The Agent Skills Directory

[COMMAND_EXECUTION]: The skill features a 'Metric Verification Mode' that accepts and executes arbitrary shell commands through the Metric and Guard parameters. These commands are used to automate measurement and verification during iterative loops.
[COMMAND_EXECUTION]: The skill performs automated Git operations including git add -A, git commit, and git revert as part of its experiment and rollback lifecycle.
[PROMPT_INJECTION]: The skill is configured to automatically activate based on specific natural language triggers such as "loop X times" or "loop around N times". This allows for seamless integration but acts as a wide trigger surface for iterative task execution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests configuration data from a local file at ~/.claude-octopus/loop-config.conf to override its self-regulation weights and thresholds.
Ingestion points: Reads configuration from ~/.claude-octopus/loop-config.conf (SKILL.md).
Boundary markers: Absent; the skill directly parses key=value pairs from the configuration file.
Capability inventory: Arbitrary shell command execution via Metric/Guard parameters and Git repository manipulation via git commands.
Sanitization: Absent; the skill uses the values directly to adjust its safety logic and iteration limits.
[DYNAMIC_EXECUTION]: The skill dynamically assembles and executes shell command strings for metric measurement and guard verification based on user-provided input parameters.

skill-iterative-loop