Opencode-controller
Analysis
This is a coherent instruction-only controller for Opencode, with no embedded code or install behavior, but users should verify Opencode, authentication links, session reuse, and delegated code changes.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
Required binaries (all must exist): none ... No install spec — this is an instruction-only skill.
The skill has no install package or declared binary requirement, while its instructions direct use of Opencode. This is purpose-aligned, but the user should rely only on a trusted local Opencode installation.
- Ask Opencode to implement the approved plan.
The skill delegates implementation to Opencode Build mode. That is the stated purpose, but it can result in project file changes.
Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.
- Ask the user which AI provider to use. - Ask how the provider should be authenticated. - Do not proceed without confirmation.
The skill involves provider authentication, which is expected for model selection. The artifact requires user confirmation and does not show hardcoded credentials or credential logging.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
- Opencode keeps a history of projects - The same project must always use the same session - Reusing sessions preserves context and decisions
The skill intentionally uses persistent Opencode sessions. This supports continuity, but retained context may contain sensitive project details or stale decisions.