The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install external binaries (flighty-pp-cli, flighty-pp-mcp) from the vendor's GitHub repository via go install and from the npm registry via npx.
[DATA_EXFILTRATION]: The CLI tool provides a --deliver flag that supports a webhook:<url> sink. This functionality allows for the transmission of command outputs and airport data to arbitrary external network endpoints.
[DYNAMIC_EXECUTION]: The skill implements a persistent learning loop (recall, teach, playbook) that journals interactions and auto-synthesizes command "playbooks." These playbooks contain sequences of shell commands that are dynamically executed in future sessions to automate workflows.
[INDIRECT_PROMPT_INJECTION]: The skill has a large attack surface for indirect injection as it ingests data from external Flighty SSR payloads and interpolates untrusted user-supplied strings into recall, teach, and playbook amend commands. This could allow malicious data to influence the agent's learned behaviors and future command execution through the synthesized playbooks.
Ingestion points: Flighty SSR payloads (HTML/JSON), user-provided queries in recall and teach commands.
Boundary markers: None identified for processed payload data or taught queries.
Capability inventory: The skill uses Read Bash to execute the flighty-pp-cli binary, which has broad capabilities including file system access (SQLite, config files) and network delivery via webhooks.
Sanitization: The documentation notes a PII rule for manual input but does not detail automated sanitization or escaping of ingested data before it is stored in the local learning database.

pp-flighty