| name | Report Engine |
| description | Shared reporting engine for generating session, verification, and compliance reports across workspace workflows. |
| version | 1.0.0 |
Skill: Report Engine
Workspace Skill for the Report Engine. Provides standard templates and rules for generating session summary reports, verification logs, and compliance audits.
Pattern: Workflow = Logic ➔ Sidecar Skill = Data & Template Router.
Any workflow invoking the --report option loads this skill to generate standardized outputs.
When to Load
- Triggered when the user runs any workflow with the
--report flag (e.g., /plan end --report, /verify --report).
- Loaded by Agent to format and populate objective summaries at session completion or verification phases.
References
| File | When | Purpose |
|---|
references/dev-session-report-template.md | Called during plan closure (/plan end --report) | Standard structure for comprehensive dev session reports |
Data Collection & Analysis Instructions
When generating a report, the Agent MUST collect and analyze statistics across the following 6 dimensions, compile a General Overview, and execute a Deep-Dive Focus analysis.
General Overview
- Accomplishments: Provide a concise summary of all features implemented, bugs fixed, and goals achieved.
- Code Quality Evaluation: Rate the overall cleanliness, type safety, and architectural integrity of the changes.
- Strengths & Risks: Highlight the most robust aspects of the implementation and note any remaining risks, code debt, or warnings.
Core 6 Dimensions
1. Token & Performance
- Token Estimation: Estimate the token consumption of the session based on context length or session transcript logs.
- Latency & Performance: List execution times for major test suites, build scripts, or AST building. Note performance degradation if any command exceeded normal limits.
2. CSA Compliance
- CSA Audit: Run the MCP tool
graph_audit_csa (or CLI npx para-graph audit csa).
- Traceability: Document final spec coverage %, active double-binding comments (
// @para-doc), and identify if any spec anchors are dangling.
3. Snapshot & Drift
- Drift Detection: Run the MCP tool
project_snapshot followed by project_diff to get the list of physically added, modified, or deleted files.
- Comparison: Cross-check the modified file list against the plan's File Inventory.
- Watchlist Check: Verify the integrity of files registered under
project_protected_files to confirm no drift occurred on core structures.
4. TDD Execution
- Evidence Count: Read
artifacts/tests/tdd-evidence.log (if TDD was active).
- Audit: Count the exact number of RED (failing test) and GREEN (minimal code passing) states. Confirm that no PASS occurred before a FAIL for each respective feature.
5. Session KI Usage
- Compaction Status: Check if
project_session_compact was run and confirm that session.md is updated.
- Recovery Analytics: Document how many times JIT Context Recovery reloaded indices from
session.md versus falling back to raw rule parsing.
6. Tool & Terminal Usage
- Native vs Bash vs MCP: Count the tool invocations from the session log:
- Native API:
view_file, grep_search, replace_file_content, write_to_file.
- Bash commands:
run_command pipeline executions.
- MCP tools:
para-graph tools.
- Highlight if the specificity priority rule (
MCP -> Native -> Bash) was obeyed.
Deep-Dive Focus Analysis
The report MUST include a dedicated section focusing deeply on one of the 6 dimensions chosen by the user (the "Focus Area").
Focus Area Resolution:
- If
--report-focus flag is passed, use the matching dimension as the Focus Area.
- If
--report-focus is not passed, prompt the user: "Which dimension would you like to deep-dive into? (csa, tdd, performance, drift, tools, session)". If the user doesn't specify, default to the area with the most active changes or complexity.
Deep-Dive Guidelines per Focus Area:
- CSA Compliance: List every spec anchor bound, quote the corresponding spec requirement text, analyze why the implementation meets or exceeds it, and trace the semantic path in the graph.
- TDD: Step-by-step breakdown of every RED-to-GREEN transition, detail how the minimal code was written, the refactoring steps performed, and any test coverage challenges.
- Performance: Analyze code hotspots, algorithmic complexity of new functions, trace latency bottlenecks in test runs, and propose future optimization steps.
- Snapshot & Drift: Detailed breakdown of physical file modifications, explain why any deviating files were modified, and analyze structural changes in the codebase layout.
- Tools: Analyze tool call patterns, evaluate tool efficiency, and identify any anti-patterns (e.g. over-reliance on bash commands instead of native tools).
- Session KI: Trace context size progression across the session, list specific JIT context recovery events, and evaluate the clarity of compressed guidelines in
session.md.