The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on a suite of local bash scripts located in ${HOME}/.claude-octopus/plugin/scripts/ to perform its operations. These include orchestrate.sh, build-fleet.sh, state-manager.sh, and check-providers.sh. These scripts are used for provider checks, fleet building, and managing project state.
[DATA_EXFILTRATION]: As part of its core functionality, the skill sends user research prompts to external providers including OpenAI (via Codex CLI), Antigravity CLI, and Perplexity. While this is the intended use case, it involves transmitting data to third-party services and consumes associated API credits.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and synthesize data from various external AI models, creating a vulnerability surface for indirect prompt injection. It proactively addresses this risk by mandating a 'Security Frame' for all external content. This framing include instructions to treat external content as informational rather than directive and uses clear boundary markers (ASCII boxes) to isolate untrusted data.

flow-discover