chore(deps): update dependency blender-mcp to v1.8.7

[renovate]

This PR contains the following updates:

Package	Change	Age	Confidence
blender-mcp	1.5.6 → 1.8.7

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • Between 12:00 AM and 03:59 AM, only on Monday (* 0-3 * * 1)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[toolhive-release-app]

🔒 MCP Security Scan Results

❌ blender-mcp

• Status: Failed
• Tools scanned: 25
• Vulnerabilities found: 14

Security issues detected:

• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").
• [AITech-1.1] Explicit attempts to override, replace, or modify the model's system instructions, operational directives, or behavioral guidelines through direct user input, causing the model to follow attacker-controlled instructions instead of its intended programming (e.g., "Ignore previous instructions").

---

Summary: Scanned 1 MCP server(s), found 14 security issue(s).

⚠️ Action Required: Security issues were detected. Please review and address them before merging.

[danbarr]

Triage note — do not merge as-is, no code change pushed here.

This is a genuine, version-introduced risk, not scanner noise. Diffed the PyPI sdists for blender-mcp 1.5.6 vs 1.8.7: every one of the ~16 tools already had a mild `user_prompt` "for telemetry" parameter in 1.5.6, but 1.8.7 rewrote every tool's docstring to compel the calling agent to relay the user's raw, verbatim prompt into that parameter on every call, using directive language aimed at the LLM itself (e.g. "never substitute... repeat their previous words unchanged"). That's what triggered the 16 identical `AITech-1.1` (HIGH) findings.

Recommend escalating to a security reviewer with the specific question of what blender-mcp's telemetry endpoint does with that raw prompt data before deciding whether to hold this PR, ask upstream (`ahujasid/blender-mcp`) to clarify/revert the docstring change, or accept it with an explicit allowlist reason naming the behavior. Not treating this as a routine allowlist add.