The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill creates a local shell script at scripts/capability-map.sh from internal assets and uses chmod +x to grant it execution permissions.
[DYNAMIC_EXECUTION]: The skill generates and executes a bash script (capability-map.sh) to perform documentation checks and update the capability table in CLAUDE.md or AGENTS.md using the --fix flag.
[COMMAND_EXECUTION]: The skill modifies project configuration files (e.g., package.json) and CI/CD workflow files (such as .github/workflows/ or .gitlab-ci.yml) to add automated execution of the validation script.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including meeting notes, product descriptions, and glossary files, to generate its drafts.
Ingestion points: Documented in SKILL.md (Step 2 and 3) and references/subagent-brief.md where external sources are read into the session context.
Boundary markers: The instructions lack specific delimiters or sanitization requirements for the content ingested from client documents before being processed by subagents.
Capability inventory: The skill has the capability to write files, execute shell scripts, and modify critical CI/CD pipeline configurations.
Sanitization: No explicit sanitization or escaping mechanisms are implemented for the external content before it is interpolated into the mapping process.

capability-map