Home / Skills / planning-with-files

OthmanAdi/planning-with-files

OthmanAdi / planning-with-files

Agent SkillsShellMIT

Persistent file-based planning for AI coding agents and long-running tasks. Crash-proof markdown plans, session recovery after /clear and compaction, per-turn re-injection against context rot, deterministic completion gate. Manus-style. …

27.3kstars
2.3kforks
7open issues
2026-10-01last pushed
MITlicense

What it is, and how people use it

Persistent file-based planning for AI coding agents and long-running tasks. Crash-proof markdown plans, session recovery after /clear and compaction, per-turn re-injection against context rot, deterministic completion gate. Manus-style. … It sits on the Agent Skills shelf of the catalog, maintained by @OthmanAdi.

✓Drop-in for Claude Code, Codex CLI and Cursor - install and it just runs.
✓Read the SKILL.md / README first; most setups are one command from the docs.
✓Teams use it to standardise how agents behave across every repo in the org.
agent-skillsautonomous-agentsclaudeclaude-codeclaude-code-skills
www.skills.sh/othmanadi/planning-with-files/planning-with-files

Why builders keep it installedAgent Skills benefits

Your agent stops guessing the workflow

A skill packages the exact steps, file conventions and checklists an expert would follow. The model spends its context on your feature instead of figuring out how the tool works.

One install, every repo behaves the same

Commit the skill to your project and every teammate's Claude Code, Codex or Cursor run inherits the same quality bar - including the new teammate on day one.

Cheaper runs, fewer retries

Tuned instructions cut failed attempts and half-baked output. Fewer retries means fewer tokens burned and noticeably shorter sessions on paid model plans.

Survives the next model release

Skills are plain markdown and scripts the community versions together. A model upgrade changes behaviour, not your workflow - unlike a bespoke prompt that breaks silently.

Contributorsloading from GitHub…

Languagesestimated from snapshot

  • TypeScript 58.6%
  • Shell 41.4%

Community pulsesnapshot ratios, normalized

starsforksissues ×3watchers