Agent Performance Analytics

[OBenner]

Track and display detailed analytics on agent performance including success rates, completion times, error patterns, and quality metrics. Compare performance over time and across different task types.

Summary by CodeRabbit

• New Features

  • Comprehensive analytics dashboard: overview, per-agent stats, trends and QA report.
  • CLI and desktop bridge to fetch analytics as JSON; frontend exposes analytics API and IPC channels.
  • Sidebar navigation and mock data support for analytics; configurable trend window (days).

• Documentation

  • Added analytics service documentation detailing architecture, usage, CLI and examples.

[pantoaibot]

PR Summary:

Adds a new Agent Performance Analytics feature (backend data aggregation, CLI, frontend UI and IPC integration) to collect and display agent success rates, costs, QA metrics and trends.

Key changes:

• Backend analytics

  • New analytics service: apps/backend/services/analytics.py — aggregates data from spec folders (cost_report.json, attempt_history.json, implementation_plan.json, qa_report.md). Provides MetricsSummary, AgentStats, TaskComplexityStats, QAStats, TrendDataPoint and methods: get_metrics_summary, get_agent_stats, get_trend_data, get_analytics_report.
  • New CLI: apps/backend/cli/analytics_cli.py — thin CLI wrapper returning JSON for IPC usage.
  • Documentation: apps/backend/services/README_ANALYTICS.md describing architecture, usage, data sources, and caveats (I/O, caching).
  • Export: apps/backend/services/init.py exposes AnalyticsService.

• Frontend integration

  • New IPC handlers: apps/frontend/src/main/ipc-handlers/analytics-handlers.ts — spawns the Python CLI, parses JSON, and exposes handlers for summary, agent-stats, trends, report.
  • Register analytics handlers in IPC module: apps/frontend/src/main/ipc-handlers/index.ts.
  • New preload API module: apps/frontend/src/preload/api/modules/analytics-api.ts and integrated into agent-api and Electron API (createAgentAPI / createElectronAPI).
  • New shared IPC channels: added ANALYTICS_GET_* constants in shared constants.
  • New TypeScript types: apps/frontend/src/shared/types/analytics.ts and exported via shared types index; updated IPC types to include analytics methods.
  • Frontend UI: new Analytics React component (apps/frontend/src/renderer/components/Analytics.tsx) with Overview / Agents / Trends / QA views; integrated into App.tsx and Sidebar navigation (icon, route, i18n).
  • i18n: added English and French analytics strings and wired into i18n config.
  • README update: README.md mentions "Analytics" feature.

• Behavior / implementation notes

  • Frontend handlers run the Python CLI using configured Python path; results parsed from stdout JSON.
  • Analytics service scans .auto-claude/specs/ (configurable via --specs-dir) and returns aggregated, read-only metrics. It skips missing files and returns partial data on parse errors.
  • Performance considerations documented (lazy loading, potential need for caching/pagination for many specs).

• Potential breaking / integration impacts

  • New exports and TypeScript surface area (AnalyticsAPI added to AgentAPI and ElectronAPI). Consumers relying on strict AgentAPI shape may need to handle the added analytics property.
  • Running analytics requires a working Python environment and accessible spec directories; frontend assumes Python CLI is present and configured (via existing python-env manager).
  • No external dependency upgrades detected.

Files added:

• Backend: services/analytics.py, cli/analytics_cli.py, services/README_ANALYTICS.md
• Frontend: main/ipc-handlers/analytics-handlers.ts, renderer/components/Analytics.tsx, preload API module analytics-api.ts, types analytics.ts, i18n locales (en/fr), and multiple integration touchpoints (ipc registration, agent-api, createElectronAPI, App.tsx, Sidebar).

Files modified:

• README.md (feature list)
• apps/backend/services/init.py
• apps/frontend/src/main/ipc-handlers/README.md, index.ts
• apps/frontend/src/preload/api/* (agent-api.ts, index.ts, modules/index.ts)
• apps/frontend/src/renderer/App.tsx, Sidebar.tsx
• shared constants, types, i18n entries

Testing / usage:

• Use CLI locally: python analytics_cli.py --specs-dir
• Frontend Analytics UI calls IPC channels which run the CLI; ensure configured Python path and specs dir.

Reviewed by Panto AI

[pantoaibot]

[CRITICAL_BUG] Successful/failed attempt counting is incorrect. In get_metrics_summary you increment stats.successful_attempts and stats.failed_attempts by 1 per spec when plan_data reports completed/failed subtasks (lines 406-410). This yields wrong per-agent success/failure counts because agent usage (usage["count"]) may be >1 or attempts should be derived from attempt_history. Instead: (a) use attempt_history data (preferred) to map attempts->success/failure per agent, or (b) at minimum increment by usage["count"] rather than +1 so agent attempt counts align with agent_usage. Update logic to compute successes/failures per-agent based on accurate sources.

# Inside get_metrics_summary, when updating agent_stats

# Use attempt history to derive per-agent success/failure where possible
attempt_subtasks = attempt_data.get("subtasks", {})

for agent_type, usage in cost_data["agent_usage"].items():
    if agent_type not in summary.agent_stats:
        summary.agent_stats[agent_type] = AgentStats(agent_type=agent_type)

    stats = summary.agent_stats[agent_type]
    stats.total_attempts += usage["count"]
    stats.total_cost += usage["cost"]
    stats.total_tokens += usage["tokens"]

    # Prefer attempt_history if it has per-agent outcome data
    agent_attempts = attempt_subtasks.get(agent_type)
    if isinstance(agent_attempts, list):
        for attempt in agent_attempts:
            outcome = attempt.get("status") or attempt.get("outcome")
            if outcome == "success":
                stats.successful_attempts += 1
            elif outcome in {"failed", "error"}:
                stats.failed_attempts += 1
    else:
        # Fallback: align successes/failures with usage count so rates are consistent
        if plan_data["completed_subtasks"] > 0:
            stats.successful_attempts += usage["count"]
        if plan_data["failed_subtasks"] > 0:
            stats.failed_attempts += usage["count"]

[pantoaibot]

[CRITICAL_BUG] The code mixes module-based Python invocation and direct-script invocation. The comment says "python -m apps.backend.cli.analytics_cli" but the implementation builds fullArgs with pythonBaseArgs and then appends scriptPath (line 43). If parsePythonCommand returns base args containing -m/module, passing a script path afterwards will produce an invalid command. Choose one approach and make it explicit: either (A) use module invocation: args = [...pythonBaseArgs, 'apps.backend.cli.analytics_cli', command, ...] and ensure pythonBaseArgs contains '-m' if needed, or (B) invoke script path directly: args = [...pythonBaseArgs (should be empty), scriptPath, command, ...]. Validate parsePythonCommand output and document expected shape. Add unit tests or runtime checks to fail fast if the combination is invalid.

async function executeAnalyticsCommand(
  projectPath: string,
  command: string,
  args: string[]
): Promise<any> {
  return new Promise((resolve, reject) => {
    const pythonPath = getConfiguredPythonPath();
    const [pythonCommand, pythonBaseArgs] = parsePythonCommand(pythonPath);
    const sourcePath = getEffectiveSourcePath();
    const backendPath = path.join(sourcePath, "apps", "backend");

    // Use module-based invocation to match comment and avoid mixing styles
    const fullArgs = [...pythonBaseArgs, "-m", "apps.backend.cli.analytics_cli", command, ...args];

    const env = getAugmentedEnv();
    const pythonProcess = spawn(pythonCommand, fullArgs, {
      cwd: backendPath,
      env,
    });

    let stdout = "";
    let stderr = "";

    pythonProcess.stdout?.on("data", (data) => {
      stdout += data.toString();
    });

    pythonProcess.stderr?.on("data", (data) => {
      stderr += data.toString();
    });

    pythonProcess.on("close", (code) => {
      if (code !== 0) {
        debugError(`[Analytics] Command failed with code ${code}:`, stderr);
        reject(new Error(stderr || `Process exited with code ${code}`));
      } else {
        try {
          const result = JSON.parse(stdout);
          resolve(result);
        } catch (error) {
          debugError("[Analytics] Failed to parse JSON output:", error);
          reject(new Error(`Failed to parse JSON: ${error}`));
        }
      }
    });

    pythonProcess.on("error", (error) => {
      debugError("[Analytics] Process error:", error);
      reject(error);
    });
  });
}

[pantoaibot]

[REFACTORING] analytics_cli inserts parent.parent into sys.path (line 24) to import services.analytics. This is brittle for packaging and when the script is executed from different working directories. Prefer installing the backend package (or running via 'python -m apps.backend.cli.analytics_cli') and using package imports, or compute a stable project root through environment/configuration rather than mutating sys.path. Add a comment explaining the intention if keeping this hack.

#!/usr/bin/env python3
"""
Analytics CLI
=============

Command-line interface for accessing analytics data from the frontend.
Provides JSON output for IPC communication.

Usage:
    python -m apps.backend.cli.analytics_cli summary --specs-dir <path>
    python -m apps.backend.cli.analytics_cli agent-stats --specs-dir <path>
    python -m apps.backend.cli.analytics_cli trends --specs-dir <path> --days <n>
    python -m apps.backend.cli.analytics_cli report --specs-dir <path>
"""

from __future__ import annotations

import argparse
import json
import sys
from pathlib import Path

# When run as a module (python -m apps.backend.cli.analytics_cli), package imports
# work without mutating sys.path. Direct script execution is supported via the
# module form documented above.

from apps.backend.services.analytics import AnalyticsService


def command_summary(args: argparse.Namespace) -> None:
    """Get metrics summary."""
    service = AnalyticsService(specs_dir=Path(args.specs_dir))
    summary = service.get_metrics_summary()
    print(json.dumps(summary.to_dict(), indent=2))

# ... rest of file unchanged ...

[pantoaibot]

[REFACTORING] OverviewView (line 246), AgentsView (line 371), TrendsView (line 452) and QAView (line 496) are typed using any for props. Replace these any types with the concrete types from shared/types/analytics (e.g. AnalyticsReport, MetricsSummary, AgentStats, TrendDataPoint, QAStats) to get compile-time safety and avoid runtime mistakes when fields are renamed or shapes change.

import type {
  MetricsSummary,
  AgentStats as AgentStatsType,
  QAStats as QAStatsType,
  TrendDataPoint,
} from '../../shared/types';

interface OverviewViewProps {
  summary: MetricsSummary;
  formatCurrency: (amount: number) => string;
  formatPercentage: (value: number) => string;
  formatNumber: (value: number) => string;
  t: typeof import('react-i18next').useTranslation extends (...args: any) => infer R
    ? R[0]
    : (key: string, options?: any) => string;
}

function OverviewView({ summary, formatCurrency, formatPercentage, formatNumber, t }: OverviewViewProps) {
  // ...body unchanged...
}

interface AgentsViewProps {
  agentStats: Record<string, AgentStatsType>;
  formatCurrency: (amount: number) => string;
  formatPercentage: (value: number) => string;
  formatDuration: (seconds: number) => string;
  formatNumber: (value: number) => string;
  t: OverviewViewProps['t'];
}

function AgentsView({ agentStats, formatCurrency, formatPercentage, formatDuration, formatNumber, t }: AgentsViewProps) {
  // ...body unchanged...
}

interface TrendsViewProps {
  trends: TrendDataPoint[];
  formatCurrency: (amount: number) => string;
  formatPercentage: (value: number) => string;
  formatNumber: (value: number) => string;
  t: OverviewViewProps['t'];
}

function TrendsView({ trends, formatCurrency, formatPercentage, formatNumber, t }: TrendsViewProps) {
  // ...body unchanged...
}

interface QAViewProps {
  qaStats: QAStatsType;
  formatPercentage: (value: number) => string;
  formatNumber: (value: number) => string;
  t: OverviewViewProps['t'];
}

function QAView({ qaStats, formatPercentage, formatNumber, t }: QAViewProps) {
  // ...body unchanged...
}

[pantoaibot]

Reviewed up to commit:b6df784114b95a9f6a052fea4ba8b53df528c9a7

Additional Suggestion

apps/backend/services/analytics.py, line:264-276

Implementation assumes attempt_history.json lives under spec_dir/memory/attempt_history.json (lines 266-267) but documentation and other parts of the repo indicate attempt_history.json may be at the spec root. This mismatch can cause attempt data to be missed and affect success/failure and timing metrics. Add a fallback to check both locations (spec_dir/attempt_history.json and spec_dir/memory/attempt_history.json) and normalize the structure. Use attempt history to compute attempt-level metrics rather than inferring from subtasks only.

def _parse_attempt_history(self, spec_dir: Path) -> dict[str, Any]:
    """Parse attempt_history.json from a spec directory.

    Supports both legacy and new locations:
      - <spec_dir>/attempt_history.json
      - <spec_dir>/memory/attempt_history.json
    """
    # Preferred new location
    memory_dir = spec_dir / "memory"
    attempt_file = memory_dir / "attempt_history.json"

    data = self._load_json_file(attempt_file)
    if not data:
        # Fallback: legacy location at spec root
        legacy_file = spec_dir / "attempt_history.json"
        data = self._load_json_file(legacy_file)

    if not data:
        return {"subtasks": {}, "stuck_subtasks": []}

    # Normalize structure
    return {
        "subtasks": data.get("subtasks", {}),
        "stuck_subtasks": data.get("stuck_subtasks", []),
    }

apps/backend/services/analytics.py, line:481-496

get_trend_data obtains spec date from 'updated_at' or 'metadata.created_at' (lines 487-491) but many specs may lack these fields. As a robust fallback, use the file's filesystem modification time (spec_dir / 'implementation_plan.json' mtime) or the spec directory mtime when metadata is missing. This will increase coverage of trend data and avoid silently skipping specs without metadata.

def get_trend_data(self, days: int = 30) -> list[TrendDataPoint]:
    ...
    cutoff_date = datetime.now(timezone.utc) - timedelta(days=days)
    daily_data = defaultdict(lambda: {"total": 0, "completed": 0, "cost": 0.0})

    spec_dirs = self._get_all_spec_dirs()

    for spec_dir in spec_dirs:
        plan_data = self._parse_implementation_plan(spec_dir)
        cost_data = self._parse_cost_report(spec_dir)

        plan_file = spec_dir / "implementation_plan.json"
        plan_json = self._load_json_file(plan_file)
        date_str = None
        if plan_json:
            date_str = plan_json.get("updated_at") or plan_json.get("metadata", {}).get("created_at")

        # Fallback to filesystem mtime if metadata is missing
        if not date_str:
            try:
                mtime = (spec_dir / "implementation_plan.json").stat().st_mtime
            except OSError:
                try:
                    mtime = spec_dir.stat().st_mtime
                except OSError:
                    continue
            spec_date = datetime.fromtimestamp(mtime, tz=timezone.utc)
        else:
            try:
                spec_date = datetime.fromisoformat(str(date_str).replace("Z", "+00:00"))
            except (TypeError, ValueError):
                continue

        if spec_date < cutoff_date:
            continue

        date_key = spec_date.strftime("%Y-%m-%d")
        daily = daily_data[date_key]
        daily["total"] += 1
        daily["cost"] += cost_data["total_cost"]
        if plan_data["status"] == "completed":
            daily["completed"] += 1

    ...  # existing aggregation into TrendDataPoint

apps/frontend/src/renderer/components/Analytics.tsx, line:72-80

formatCurrency constructs a new Intl.NumberFormat on every call (lines 72-80). Intl.NumberFormat is relatively expensive to create repeatedly. Memoize the formatter (useMemo or create it once outside the component) or reuse an instance to avoid unnecessary allocations on every render.

// Outside the component, reuse a single formatter
const usdFormatter = new Intl.NumberFormat('en-US', {
  style: 'currency',
  currency: 'USD',
  minimumFractionDigits: 2,
  maximumFractionDigits: 2,
});

// Inside Analytics component
const formatCurrency = (amount: number) => usdFormatter.format(amount);

apps/frontend/src/shared/types/analytics.ts, line:25-34

TaskComplexityStats and MetricsSummary use Record for complexity_stats (lines 27 and 61). Consider tightening the type to Record where appropriate so callers get correct autocompletion and you eliminate accidental keys. Also consider whether TaskComplexityStats should omit the 'complexity' property when stored in a keyed record (it can be redundant).

// In MetricsSummary
export interface MetricsSummary {
  total_specs: number;
  completed_specs: number;
  failed_specs: number;
  in_progress_specs: number;
  overall_success_rate: number; // Percentage (0-100)
  total_cost: number; // USD
  total_tokens: number;
  agent_stats: Record<string, AgentStats>; // Agent type -> stats
  complexity_stats: Record<SpecComplexity, TaskComplexityStats>; // Complexity -> stats
  qa_stats: QAStats;
  last_updated: string; // ISO 8601 timestamp
}

// Optionally, if complexity is always implied by the key, you could remove it:
export interface TaskComplexityStats {
  // complexity: SpecComplexity; // derived from the record key
  total_tasks: number;
  successful_tasks: number;
  success_rate: number; // Percentage (0-100)
  avg_completion_time: number; // Seconds
  avg_cost: number; // USD
}

Reviewed by Panto AI

[coderabbitai]

Warning

Rate limit exceeded

@OBenner has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 1 minutes and 36 seconds before requesting another review.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

📥 Commits

Reviewing files that changed from the base of the PR and between 87df473 and 969ec01.

⛔ Files ignored due to path filters (1)

• package-lock.json is excluded by !**/package-lock.json, !**/package-lock.json

📒 Files selected for processing (4)

• apps/backend/services/analytics.py
• apps/frontend/package.json
• apps/frontend/src/renderer/components/Analytics.tsx
• apps/frontend/src/renderer/components/Sidebar.tsx

📝 Walkthrough

Walkthrough

Adds a new AnalyticsService (backend) with a CLI and end-to-end frontend integration: IPC handlers, preload API, types, UI dashboard, localization, mocks, and documentation to surface aggregated agent/spec analytics to the app.

Changes

Cohort / File(s)	Summary
Backend Analytics Core apps/backend/services/analytics.py, apps/backend/cli/analytics_cli.py, apps/backend/services/__init__.py	New AnalyticsService and dataclasses (AgentStats, TaskComplexityStats, QAStats, MetricsSummary, TrendDataPoint). CLI exposes commands summary, agent-stats, trends, report. Service exported from services package.
Backend Docs apps/backend/services/README_ANALYTICS.md	New comprehensive analytics documentation describing architecture, data flow, usage, CLI examples, and integration notes.
Frontend IPC Handlers apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts, apps/frontend/src/main/ipc-handlers/index.ts	New IPC handlers that spawn the analytics CLI, parse JSON output, enforce timeout, and register channels for summary/agent-stats/trends/report. Wired into setupIpcHandlers.
Preload API & Composition apps/frontend/src/preload/api/modules/analytics-api.ts, apps/frontend/src/preload/api/modules/index.ts, apps/frontend/src/preload/api/agent-api.ts, apps/frontend/src/preload/api/index.ts	New AnalyticsAPI with getSummary/getAgentStats/getTrends/getReport. Integrated into createAgentAPI and exposed in preload API types and exports.
Frontend Types apps/frontend/src/shared/types/analytics.ts, apps/frontend/src/shared/types/ipc.ts, apps/frontend/src/shared/types/index.ts	New TypeScript interfaces for analytics data (AgentStats, MetricsSummary, TrendDataPoint, AnalyticsReport, query/state types) and IPC ElectronAPI analytics namespace.
Frontend UI & Navigation apps/frontend/src/renderer/components/Analytics.tsx, apps/frontend/src/renderer/App.tsx, apps/frontend/src/renderer/components/Sidebar.tsx	New Analytics React component with overview/agents/trends/qa views, integrated into App routing and sidebar (adds productivity view and adjusts analytics placement).
Frontend Mocks & Docs apps/frontend/src/renderer/lib/browser-mock.ts, apps/frontend/src/main/ipc-handlers/README.md	Browser mock extended with analytics endpoints. IPC handlers README updated to document analytics actions.
IPC Constants apps/frontend/src/shared/constants/ipc.ts	Added ANALYTICS_GET_SUMMARY, ANALYTICS_GET_AGENT_STATS, ANALYTICS_GET_TRENDS, ANALYTICS_GET_REPORT constants (note: additions appear duplicated in file).
Localization apps/frontend/src/shared/i18n/locales/en/analytics.json, apps/frontend/src/shared/i18n/locales/fr/analytics.json, apps/frontend/src/shared/i18n/locales/en/navigation.json, apps/frontend/src/shared/i18n/locales/fr/navigation.json, apps/frontend/src/shared/i18n/index.ts	Added English and French analytics translations and inserted navigation keys; integrated analytics locale resources into i18n config.
Small Frontend Tweaks apps/frontend/src/renderer/components/ImprovementTracker.tsx, apps/frontend/src/renderer/components/TaskProgress.tsx, apps/frontend/src/renderer/components/task-detail/TaskOverview.tsx	Minor UI adjustments: fragment removal and removal of explicit 'pending' status cases to fall through to default styling.

Sequence Diagram

sequenceDiagram
    actor User
    participant Frontend as Frontend App
    participant Preload as Preload/API
    participant IPC as IPC Handler
    participant CLI as Python CLI
    participant Service as AnalyticsService
    participant FS as Spec Files

    User->>Frontend: open Analytics view
    Frontend->>Preload: analytics.getReport(projectId)
    Preload->>IPC: invokeIpc(ANALYTICS_GET_REPORT, projectId)
    IPC->>CLI: spawn analytics_cli.py report --specs-dir
    CLI->>Service: AnalyticsService(specs_dir)
    Service->>FS: enumerate spec dirs and read reports
    loop per-spec
        Service->>FS: load cost_report.json / attempt_history.json / implementation_plan / qa_report.md
    end
    Service->>Service: aggregate metrics, build trends & report
    Service-->>CLI: JSON report (stdout)
    CLI-->>IPC: stdout (parsed JSON)
    IPC-->>Preload: IPCResult<AnalyticsReport>
    Preload-->>Frontend: resolve analytics.getReport
    Frontend->>Frontend: render dashboard UI

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

Possibly related PRs

• Multi-Model Provider Support #4 — Related via shared cost_report.json surface and multi-provider cost tracking types that overlap with analytics cost aggregation.

Poem

🐰 Hop! I dig through specs and logs, so spry,

I count the costs and watch trends fly,
From Python burrow to the app's bright screen,
Agents dance in charts, precise and keen,
A carrot-shaped graph—hooray! Analytics delight.

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'Agent Performance Analytics' directly summarizes the main feature being added across the entire changeset—a comprehensive analytics system for tracking and displaying agent performance metrics.
Docstring Coverage	✅ Passed	Docstring coverage is 85.37% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch auto-claude/048-agent-performance-analytics

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 11

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (2)

apps/frontend/src/shared/i18n/locales/fr/navigation.json (1)

15-24: ⚠️ Potential issue | 🟠 Major

Remove duplicate analytics key from French locale items.

items.analytics appears twice. Keep exactly one key to avoid parser/tooling ambiguity.

Suggested fix

     "worktrees": "Worktrees",
     "agentTools": "Aperçu MCP",
-    "analytics": "Analytique",
     "mergeAnalytics": "Analyse des fusions",

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/shared/i18n/locales/fr/navigation.json` around lines 15 -
24, Duplicate "analytics" key exists in the French navigation.json items; remove
the redundant entry so only a single "analytics" key remains (keep the first
occurrence "Analytique" and delete the second duplicate) to prevent
parser/tooling ambiguity and ensure unique keys in the items object.

apps/frontend/src/shared/i18n/locales/en/navigation.json (1)

15-24: ⚠️ Potential issue | 🟠 Major

Remove duplicate analytics key from items.

items.analytics is declared twice. Depending on parser/tooling, one value is dropped or file parsing can fail in strict modes. Keep a single declaration only.

Suggested fix

     "worktrees": "Worktrees",
     "agentTools": "MCP Overview",
-    "analytics": "Analytics",
     "mergeAnalytics": "Merge Analytics",

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/shared/i18n/locales/en/navigation.json` around lines 15 -
24, The JSON has a duplicate "analytics" key in the navigation items; remove the
redundant "analytics" entry so only one "analytics" key exists (leave the
desired label value intact) and validate the JSON syntax after editing; locate
the duplicate by searching for the "analytics" key in the navigation items block
(items.analytics) and delete the extra occurrence.

♻️ Duplicate comments (2)

apps/backend/cli/analytics_cli.py (1)

23-31: ⚠️ Potential issue | 🟠 Major

Avoid mutating sys.path for runtime imports.

Lines 23-31 make imports execution-mode dependent and brittle. Prefer package import (apps.backend.services.analytics) and invoke this CLI via module mode (python -m apps.backend.cli.analytics_cli).

♻️ Proposed direction

-# Ensure the backend package root is on sys.path so that
-# `services.analytics` resolves correctly when invoked as a script.
-# This is necessary because the frontend spawns this file directly
-# (not via `python -m`).
-_backend_root = str(Path(__file__).resolve().parent.parent)
-if _backend_root not in sys.path:
-    sys.path.insert(0, _backend_root)
-
-from services.analytics import AnalyticsService
+from apps.backend.services.analytics import AnalyticsService

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/backend/cli/analytics_cli.py` around lines 23 - 31, The current file
mutates sys.path via _backend_root and inserts it to enable from
services.analytics import AnalyticsService; remove the _backend_root,
sys.path.insert checks and the related comments and instead import the service
using its package path (from apps.backend.services.analytics import
AnalyticsService) so imports no longer depend on execution mode; update any
README or invocation notes to run the CLI as a module (python -m
apps.backend.cli.analytics_cli) to ensure the package import works.

apps/backend/services/analytics.py (1)

428-432: ⚠️ Potential issue | 🔴 Critical

Prevent fallback outcome double-counting.

Lines 429-432 can increment both successful_attempts and failed_attempts for the same usage["count"], which can push success/failure totals beyond total_attempts.

🛠️ Proposed fix

-                else:
-                    # Fallback: align successes/failures with usage count
-                    if plan_data["completed_subtasks"] > 0:
-                        stats.successful_attempts += usage["count"]
-                    if plan_data["failed_subtasks"] > 0:
-                        stats.failed_attempts += usage["count"]
+                else:
+                    # Fallback: use mutually exclusive spec-level status
+                    if status == "completed":
+                        stats.successful_attempts += usage["count"]
+                    elif status == "failed":
+                        stats.failed_attempts += usage["count"]

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/backend/services/analytics.py` around lines 428 - 432, The fallback
currently increments both stats.successful_attempts and stats.failed_attempts
for the same usage["count"] when plan_data["completed_subtasks"] and
plan_data["failed_subtasks"] are both > 0; change the logic to make the branches
mutually exclusive (e.g., use if/elif) so a single usage["count"] only
increments one counter—prefer incrementing stats.failed_attempts when
plan_data["failed_subtasks"] > 0, otherwise increment
stats.successful_attempts—ensuring totals (stats.total_attempts) are not
exceeded; update the block around plan_data, stats and usage to reflect this
exclusive decision.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@apps/backend/services/analytics.py`:
- Around line 516-533: The parsed spec_date from datetime.fromisoformat may be
naive; ensure spec_date is always timezone-aware UTC before comparing to
cutoff_date: after the datetime.fromisoformat call (and also after
datetime.fromtimestamp fallback), if spec_date.tzinfo is None set its tzinfo to
timezone.utc, otherwise convert it to UTC with astimezone(timezone.utc); apply
this normalization to the variable spec_date used in the comparison with
cutoff_date so the offset-naive vs offset-aware TypeError is avoided.

In `@apps/backend/services/README_ANALYTICS.md`:
- Around line 32-191: Several Markdown headings (e.g., "### MetricsSummary",
"### AgentStats", "### TaskComplexityStats", "### QAStats", "###
TrendDataPoint", "## Usage", "### Python Backend", "### CLI Access", "###
Frontend Integration", "## Features", "### Dashboard Views", "#### Overview
View", etc.) are missing the required blank lines; to fix, ensure there is an
empty line both above and below each heading so they comply with markdownlint
MD022 (insert a single blank line before and after every heading block in the
README_ANALYTICS.md content).
- Around line 20-28: Update the fenced code block in README_ANALYTICS.md so it
includes a language identifier (e.g., change ``` to ```text) for the block that
begins with "analytics.py (19KB)" to satisfy markdownlint MD040; locate the
triple-backtick fence around that ASCII tree and add the language tag
immediately after the opening backticks.

In `@apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts`:
- Around line 48-51: Replace the direct use of spawn (the pythonProcess created
from spawn(pythonCommand, fullArgs, { cwd: projectPath, env })) with the
centralized platform execution utility and path-joining helpers: stop calling
spawn directly and instead call the platform exec helper (the app's standardized
execution API) with pythonCommand and fullArgs, passing projectPath via the
platform helper's cwd option and env via its env option; also build any file
paths using the platform path-join utility rather than manual concatenation so
that pythonCommand, fullArgs and projectPath are executed through the platform
abstraction to preserve shared behavior.
- Around line 93-115: The handler for ANALYTICS_GET_SUMMARY currently returns an
object shaped as { summary: MetricsSummary } when the specs directory is
missing, which breaks the expected MetricsSummary contract; change the
early-return to return the MetricsSummary object itself (not wrapped in a
summary property) so callers receive a MetricsSummary with zeros and
last_updated, keeping the same fields (total_specs, completed_specs,
failed_specs, in_progress_specs, overall_success_rate, total_cost, total_tokens,
agent_stats, complexity_stats, qa_stats, last_updated); locate the check using
fileExists(specsDir) in agent-analytics-handlers.ts and replace the wrapped
return with the direct MetricsSummary value to match the ANALYTICS_GET_SUMMARY
return type.
- Around line 48-83: Add a 60-second timeout and cleanup to the spawned Python
process in the promise that creates pythonProcess: start a timer (setTimeout)
after spawn that, when fired, kills pythonProcess (pythonProcess.kill()), clears
listeners, and rejects the promise with a timeout Error; ensure the timer is
cleared on normal completion in the pythonProcess.on('close') and on('error')
handlers and guard resolve/reject so only one path runs (e.g., a local
"finished" boolean). Update the handlers around stdout/stderr,
pythonProcess.on('close') and pythonProcess.on('error') to clear the timeout and
not attempt to resolve/reject if the timeout already fired.

In `@apps/frontend/src/main/ipc-handlers/README.md`:
- Around line 92-93: The new heading "#### `analytics-handlers.ts`" needs a
blank line above it to satisfy markdownlint MD022; edit the README block that
contains the heading (look for the heading text analytics-handlers.ts) and
insert one empty line immediately before that heading so the heading is
separated from the previous paragraph or list.
- Around line 92-95: Update the documentation in the analytics-handlers.ts doc
block to use the correct IPC action name: replace the incorrect `ANALYTICS_GET`
entry with `ANALYTICS_GET_SUMMARY` so it matches the implemented action in
analytics-handlers.ts and avoids confusion with `ANALYTICS_GET_AGENT_STATS`.

In `@apps/frontend/src/renderer/App.tsx`:
- Around line 1046-1048: The analytics view is being rendered twice because both
Analytics and ProductivityDashboard are gated by the same activeView ===
'analytics' condition; update the view key so only one component owns
'analytics' (e.g., keep Analytics under activeView === 'analytics' and change
ProductivityDashboard to use a distinct key such as 'productivity'), and ensure
the conditional checks around Analytics (using activeView, activeProjectId,
selectedProjectId) and ProductivityDashboard are updated accordingly so they no
longer share the same activeView string.

In `@apps/frontend/src/renderer/components/Analytics.tsx`:
- Line 423: Replace the hardcoded literals in the Analytics component JSX (the
string "tokens" next to formatNumber(stats.total_tokens) and the "0%"
occurrence) with i18n lookups using react-i18next (e.g.,
t('analytics:agents.tokens') and a suitable key for the percent text, e.g.,
t('analytics:agents.zeroPercent', { value: percentValue }) or format the percent
via a translator-aware formatter), update the Analytics.tsx imports to use the
useTranslation hook, and ensure you pass any dynamic values (like formatted
numbers or percent) to the translation interpolation rather than embedding raw
strings; then add the new translation entries (analytics:agents.tokens and the
percent key) to both English and French locale files.

---

Outside diff comments:
In `@apps/frontend/src/shared/i18n/locales/en/navigation.json`:
- Around line 15-24: The JSON has a duplicate "analytics" key in the navigation
items; remove the redundant "analytics" entry so only one "analytics" key exists
(leave the desired label value intact) and validate the JSON syntax after
editing; locate the duplicate by searching for the "analytics" key in the
navigation items block (items.analytics) and delete the extra occurrence.

In `@apps/frontend/src/shared/i18n/locales/fr/navigation.json`:
- Around line 15-24: Duplicate "analytics" key exists in the French
navigation.json items; remove the redundant entry so only a single "analytics"
key remains (keep the first occurrence "Analytique" and delete the second
duplicate) to prevent parser/tooling ambiguity and ensure unique keys in the
items object.

---

Duplicate comments:
In `@apps/backend/cli/analytics_cli.py`:
- Around line 23-31: The current file mutates sys.path via _backend_root and
inserts it to enable from services.analytics import AnalyticsService; remove the
_backend_root, sys.path.insert checks and the related comments and instead
import the service using its package path (from apps.backend.services.analytics
import AnalyticsService) so imports no longer depend on execution mode; update
any README or invocation notes to run the CLI as a module (python -m
apps.backend.cli.analytics_cli) to ensure the package import works.

In `@apps/backend/services/analytics.py`:
- Around line 428-432: The fallback currently increments both
stats.successful_attempts and stats.failed_attempts for the same usage["count"]
when plan_data["completed_subtasks"] and plan_data["failed_subtasks"] are both >
0; change the logic to make the branches mutually exclusive (e.g., use if/elif)
so a single usage["count"] only increments one counter—prefer incrementing
stats.failed_attempts when plan_data["failed_subtasks"] > 0, otherwise increment
stats.successful_attempts—ensuring totals (stats.total_attempts) are not
exceeded; update the block around plan_data, stats and usage to reflect this
exclusive decision.

---

ℹ️ Review info

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 31d45c4 and b703c86.

📒 Files selected for processing (24)

• apps/backend/cli/analytics_cli.py
• apps/backend/services/README_ANALYTICS.md
• apps/backend/services/__init__.py
• apps/backend/services/analytics.py
• apps/frontend/src/main/ipc-handlers/README.md
• apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts
• apps/frontend/src/main/ipc-handlers/index.ts
• apps/frontend/src/preload/api/agent-api.ts
• apps/frontend/src/preload/api/index.ts
• apps/frontend/src/preload/api/modules/analytics-api.ts
• apps/frontend/src/preload/api/modules/index.ts
• apps/frontend/src/renderer/App.tsx
• apps/frontend/src/renderer/components/Analytics.tsx
• apps/frontend/src/renderer/components/Sidebar.tsx
• apps/frontend/src/renderer/lib/browser-mock.ts
• apps/frontend/src/shared/constants/ipc.ts
• apps/frontend/src/shared/i18n/index.ts
• apps/frontend/src/shared/i18n/locales/en/analytics.json
• apps/frontend/src/shared/i18n/locales/en/navigation.json
• apps/frontend/src/shared/i18n/locales/fr/analytics.json
• apps/frontend/src/shared/i18n/locales/fr/navigation.json
• apps/frontend/src/shared/types/analytics.ts
• apps/frontend/src/shared/types/index.ts
• apps/frontend/src/shared/types/ipc.ts

[coderabbitai]

⚠️ Potential issue | 🟠 Major

Completion-time metrics are currently always zero.

AgentStats.avg_completion_time and TaskComplexityStats.avg_completion_time are declared, but there is no accumulation/normalization path in get_metrics_summary, so time-based analytics are misleading.

Also applies to: 441-447, 465-467

[coderabbitai]

🛠️ Refactor suggestion | 🟠 Major

Use the centralized platform execution utility instead of direct spawn.

Direct child-process execution here bypasses shared platform behavior and increases drift across OS/runtime paths.

As per coding guidelines "Use path joining utilities and execution utilities from the platform abstraction module instead of manual string concatenation for file paths and command execution".

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts` around lines
48 - 51, Replace the direct use of spawn (the pythonProcess created from
spawn(pythonCommand, fullArgs, { cwd: projectPath, env })) with the centralized
platform execution utility and path-joining helpers: stop calling spawn directly
and instead call the platform exec helper (the app's standardized execution API)
with pythonCommand and fullArgs, passing projectPath via the platform helper's
cwd option and env via its env option; also build any file paths using the
platform path-join utility rather than manual concatenation so that
pythonCommand, fullArgs and projectPath are executed through the platform
abstraction to preserve shared behavior.

[coderabbitai]

⚠️ Potential issue | 🟡 Minor

Add a blank line before the new heading to satisfy markdownlint (MD022).

The heading insertion is currently not surrounded by the required blank lines.

🧰 Tools

🪛 markdownlint-cli2 (0.21.0)

[warning] 92-92: Headings should be surrounded by blank lines
Expected: 1; Actual: 0; Below

(MD022, blanks-around-headings)

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/main/ipc-handlers/README.md` around lines 92 - 93, The new
heading "#### `analytics-handlers.ts`" needs a blank line above it to satisfy
markdownlint MD022; edit the README block that contains the heading (look for
the heading text analytics-handlers.ts) and insert one empty line immediately
before that heading so the heading is separated from the previous paragraph or
list.

[coderabbitai]

⚠️ Potential issue | 🟠 Major

Replace hardcoded UI literals with translation/formatter output.

Line 423 ("tokens") and Line 558 ("0%") are hardcoded in JSX, so localization can drift and bypass i18n.

🛠️ Proposed fix

-                  <div className="text-xs text-muted-foreground">{formatNumber(stats.total_tokens)} tokens</div>
+                  <div className="text-xs text-muted-foreground">
+                    {formatNumber(stats.total_tokens)} {t('analytics:agents.tokens')}
+                  </div>
@@
-                : '0%'
+                : formatPercentage(0)

Also add analytics:agents.tokens in English and French locale files.

As per coding guidelines "All user-facing text must use i18n translation keys from react-i18next with format namespace:section.key. Never use hardcoded strings in JSX/TSX. Update all language files (minimum: English and French) when adding new text."

Also applies to: 556-559

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/renderer/components/Analytics.tsx` at line 423, Replace the
hardcoded literals in the Analytics component JSX (the string "tokens" next to
formatNumber(stats.total_tokens) and the "0%" occurrence) with i18n lookups
using react-i18next (e.g., t('analytics:agents.tokens') and a suitable key for
the percent text, e.g., t('analytics:agents.zeroPercent', { value: percentValue
}) or format the percent via a translator-aware formatter), update the
Analytics.tsx imports to use the useTranslation hook, and ensure you pass any
dynamic values (like formatted numbers or percent) to the translation
interpolation rather than embedding raw strings; then add the new translation
entries (analytics:agents.tokens and the percent key) to both English and French
locale files.

[coderabbitai]

Actionable comments posted: 4

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

apps/frontend/src/renderer/components/Sidebar.tsx (1)

94-107: ⚠️ Potential issue | 🟠 Major

Keyboard shortcut collision: 'P' is assigned to both productivity and github-prs.

Both productivity (line 95) and github-prs (line 107) use shortcut 'P'. Since baseNavItems is processed before githubNavItems in visibleNavItems, pressing 'P' will always navigate to productivity, making the github-prs shortcut unreachable when both are visible.

🔧 Proposed fix - change github-prs shortcut

 const githubNavItems: NavItem[] = [
   { id: 'github-issues', labelKey: 'navigation:items.githubIssues', icon: Github, shortcut: 'G' },
-  { id: 'github-prs', labelKey: 'navigation:items.githubPRs', icon: GitPullRequest, shortcut: 'P' }
+  { id: 'github-prs', labelKey: 'navigation:items.githubPRs', icon: GitPullRequest, shortcut: 'H' }
 ];

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/renderer/components/Sidebar.tsx` around lines 94 - 107,
There is a keyboard shortcut collision: both the NavItem with id "productivity"
(in baseNavItems) and the NavItem with id "github-prs" (in githubNavItems) use
shortcut 'P', causing the base item to always win when visible; to fix, update
the "github-prs" NavItem in githubNavItems to use a different unused shortcut
(e.g., change 'P' to another letter like 'R' or 'H') and ensure the new key does
not conflict with other entries in baseNavItems or visibleNavItems so the GitHub
PR shortcut becomes reachable.

♻️ Duplicate comments (4)

apps/frontend/src/main/ipc-handlers/README.md (1)

92-99: ⚠️ Potential issue | 🟡 Minor

Add a blank line before the new heading.

The heading #### \agent-analytics-handlers.ts` at line 92 lacks a blank line above it (after line 90's content), violating markdownlint MD022. The action names are now correct (ANALYTICS_GET_SUMMARY` etc.).

📝 Proposed fix

 - Session management (list, new, switch, delete, rename)
 
+
 #### `agent-analytics-handlers.ts`

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/main/ipc-handlers/README.md` around lines 92 - 99, The
markdown heading "#### `agent-analytics-handlers.ts`" is missing a blank line
above it, violating MD022; insert a single blank line immediately before the
heading in apps/frontend/src/main/ipc-handlers/README.md so there is an empty
line between the previous paragraph (end of line ~90 content) and the heading
`agent-analytics-handlers.ts`, keeping the rest of the list
(`ANALYTICS_GET_SUMMARY`, `ANALYTICS_GET_AGENT_STATS`, etc.) unchanged.

apps/backend/services/analytics.py (2)

48-48: ⚠️ Potential issue | 🟠 Major

avg_completion_time is never populated.

AgentStats.avg_completion_time and TaskComplexityStats.avg_completion_time are declared but no code path accumulates or normalizes completion times. These fields will always be zero, making time-based analytics misleading.

Consider either:

1. Implementing time tracking by parsing timestamps from attempt history or plan metadata
2. Removing or marking these fields as "not yet implemented" to avoid confusion

Also applies to: 80-80

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/backend/services/analytics.py` at line 48,
AgentStats.avg_completion_time and TaskComplexityStats.avg_completion_time are
declared but never calculated; update the stats aggregation to compute average
completion time by extracting start/end timestamps from the relevant attempt
history or plan metadata (e.g., parse Attempt.history entries or
Plan.completed_at/started_at fields), sum durations per item and divide by count
(handle missing timestamps and zero-counts), or alternatively remove/flag these
fields as "not implemented" in AgentStats and TaskComplexityStats to avoid
misleading zeros; update the aggregation function(s) that populate AgentStats
and TaskComplexityStats to perform this calculation or mark the fields
accordingly.

---

519-520: 🧹 Nitpick | 🔵 Trivial

Add explanatory comment for the empty except block.

The pass after catching TypeError, ValueError is intentional (allows fallback to mtime), but lacks explanation. Adding a brief comment satisfies code clarity and addresses the static analysis hint.

📝 Proposed fix

                 except (TypeError, ValueError):
-                    pass
+                    pass  # Invalid date format; fall back to filesystem mtime below

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/backend/services/analytics.py` around lines 519 - 520, The empty except
block catching (TypeError, ValueError) should include a brief explanatory
comment clarifying that the pass is intentional to allow the code to fall back
to using file mtime; update the block (the except (TypeError, ValueError): pass)
to add a one-line comment such as "# Intentionally ignore parsing errors so we
can fall back to mtime" directly above or on the same line as the pass to
satisfy clarity and static-analysis checks.

apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts (1)

48-51: 🛠️ Refactor suggestion | 🟠 Major

Use the centralized platform execution utility instead of direct spawn.

Direct child-process execution bypasses shared platform behavior. The coding guidelines require using execution utilities from the platform abstraction module.

As per coding guidelines "Use path joining utilities and execution utilities from the platform abstraction module instead of manual string concatenation for file paths and command execution".

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts` around lines
48 - 51, Replace the direct child-process spawn call that creates pythonProcess
with the centralized execution utility from our platform abstraction: stop using
spawn and instead import and call the platform exec/run utility (e.g.,
runProcess/executeCommand) passing pythonCommand, fullArgs, and an options
object containing cwd: projectPath and env; also replace any manual path
concatenation around projectPath/script paths with the platform path-join
utility. Remove the direct spawn usage and ensure the returned handle/error
streams from the platform utility are used where pythonProcess was previously
referenced.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@apps/backend/services/analytics.py`:
- Line 381: Replace the deprecated naive call datetime.utcnow() used when
constructing MetricsSummary (e.g., summary =
MetricsSummary(last_updated=datetime.utcnow().isoformat() + "Z")) with a
timezone-aware timestamp: use datetime.now(timezone.utc).isoformat() (and remove
the manual "+ 'Z'" if present). Add the timezone import from datetime (ensure
timezone is imported) and apply the same change to the other occurrence at the
later MetricsSummary usage.

In `@apps/backend/services/README_ANALYTICS.md`:
- Line 227: Replace the vague phrase in the Memory bullet ("consider pagination
for very large datasets") with a precise technical recommendation: state that
the process loads all spec data into memory and advise using pagination or
streaming for datasets that exceed available memory or a defined record
threshold (e.g., millions of rows), and mention measuring memory usage or
applying server-side pagination/streaming as mitigation; update the sentence
that currently reads "**Memory**: Loads all spec data into memory - consider
pagination for very large datasets" to this tightened wording.

In `@apps/frontend/src/renderer/components/Analytics.tsx`:
- Around line 593-594: The sort comparator on
Object.entries(qaStats.common_issues) uses `any` for the destructured tuple;
replace that with proper tuple typing since common_issues is a Record<string,
number>. Update the comparator signature to use tuple types like ([, a]:
[string, number], [, b]: [string, number]) or ensure qaStats.common_issues is
strongly typed so TypeScript infers entries as [string, number], then use
(.sort(([, a], [, b]) => b - a)) to remove `any` and keep correct number
comparisons; reference qaStats.common_issues and the sort callback in your
change.
- Around line 74-76: The effect references loadAnalytics but doesn't list it as
a dependency, risking a stale closure; fix by either wrapping loadAnalytics in
useCallback with the appropriate dependencies (e.g., projectId and any
state/setters it uses) and then include loadAnalytics in the useEffect
dependency array, or move the fetch/logic currently in loadAnalytics inline into
the useEffect and remove loadAnalytics from dependencies; ensure the chosen
approach includes all variables used by the logic (projectId, state setters) in
the dependency list so the effect updates correctly.

---

Outside diff comments:
In `@apps/frontend/src/renderer/components/Sidebar.tsx`:
- Around line 94-107: There is a keyboard shortcut collision: both the NavItem
with id "productivity" (in baseNavItems) and the NavItem with id "github-prs"
(in githubNavItems) use shortcut 'P', causing the base item to always win when
visible; to fix, update the "github-prs" NavItem in githubNavItems to use a
different unused shortcut (e.g., change 'P' to another letter like 'R' or 'H')
and ensure the new key does not conflict with other entries in baseNavItems or
visibleNavItems so the GitHub PR shortcut becomes reachable.

---

Duplicate comments:
In `@apps/backend/services/analytics.py`:
- Line 48: AgentStats.avg_completion_time and
TaskComplexityStats.avg_completion_time are declared but never calculated;
update the stats aggregation to compute average completion time by extracting
start/end timestamps from the relevant attempt history or plan metadata (e.g.,
parse Attempt.history entries or Plan.completed_at/started_at fields), sum
durations per item and divide by count (handle missing timestamps and
zero-counts), or alternatively remove/flag these fields as "not implemented" in
AgentStats and TaskComplexityStats to avoid misleading zeros; update the
aggregation function(s) that populate AgentStats and TaskComplexityStats to
perform this calculation or mark the fields accordingly.
- Around line 519-520: The empty except block catching (TypeError, ValueError)
should include a brief explanatory comment clarifying that the pass is
intentional to allow the code to fall back to using file mtime; update the block
(the except (TypeError, ValueError): pass) to add a one-line comment such as "#
Intentionally ignore parsing errors so we can fall back to mtime" directly above
or on the same line as the pass to satisfy clarity and static-analysis checks.

In `@apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts`:
- Around line 48-51: Replace the direct child-process spawn call that creates
pythonProcess with the centralized execution utility from our platform
abstraction: stop using spawn and instead import and call the platform exec/run
utility (e.g., runProcess/executeCommand) passing pythonCommand, fullArgs, and
an options object containing cwd: projectPath and env; also replace any manual
path concatenation around projectPath/script paths with the platform path-join
utility. Remove the direct spawn usage and ensure the returned handle/error
streams from the platform utility are used where pythonProcess was previously
referenced.

In `@apps/frontend/src/main/ipc-handlers/README.md`:
- Around line 92-99: The markdown heading "#### `agent-analytics-handlers.ts`"
is missing a blank line above it, violating MD022; insert a single blank line
immediately before the heading in apps/frontend/src/main/ipc-handlers/README.md
so there is an empty line between the previous paragraph (end of line ~90
content) and the heading `agent-analytics-handlers.ts`, keeping the rest of the
list (`ANALYTICS_GET_SUMMARY`, `ANALYTICS_GET_AGENT_STATS`, etc.) unchanged.

---

ℹ️ Review info

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between b703c86 and 87df473.

📒 Files selected for processing (12)

• apps/backend/services/README_ANALYTICS.md
• apps/backend/services/analytics.py
• apps/frontend/src/main/ipc-handlers/README.md
• apps/frontend/src/main/ipc-handlers/agent-analytics-handlers.ts
• apps/frontend/src/renderer/App.tsx
• apps/frontend/src/renderer/components/Analytics.tsx
• apps/frontend/src/renderer/components/ImprovementTracker.tsx
• apps/frontend/src/renderer/components/Sidebar.tsx
• apps/frontend/src/renderer/components/TaskProgress.tsx
• apps/frontend/src/renderer/components/task-detail/TaskOverview.tsx
• apps/frontend/src/shared/i18n/locales/en/navigation.json
• apps/frontend/src/shared/i18n/locales/fr/navigation.json

💤 Files with no reviewable changes (2)

• apps/frontend/src/renderer/components/TaskProgress.tsx
• apps/frontend/src/renderer/components/task-detail/TaskOverview.tsx

[coderabbitai]

⚠️ Potential issue | 🟡 Minor

Replace deprecated datetime.utcnow() with datetime.now(timezone.utc).

datetime.utcnow() is deprecated in Python 3.12+ and returns a naive datetime. Use datetime.now(timezone.utc) for consistency with the timezone-aware handling elsewhere in this file.

🔧 Proposed fix

-        summary = MetricsSummary(last_updated=datetime.utcnow().isoformat() + "Z")
+        summary = MetricsSummary(last_updated=datetime.now(timezone.utc).isoformat().replace("+00:00", "Z"))

-            "generated_at": datetime.utcnow().isoformat() + "Z",
+            "generated_at": datetime.now(timezone.utc).isoformat().replace("+00:00", "Z"),

Also applies to: 581-581

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/backend/services/analytics.py` at line 381, Replace the deprecated naive
call datetime.utcnow() used when constructing MetricsSummary (e.g., summary =
MetricsSummary(last_updated=datetime.utcnow().isoformat() + "Z")) with a
timezone-aware timestamp: use datetime.now(timezone.utc).isoformat() (and remove
the manual "+ 'Z'" if present). Add the timezone import from datetime (ensure
timezone is imported) and apply the same change to the other occurrence at the
later MetricsSummary usage.

[coderabbitai]

⚠️ Potential issue | 🟡 Minor

Tighten wording at Line 227 for stronger technical tone.

“Very large datasets” is vague; use a more precise phrase.

Suggested doc tweak

-- **Memory**: Loads all spec data into memory - consider pagination for very large datasets
+- **Memory**: Loads all spec data into memory - consider pagination for large datasets

📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change

	- **Memory**: Loads all spec data into memory - consider pagination for very large datasets
	- **Memory**: Loads all spec data into memory - consider pagination for large datasets

🧰 Tools

🪛 LanguageTool

[style] ~227-~227: As an alternative to the over-used intensifier ‘very’, consider replacing this phrase.
Context: ...a into memory - consider pagination for very large datasets ## Internationalization The ...

(EN_WEAK_ADJECTIVE)

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/backend/services/README_ANALYTICS.md` at line 227, Replace the vague
phrase in the Memory bullet ("consider pagination for very large datasets") with
a precise technical recommendation: state that the process loads all spec data
into memory and advise using pagination or streaming for datasets that exceed
available memory or a defined record threshold (e.g., millions of rows), and
mention measuring memory usage or applying server-side pagination/streaming as
mitigation; update the sentence that currently reads "**Memory**: Loads all spec
data into memory - consider pagination for very large datasets" to this
tightened wording.

[sonarqubecloud]

Quality Gate passed

Issues
21 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
2.0% Duplication on New Code

See analysis details on SonarQube Cloud