depth security, local-first. Apache 2.0.

All releases →