The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install external command-line tools from vendor-managed sources.
It uses npx to execute the installer from the @mvanhorn/printing-press-library package on the NPM registry.
It provides fallback installation instructions using go install targeting the github.com/mvanhorn/printing-press-library repository.
[COMMAND_EXECUTION]: The skill's primary functionality is delivered through the execution of the techtwitter-pp-cli binary via the Bash tool.
The instructions define various subcommands for searching (tweets search), summarizing (digest), and analyzing topic momentum (momentum).
[DATA_EXFILTRATION]: The CLI tool includes a built-in feature to deliver command output to external endpoints.
The --deliver webhook:<url> flag allows the tool to POST result data directly to a specified URL in JSON or NDJSON format.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted content from the internet (curated tweets).
Ingestion points: External tweet content is ingested into the agent's context through commands like digest, tweets search, and agent in the SKILL.md instructions.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the fetched tweet text as untrusted or to ignore embedded instructions.
Capability inventory: The skill enables the Bash tool and filesystem access (via the CLI's SQLite mirror).
Sanitization: No sanitization or filtering of the fetched tweet content is described in the skill instructions.

pp-techtwitter