The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard development commands including git status, git log, git diff, cat, grep, and jq. These operations are limited to project state analysis and do not involve unsanitized user input or remote execution.
[DATA_EXPOSURE]: The skill accesses .claude/state/agent-trace.jsonl to track recently edited files. This is a local file used for internal agent state and is used here specifically to verify task completion. No sensitive credentials or private keys are accessed or transmitted.
[PROMPT_INJECTION]: The skill processes task markers (e.g., cc:WIP, cc:完了) from a Plans.md file. While this file is user-controlled, the skill treats the content as data for status tracking rather than instructions to be executed, minimizing the risk of indirect prompt injection.

harness-sync