Report.prompt

---
description: Generate a report on completed work, a feature area, or a specific topic — saved to docs/reports/
agent: agent
---

# Generate Report

You are generating a structured report based on this chat session or a user-specified topic. The report will be saved as a permanent markdown file in `docs/reports/`.

**CRITICAL — Voice & Authorship:** The report must be written **in the user's voice**, as if the user personally performed the work, made the decisions, and is presenting the findings. The user needs to be able to share these reports directly with colleagues or management without editing. Never reference yourself (the agent, Copilot, AI, etc.) anywhere in the report. Use first person ("I", "my") or neutral professional voice ("We audited...", "The team implemented..."). Never say "the user asked" or "the user requested" — instead say "I identified...", "after reviewing...", or describe the motivation directly.

## What You Receive

The user may:
- Ask you to report on the work just completed in this chat session
- Ask you to report on a specific feature, module, or area of the codebase
- Ask you to report on a specific topic (e.g., architecture decisions, audit results, performance analysis, migration status)
- Provide a custom focus or scope for the report

If the user doesn't specify a scope, default to **summarizing all work completed in this chat session**.

## Steps

### 1. Gather Context

- **If reporting on completed work:** Scan this conversation to identify all changes made — code changes, bug fixes, features, refactors, config changes, documentation updates. Note which files were touched and what the outcomes were.
- **If reporting on a specific topic:** Search the codebase, read relevant files, and gather all information needed to produce an accurate, thorough report on the requested topic.
- **Check recent task logs** in `docs/task/logs/` for additional context on recent work.
- **Read relevant `docs/logic/` files** to understand the documented state of affected areas.
- **Read agent observation logs** in `docs/agent-observations/` (`critical.md`, `recommendations.md`, `anomalies.md`) — check for unresolved observations that may be relevant.

### 2. Determine Report Type & Title

Based on what you're reporting on, choose the most appropriate report type:

| Type | When to use |
|------|------------|
| `work-summary` | Summarizing work completed in this session or a date range |
| `feature-report` | Deep dive on a specific feature or module |
| `audit` | Code quality, security, performance, or compliance analysis |
| `architecture` | Architecture decisions, patterns, or technical debt assessment |
| `migration` | Status of a data, code, or infrastructure migration |
| `incident` | Bug investigation, root cause analysis, resolution |
| `comparison` | Evaluating options, tools, or approaches |
| `status` | Project or sprint status update |
| `integration` | Payment provider integration status, API compatibility |
| `custom` | Anything that doesn't fit the above — describe the type in the report header |

Create a descriptive, searchable title (e.g., "Webhook Signature Verification Implementation", "Glitch API Integration Audit", "Refund Processing Architecture Decision").

### 3. Generate the Report File

**File path:** `docs/reports/{YYYY-MM-DD}-{slug}.md`

- `{YYYY-MM-DD}` = today's date
- `{slug}` = kebab-case summary of the report topic (e.g., `webhook-verification-implementation`, `api-integration-audit`, `refund-architecture-decision`)

If `docs/reports/` doesn't exist, create it.

### 4. Report Structure

Use this template, adapting sections based on report type: