rembric-plugin-development
Facilitates the development and management of Rembric agent plugins, ensuring compliance with specifications and validation processes.
Install this skill
or
0/100
Security score
The rembric-plugin-development skill was audited on May 26, 2026 and we found 8 security issues across 3 threat categories, including 6 high-severity. Review the findings below before installing.
Categories Tested
Security Issues
high line 28
Template literal with variable interpolation in command context
SourceSKILL.md
| 28 | - **Claude Code**: `${user_config.*}` works in both `mcp.json::env` AND hook commands; keychain is the SoT for credentials. |
high line 29
Template literal with variable interpolation in command context
SourceSKILL.md
| 29 | - **Codex CLI**: `${user_config.*}` is NOT substituted; subprocess env is **cleared** before MCP spawn → MUST list every needed var in `env_vars: [...]`; `${CLAUDE_PLUGIN_ROOT}` doesn't work in MCP ar |
high line 38
Template literal with variable interpolation in command context
SourceSKILL.md
| 38 | - **Legitimately divergent today**: `hooks/hooks.json` vs `hooks/hooks.codex.json` (env-substitution rules differ); `.claude-plugin/mcp.json` vs `.codex-plugin/mcp.json` (`${CLAUDE_PLUGIN_ROOT}` works |
medium line 31
Access to hidden dotfiles in home directory
SourceSKILL.md
| 31 | - **opencode**: every named export of a plugin file is invoked as a Plugin function — export ONLY `RembricPlugin`. The bridge MUST live outside `~/.config/opencode/plugins/`. Sub-agent filtering (`par |
medium line 44
Access to hidden dotfiles in home directory
SourceSKILL.md
| 44 | - **Never edit the user's agent config file** (`settings.json`, `config.toml`, `opencode.json`, `~/.hermes/config.yaml`). Print the snippet with `<PLACEHOLDERS>`; the user pastes. |
high line 28
Access to system keychain/keyring
SourceSKILL.md
| 28 | - **Claude Code**: `${user_config.*}` works in both `mcp.json::env` AND hook commands; keychain is the SoT for credentials. |
high line 78
Access to system keychain/keyring
SourceSKILL.md
| 78 | **If you cannot drive the agent TUI** (live LLM cost, or you're testing keychain integration you can't script): |
high line 83
Access to system keychain/keyring
SourceSKILL.md
| 83 | **If local e2e is genuinely impossible** (Codex `plugin_hooks` feature gate, Claude Code keychain integration, tool you don't have installed): say so out loud. Tell the user: |
Scanned on May 26, 2026
View Security Dashboard