Menu
Discover all actionable tasks in the current project and add them to the task list
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | suggest-tasks |
| displayName | Suggest Project Tasks |
| description | Discover all actionable tasks in the current project and add them to the task list |
| version | 1.0.0 |
| tags | ["productivity","tasks","project-management","analysis"] |
Performs comprehensive project analysis and creates tasks for everything that needs attention.
Rails/Ruby:
Node.js/TypeScript:
Python:
Rust:
When the user runs /suggest-tasks, perform these steps:
Look for cached scan results at /tmp/project-scan-*.json. If recent (< 5 minutes), use those results. Otherwise, run a fresh scan.
Run the project detection to identify what kind of project this is:
~/.claude/hooks/analyzers/detect-project.sh .
Based on the project type, run the appropriate analyzers. Always run the universal analyzers:
~/.claude/hooks/analyzers/analyze-git.sh .
~/.claude/hooks/analyzers/analyze-todos.sh .
~/.claude/hooks/analyzers/analyze-tests.sh .
~/.claude/hooks/analyzers/analyze-deps.sh .
~/.claude/hooks/analyzers/analyze-docs.sh .
~/.claude/hooks/analyzers/analyze-security.sh .
Then run framework-specific analyzers based on detected types:
~/.claude/hooks/analyzers/analyze-rails.sh .~/.claude/hooks/analyzers/analyze-node.sh .~/.claude/hooks/analyzers/analyze-python.sh .~/.claude/hooks/analyzers/analyze-rust.sh .Each analyzer outputs JSON lines in this format:
{"category": "git", "task": "Task title", "description": "Details", "priority": "high|medium|low|critical"}
Collect all results and group by category.
For each discovered item, use TaskCreate to add it to the task list:
TaskCreate:
subject: "[category] Task title"
description: "Details from analyzer"
activeForm: "Checking task..."
After creating tasks, present a summary grouped by category:
## Project Analysis Complete
**Project Type:** Rails, TypeScript
**Tasks Created:** 24
### By Category:
#### Git & Version Control (3 tasks)
- [ ] Commit staged changes (high)
- [ ] Clean up merged branches (low)
- [ ] Push unpushed commits (medium)
#### Code Quality (5 tasks)
- [ ] Resolve 12 TODO comments (medium)
- [ ] Fix 3 FIXME comments (high)
...
#### Security (2 tasks)
- [ ] Review potential hardcoded secrets (critical)
- [ ] Update .gitignore for secrets (high)
...
If the user provides a filter argument, only show tasks matching that category:
/suggest-tasks git - Only git-related tasks/suggest-tasks security - Only security tasks/suggest-tasks rails - Only Rails-specific tasks/suggest-tasks node - Only Node.js-specific tasks/suggest-tasks python - Only Python-specific tasks/suggest-tasks rust - Only Rust-specific tasks/suggest-tasks todos - Only TODO/FIXME tasks/suggest-tasks tests - Only test-related tasks/suggest-tasks deps - Only dependency tasks/suggest-tasks docs - Only documentation tasksTasks are assigned priorities based on urgency:
/tmp/project-scan-*.json for reuseFix for webhook authorization failures caused by Base64 padding character (=) being dropped when copying secrets between systems. Use when: (1) Webhook returns "Unauthorized" or "Server Error" but the secret looks correct, (2) INTERNAL_API_SECRET, WEBHOOK_SECRET, or similar auth tokens fail validation between Railway/Vercel and Convex/database, (3) Secret comparison fails despite values appearing identical, (4) CLI tools display truncated environment variables. Common with Railway CLI, Vercel CLI, and other deployment platforms that may not display trailing = characters.
Search through Claude Code session memories and learned skills. Use when: (1) you need to find how a similar problem was solved before, (2) looking for relevant skills or knowledge, (3) checking if there's existing context about a project or technology, (4) the user asks "have we done this before?" or "what do you remember about X?". Triggers on: memory search, recall, remember, past sessions, history, previous work.
Understanding Convex + Better Auth dual-database architecture. Use when: (1) "User not found" errors during login/password reset but user appears to exist, (2) users exist in app's users table but can't authenticate, (3) need to create admin users in production, (4) debugging auth flows in Convex + Better Auth setup. Better Auth stores users in component tables separate from app tables.
Manage Better Auth user data in Convex component tables. Use when: (1) need to manually verify user email in development, (2) "Email not verified" error blocking login, (3) need to modify Better Auth user/session/account data directly, (4) can't write mutations to access component tables, (5) debugging auth issues in Convex + Better Auth setup. Covers CLI data access, export/import workflow, and common gotchas.
Fix for "Unauthorized" errors in Convex mutations called from Next.js API routes when using internal API secrets for authentication. Use when: (1) Convex mutation throws "Unauthorized" even though secret is set in .env.local, (2) webhook handlers fail with 500 status and "Unauthorized" in Convex function, (3) internal API calls work locally in Next.js but fail when calling Convex mutations. The fix is to set the secret in BOTH Next.js .env.local AND Convex environment variables.
Fix for expo-file-system migration errors when upgrading to Expo SDK 52+. Use when: (1) TypeScript error "Property 'cacheDirectory' does not exist on type", (2) "Module has no exported member 'EncodingType'" error, (3) "Property 'deleteAsync' does not exist" or similar FileSystem function errors, (4) file.info returns a function instead of file info object, (5) Upgrading Expo SDK and expo-file-system code breaks. Covers migration from legacy function-based API (FileSystem.deleteAsync, getInfoAsync) to new class-based API (File, Directory, Paths classes).