The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines workflows that execute standard shell commands (find, ls, grep, sed, printf) and version control operations (git add, git commit) to manage project planning artifacts and directory structures.
[DYNAMIC_EXECUTION]: The scripts/kata-lib.cjs utility performs dynamic path resolution at runtime to locate project templates and manages local configuration by programmatically updating .planning/config.json.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project files to drive its phase analysis logic, creating a vulnerability surface.
Ingestion points: The skill reads .planning/ROADMAP.md, .planning/STATE.md, and potentially existing CONTEXT.md files (referenced in SKILL.md and references/phase-discuss.md).
Boundary markers: No explicit boundary markers or 'ignore' instructions are used to wrap the ingested file content before it is processed by the agent.
Capability inventory: The skill is capable of writing implementation decisions to the file system (CONTEXT.md), updating configuration files, and committing changes via git.
Sanitization: The skill performs existence and basic structural validation (e.g., regex checks in kata-lib.cjs) but does not sanitize or filter the actual narrative content of the roadmap files for malicious instructions.

kata-discuss-phase