CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

todoist

Use when invoked on any codebase to perform a thorough scan of its structure, completion state, features, and gaps, then produce a ranked TODOIST.md of actionable work items with context, rationale, and competitive insight.

Quality

74%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/development-skills/todoist/SKILL.md
SKILL.md
Quality
Evals
Security

Codebase TODOIST Generator

Overview

Scan the codebase end-to-end, understand what it is and what it does, assess its completion state, compare it against proven patterns from comparable public projects, then write a TODOIST.md file at the repo root. Every item in that file is ranked, scoped, explained with a "why", and estimated — ready to execute without additional context.

Privacy guarantee: This skill never reads, logs, or outputs secrets, credentials, .env files, personal data, or anything outside the current repository's tracked files.


When to Use

  • Invoked explicitly (/codebase-todoist, use codebase-todoist skill, or generate TODOIST.md)
  • Starting work on an unfamiliar codebase and needing a prioritized action plan
  • Before a sprint, milestone, or release to find gaps and feature opportunities
  • After a large refactor, to re-assess outstanding work
  • When a project feels "stuck" and you want structured next steps

Do NOT use when:

  • The project has no committed source files (nothing to scan)
  • The user only wants a generic checklist unrelated to this codebase

Phase 1 — Codebase Discovery

Run these steps in order. Collect evidence; do not write TODOIST.md yet.

1.1 Identity & Stack

Read: README.md, package.json / pyproject.toml / go.mod / Cargo.toml / pom.xml
      composer.json, Gemfile, .nvmrc, .ruby-version, Dockerfile, docker-compose.yml
      wrangler.toml / wrangler.json, vercel.json, netlify.toml, fly.toml

Extract:

  • Project name and one-sentence purpose
  • Language(s) and primary framework(s)
  • Runtime target (browser SPA, SSR, API, CLI, mobile, library, worker, …)
  • Package manager (npm/pnpm/yarn/bun, pip/uv, cargo, go, …)
  • Deployment target (Cloudflare Workers, Vercel, Railway, Docker, bare VPS, …)

1.2 Architecture Scan

Read: src/, app/, lib/, server/, worker/, api/, pages/, routes/, components/
      (recurse up to 3 levels; do not read node_modules, .git, dist, build, .next, .cache)

Identify:

  • Entry points and main modules
  • Data-flow pattern (MVC, feature-sliced, domain-driven, flat, monorepo, …)
  • Database / persistence layer and migration state
  • Auth surface (none, JWT, session, OAuth, API keys)
  • External service integrations (APIs, queues, storage, …)
  • Test surface: presence and coverage of unit / integration / e2e tests
  • CI/CD config: .github/workflows/, .circleci/, Makefile targets

1.3 Completion State Assessment

For each major module or surface area assign one of:

StateMeaning
✅ DoneFunctional, tested, production-ready
🔧 PartialCore logic exists but incomplete, untested, or broken
❌ MissingClearly implied by the stack/spec but absent
💡 OpportunityNot present, but proven valuable in comparable projects

Scan for hard signals:

  • TODO, FIXME, HACK, XXX comments in source
  • Placeholder routes returning 501/stub responses
  • Empty or near-empty files
  • Commented-out blocks of significant size
  • Schema migrations with no corresponding business logic
  • Test files that import non-existent modules

1.4 Existing TODO Inventory

grep -rn "TODO\|FIXME\|HACK\|XXX\|PLACEHOLDER\|STUB\|WIP" \
  --include="*.ts" --include="*.tsx" --include="*.js" --include="*.jsx" \
  --include="*.py" --include="*.go" --include="*.rs" --include="*.php" \
  --include="*.rb" --include="*.java" --include="*.kt" \
  src/ app/ lib/ server/ worker/ api/ pages/ routes/ 2>/dev/null \
  | grep -v node_modules | grep -v ".git"

Record file + line + content for each hit; deduplicate by theme.


Phase 2 — Competitive & Ecosystem Research

Goal: surface features that are proven to work in comparable products, so recommendations are grounded in market evidence, not speculation.

2.1 Identify Comparable Projects

Using the project identity established in Phase 1, search for:

"[project type] open source [primary language]" GitHub stars > 500
"[project type] SaaS features" site:github.com
"[project type] alternatives" site:reddit.com OR site:hn.algolia.com

Examples by project type:

Project typeReference projects to look up
REST/GraphQL APIfastify, hono, nestjs ecosystem starter repos
React SPAbulletproof-react, t3-stack, refine
CLI toolcobra-based CLIs, oclif starters
E-commercemedusa, woocommerce, solidus
CMS / blogghost, keystatic, payload
Auth servicebetter-auth, lucia, clerk OSS
Data pipelinedagster, prefect OSS starters
Mobile (RN)expo-router starters, react-native-template-obytes

Search for feature lists, changelogs, and roadmaps of the 2–3 closest analogues. Extract features tagged as "highly requested", "v2", "shipped in X.Y", or starred in their own TODOs.

2.2 Filter for Relevance

Only carry forward competitive features that:

  1. Make sense given this project's stack and purpose
  2. Are implementable without replacing the existing architecture
  3. Provide clear user or developer value

Label these "Competitive Opportunity" in the output.


Phase 3 — Ranking Matrix

Score every candidate work item on three dimensions (1–3 each):

Dimension1 (Low)2 (Medium)3 (High)
ImpactNice-to-haveMeaningfully improves UX/DX/reliabilityUnblocks core value or prevents data loss
Effort> 2 weeks solo2–5 days< 1 day
RiskBreaks existing contracts / requires migrationIsolated change with rollback pathAdditive, zero regression risk

Priority score = Impact × 2 + Effort + Risk (range 4–15; higher = do first)

Items scoring ≥ 12 are P0 – Critical. Items scoring 9–11 are P1 – High. Items scoring 6–8 are P2 – Medium. Items scoring ≤ 5 are P3 – Low / Backlog.


Phase 4 — Write TODOIST.md

Create (or overwrite) TODOIST.md at the repository root.

Required Structure

# TODOIST — [Project Name]

> Generated: [ISO date]  
> Stack: [languages & frameworks]  
> Completion snapshot: [X% of surfaces assessed as ✅ Done]

---

## 🔴 P0 — Critical (do these first)

### [SHORT IMPERATIVE TITLE]
- **Why:** [one sentence — user or business consequence of not doing this]
- **What:** [concrete description of the deliverable]
- **How:** [key steps or approach; link to relevant file(s)]
- **Effort:** [rough estimate]
- **Source:** [InCode TODO | Architecture Gap | Competitive Opportunity | Spec]
- **Files:** [`path/to/file.ts`](path/to/file.ts) (if applicable)

[repeat for each P0 item]

---

## 🟠 P1 — High

[same template]

---

## 🟡 P2 — Medium

[same template]

---

## 🔵 P3 — Low / Backlog

[same template — can be collapsed to a simple bulleted list if > 8 items]

---

## 📊 Completion Snapshot

| Area | State | Notes |
|------|-------|-------|
| [module/surface] | ✅ / 🔧 / ❌ / 💡 | [short note] |

---

## 🌐 Competitive Insights Used

| Feature | Seen In | Why Recommended |
|---------|---------|-----------------|
| [feature] | [project] | [reason] |

---

_TODOIST.md is a living document. Re-run this skill after major milestones to refresh priorities._

Content Rules

  • No secrets, credentials, or personal data — never copy .env values, API keys, tokens, internal hostnames, or personal names into TODOIST.md
  • File paths are repo-relative — use src/routes/auth.ts, never absolute paths like /home/user/...
  • Every item has a "Why" — items without business or technical justification are cut
  • Use the project's own language — if the codebase calls something "handlers", use that word, not "controllers"
  • Competitive items are clearly labelled — never present a speculative feature as an existing gap
  • Avoid vague verbs — "Improve X" is invalid. Use "Add rate-limiting middleware to /api/auth routes", "Extract UserService from routes/user.ts", etc.
  • Keep it scannable — P0/P1 items get full template; P3 items may be brief bullets

Phase 5 — Self-Check Before Delivering

Before writing the final file, verify:

  • No absolute local paths (no /home/, /Users/, /root/)
  • No secrets, tokens, passwords, or .env values
  • No references to the local machine's username, hostname, or directory structure
  • All file references are repo-relative and exist in the tracked codebase
  • At least one item per non-empty priority tier
  • Every P0 and P1 item has Impact ≥ 2 confirmed
  • Competitive items are labelled and sourced
  • TODOIST.md is written to the repo root (not a subdirectory, not an artifact directory)

Common Mistakes

MistakeFix
Listing generic "add tests" without scopeScope to specific module: "Add unit tests for AuthService.verifyToken()"
Marking everything P0Apply the scoring matrix; 3–5 P0s is normal for most projects
Including placeholder text from templatesScan for [PROJECT NAME], TODO: fill, and remove before writing
Missing the completion snapshotAlways include the table — it's the fastest way to orient a new contributor
Copying raw grep output as-isSynthesize and deduplicate; 40 TODOs may collapse to 6 themes
Exposing local file pathsAll paths must be repo-relative

Quick Reference

Phase 1 → Discover (stack, architecture, completion state, inline TODOs)
Phase 2 → Research (comparable projects, proven features)
Phase 3 → Score (Impact × 2 + Effort + Risk → P0/P1/P2/P3)
Phase 4 → Write TODOIST.md (ranked, scoped, with why + how)
Phase 5 → Self-check (no secrets, no absolute paths, no vague items)

Deliver TODOIST.md. Tell the user what was found and how many items landed in each priority tier.

Repository

administrakt0r/pro-skills-repo-administraktor

Path

skills/development-skills/todoist/SKILL.md

Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.