The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the local twg binary and the jq utility to interact with Atlassian internal data and filter large JSON outputs. Evidence: Multiple commands in SKILL.md and OUTPUT.md describe running twg and using jq for data processing.
[INDIRECT_PROMPT_INJECTION]: The skill ingests content from external sources which could contain malicious instructions designed to manipulate the agent. 1. Ingestion points: Untrusted content is retrieved from Jira work items, Confluence pages, and Bitbucket repositories via commands like jira workitem get and docs search (detailed in SKILL.md and ROUTING.md). 2. Boundary markers: The instructions do not define delimiters or markers to isolate ingested content from the agent's core instructions. 3. Capability inventory: The skill utilizes shell command execution (twg, jq) and local filesystem writes to /tmp and $TMPDIR. 4. Sanitization: No sanitization, validation, or escaping of the retrieved content is specified before the agent processes it.
[DATA_EXPOSURE]: The skill manages high-volume payloads by writing them to the local filesystem in temporary directories. Evidence: ASSETS_GRAPH.md and OUTPUT.md provide guidance on using --output-file with /tmp and $TMPDIR for schema discovery and stdout captures.

twg