aufrank/reporting-situation

Situation Report & Tracker

Overview

This skill provides a persistent, evolving view of your work environment. Unlike simple search tools, it maintains a tracker of known items (docs, issues, pages) and a knowledge graph of entities (teams, people) to reduce redundant API calls and provide historical context.

It aggregates data from:

• Google Workspace (Drive, Docs, Comments)
• Notion (Pages, Databases)
• Jira (Issues, Projects)

Core Features

1. Entity-Based Tracking: Monitors activity for specific Collaborators, Topics, and Teams.
2. Persistent Cache: Remembers mappings (e.g., "Team X" = "Jira Project Y") and previously seen items.
3. Smart Summarization: Automatically fetches and summarizes new items and discussion threads.
4. Deduplication: Merges signals from multiple sources into unique work items.

Usage

1. Run the Situation Report Sweep

This orchestrates the data gathering, updates the tracker, and generates the raw corpus.

python scripts/orchestrator.py

• Outputs:
  • Console: A high-level list of recent updates by person/topic.
  • situation_corpus.md: A large markdown file containing the full text of relevant documents and discussions.
  • data/tracker.json: Persistent state of tracked items.

2. Generate Executive Insights (AI Analysis)

Use the slicing skill to analyze the situation_corpus.md and generate a synthesized executive summary.

python scripts/generate_insights.py

• Outputs:
  • reports/insights/Executive_Summary.md: A structured summary of key developments, risks, and action items.
  • reports/insights/: Intermediate slice artifacts (useful for debugging).

Configuration

Edit templates/config.json to manage:

• Collaborators: People to track (Name, Email, Jira Handle).
• Topics: Keywords for semantic search (e.g., "ML Platform").
• Teams: Jira Project keys and Team names.
• Providers: Jira cloud_id and base_url (optional, auto-discovered if omitted).
• Insights: Slicing runner path and LLM provider settings for generate_insights.py.

Data Structures

• tracker.json: The database of all discovered items, their summaries, and last-seen timestamps.
• config.json: User definitions and runtime settings.
• interests.json: Deprecated (use config.json instead).

Requirements

• MCP Servers: @google, @notion, @jira sessions must be active.
• Environment: A .env file must exist in the root (even if dummy values) to satisfy the slicing runner checks. The actual LLM calls use the authenticated gemini CLI via a wrapper script.
• Optional: An MCP server supporting sampling.createMessage for auto-summarization.