The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of shell commands to gather data and perform actions, including git log, gh search, and meegle. It also uses osascript to interact with the macOS Calendar application. Evidence includes commands with interpolated variables for dates and user configuration such as git log --since='{start_time}' --until='{end_time}' and gh search prs --author=@me --created='{monday}..{end_date}'.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from multiple external and untrusted sources that could contain embedded instructions intended to manipulate the agent. 1. Ingestion points: Obsidian work logs, Feishu project tasks, GitLab Merge Requests/Issues, GitHub Pull Requests/Issues, and Git commit logs. 2. Boundary markers: Absent; there are no instructions to the agent to treat fetched data as untrusted or to ignore instructions contained within it. 3. Capability inventory: The skill has capabilities for shell execution (gh, git, osascript, meegle), file reading/writing (Obsidian vaults), and system interaction via AppleScript. 4. Sanitization: No sanitization or filtering of external content is mentioned before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the meegle tool using npx @lark-project/meegle@latest install, which fetches and executes code from the npm registry.
[DATA_EXFILTRATION]: The skill accesses highly sensitive local data, including source code repositories located in ~/code/*/ and notes within Obsidian vaults. While this access is aligned with the skill's purpose of report generation, the broad read access combined with network-capable tools represents a data exposure surface.

weekly-report