The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes processes for reading and parsing GitHub Action logs, which are external sources that could potentially contain untrusted data.
Ingestion points: Output from gh run view --log and gh run view --log-failed as described in SKILL.md.
Boundary markers: None explicitly mentioned in the extraction examples.
Capability inventory: The skill utilizes the Bash tool to execute gh, git, awk, and jq.
Sanitization: The skill provides specific awk and jq patterns to filter and extract data, which limits the scope of ingested content.
[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes platform-specific markers to inject real-time context from the GitHub environment.
Evidence: The 'Context Expressions' section contains !gh run list ...`` commands.
Analysis: These commands are informational queries to the GitHub API used to provide the agent with current run status and do not involve sensitive file access or network exfiltration to external domains.

gh-workflow-monitoring