The Agent Skills Directory

[SAFE]: The skill interacts with the local file system (Claude Code session history in ~/.claude/projects) and development tools (GitHub, Jira) to gather information for standup reports. This behavior is restricted to the skill's primary stated purpose and is initiated only after explicit user approval through interactive prompts.
[COMMAND_EXECUTION]: The skill executes local shell commands and a bundled Python script to detect configured tools and extract work history.
Evidence: gh auth status, jira, ls ~/.claude/projects/*/*.jsonl, and python3 ~/.claude/skills/daily-meeting-update/scripts/claude_digest.py.
Context: All commands are used to populate the standup update with relevant technical context from the user's environment.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources (GitHub PRs, Jira tickets) and session history which may contain untrusted content that could influence the agent's output.
Ingestion points: Output from gh and jira CLI tools, and Claude Code .jsonl session log files.
Boundary markers: The instructions do not specify explicit delimiters or "ignore instructions" wrappers for the ingested content during the summarization phase.
Capability inventory: The agent has access to Bash for environmental checks and a local data processing script, as well as interactive user tools.
Sanitization: The provided claude_digest.py script parses JSON files and extracts specific metadata (titles, file paths), which effectively filters raw data before it enters the agent context.

daily-meeting-update