The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands such as sqlite3 and cat that include user-provided inputs like run_id.
Evidence: SKILL.md contains commands like sqlite3 .gorgon/checkpoints.db "SELECT ... WHERE workflow_run_id = '{run_id}'" and cat .gorgon/runs/{run_id}/executor/execution-results.json.
Risk: Lack of input validation for the run_id placeholder could lead to command injection or path traversal if the agent does not sanitize the value before interpolation into the shell commands.
[PROMPT_INJECTION]: The skill analyzes agent logs, which serves as an indirect prompt injection vector.
Ingestion points: The skill reads log files via the diagnose_failure capability from log_path as defined in both SKILL.md and schema.yaml.
Boundary markers: Absent; no instructions or delimiters are provided to separate untrusted log content from the agent's core instructions.
Capability inventory: The agent has access to powerful tools including Bash, Read, Glob, and Grep.
Sanitization: Absent; the skill does not include instructions to sanitize, escape, or filter log content before it is processed by the agent.

workflow-debugger