The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches an installation script from the vendor's GitHub repository at https://raw.githubusercontent.com/codesoda/discuss-cli/main/install.sh to set up the required CLI environment.
[REMOTE_CODE_EXECUTION]: The installation instructions direct the agent to pipe a remote script directly into a shell (curl | sh). This is the primary method for deploying the vendor's discuss tool.
[COMMAND_EXECUTION]: The agent uses the Bash and Monitor tools to execute the discuss binary and perform network requests via curl to interact with a local API endpoint on 127.0.0.1.
[PROMPT_INJECTION]: The skill is designed to ingest and act upon external data, creating a surface for indirect prompt injection.
Ingestion points: The skill reads markdown files via the Read tool, accepts markdown content via stdin, and parses real-time JSON event streams (threads and replies) from the discuss process output.
Boundary markers: The instructions do not specify any delimiters or safety markers to separate untrusted markdown content or browser-generated thread text from the agent's internal instructions.
Capability inventory: The skill utilizes Bash for shell execution, Monitor for persistent process management, and curl for interacting with a web API, providing a significant functional surface if the agent is influenced by malicious content.
Sanitization: There is no evidence of sanitization or validation performed on the ingested markdown content or the event payloads before they are processed by the agent to generate 'takes'.

discuss