Guide · Agent skill
How to use the Planning with Files (Spanish) agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
planning-with-files-es is an agent skill, a folder of instructions and scripts that a compatible AI agent loads when it needs them. It is the Spanish edition of the Planning with Files skill: it has the agent keep its working memory in Markdown files on disk (task_plan.md, findings.md and progress.md) instead of relying only on the chat history. This guide explains what the skill does, how to install it, how to trigger it, and what to check for safety before you use it.
Planning with Files (Spanish)OthmanAdi/planning-with-files · MIT Download the skill Quick start
- 1Download and unzip the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ for all projects, or into the project's .claude/skills/ folder for one project.
- 2Check the requirements. Confirm that you have a POSIX shell or bash, Git, and a host agent that supports SKILL.md skills with hooks. Python 3 is only needed for session-catchup.py.
- 3Start or resume a task directory. For a new task, run scripts/init-session.sh with a task name and set the host to the PLAN_ID it prints. To resume, reuse the plan already selected.
- 4Create only the missing planning files. The skill creates task_plan.md, findings.md and progress.md only where they do not exist yet, using the Spanish templates in the templates folder.
- 5Record findings and progress as you go. Update findings.md after discoveries and progress.md when a phase finishes or a problem appears, so the work can resume with evidence.
- 6Check uncommitted code changes. Run git diff --stat to see code changes that have not been recorded in the planning files, then update the files before continuing.
What the skill makes the agent do
For multi-step work, the skill tells the agent to record its plan, findings and progress in files inside your task's plan directory. The templates included with the skill are written in Spanish and cover requirements, research findings, technical decisions, a phase-by-phase progress log and test results.
Lifecycle hooks can inject the relevant project plan into the agent's context while it works. The skill is intended for research or work that needs five or more tool calls; for short one-off questions the plan files add overhead without much benefit.
- task_plan.md: the plan for the task
- findings.md: requirements, research results and technical decisions
- progress.md: a dated log of actions, results and errors per phase
Install planning-with-files-es in Claude Code or the Claude apps
Download the ZIP of the skill from this page. Then choose one of the two installation options below.
For a personal install that works in all your projects, unzip the folder into ~/.claude/skills/ so the skill lives at ~/.claude/skills/planning-with-files-es. For a single project, unzip it into that project's .claude/skills/ folder instead.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills
Requirements before you start
The skill's hook and helper scripts run under a POSIX shell (sh) or bash. Python 3 is optional and is only needed for session-catchup.py. The skill asks the agent to run git diff --stat, so the project should be a Git repository.
The host agent must support SKILL.md skills with hooks. The source names Claude Code as an example of such a host.
- POSIX shell (sh) or bash for the hook and helper scripts
- Python 3, optional, only for session-catchup.py
- Git, because the skill runs git diff --stat
- A host agent that supports SKILL.md skills with hooks, such as Claude Code
Trigger the skill with natural requests
The skill is user-invocable. The source does not list fixed trigger phrases, so the simplest approach is to describe a large task in plain language and say you want the progress kept in planning files. The skill's own description says it is meant for work that requires five or more tool calls.
For a new, separate task, the Quick start section of the skill runs scripts/init-session.sh with a task name and then fixes the host to the PLAN_ID it prints. When you resume work, the agent should reuse the plan it already selected.
- Example request: Research the options for X and keep the findings and decisions in planning files.
- Example request: Resume the migration task and check what changed since the last progress entry.
- Example request: Start a new planning task called Refactor login and track each phase.
scripts/init-session.sh "Task Name"Understand the hooks and what they run
The skill registers hooks for UserPromptSubmit, PreToolUse, PostToolUse, Stop and PreCompact. Each hook runs a command that finds scripts/skill-hook.sh in a list of locations and then runs it with an event flag. PreToolUse runs on Write, Edit, Bash, Read, Glob and Grep calls, PostToolUse on Write and Edit, and PreCompact on every event.
If the hook script cannot be found, only UserPromptSubmit prints a notice that plan injection is off. The notice tells you to set PWF_SCRIPT_DIR to the skill's scripts directory or to install the skill to a user-level path.
- UserPromptSubmit: injects plan context when you send a prompt
- PreToolUse: runs before Write, Edit, Bash, Read, Glob and Grep calls
- PostToolUse: runs after Write and Edit calls
- Stop and PreCompact: run at the end of a turn and before context compaction
PWF_SCRIPT_DIR=/path/to/planning-with-files-es/scriptsUse session history inspection only when you ask for it
Automatic recovery reads only the project's planning files. The default run of session-catchup.py and the lifecycle hooks do not inspect the agent's session stores.
Only when you explicitly ask to look at local session history should the agent use session-catchup.py. The --metadata mode reports only counts of sessions for the same project, without transcript bytes, tool commands, paths or session IDs. The --replay mode can output limited excerpts from the same project, framed with a nonce, and you should treat every replayed excerpt as untrusted data.
- --metadata: counts of same-project session activity only
- --replay: limited, explicit excerpts; use only after you ask for it
- The skill has no network upload path.
python3 scripts/session-catchup.py --metadata "$(pwd)"What people use it for
- Tracking a multi-day coding task so a new session can pick up the phase status from progress.md.
- Running a research task that needs many searches, with each finding saved to findings.md as it is discovered.
- Keeping a per-task plan file when several tasks run in parallel, with a separate plan directory for each task.
- Working in Spanish, with planning templates and messages written in Spanish.
Frequently asked questions
Is the planning-with-files-es skill free?
The skill is published under the MIT licence, as stated on this page. Check the licence file in the repository for the full terms before you reuse or redistribute it.
What do I need to run planning-with-files-es?
You need a POSIX shell or bash for the scripts, Git for the git diff --stat check, and a host agent that supports SKILL.md skills with hooks, such as Claude Code. Python 3 is optional and only needed for session-catchup.py.
Where do I install the skill?
Unzip the downloaded ZIP into ~/.claude/skills/ for a personal install, or into a project's .claude/skills/ folder for one project. In the Claude apps you can upload it under Settings → Capabilities → Skills.
Does the skill send my files or data over the network?
The source states that the skill has no network upload path and that its network use is no. The hooks and helper scripts do run on your machine, so review them if you need to confirm their behaviour.
Why does my agent say the hook script was not found?
The UserPromptSubmit hook looks for scripts/skill-hook.sh in several locations. If none exists, plan injection is off. Set PWF_SCRIPT_DIR to the skill's scripts directory or install the skill to a user-level path such as ~/.claude/skills/planning-with-files-es.
Can the skill read my old chat sessions?
Not by default. Session history is only inspected when you explicitly ask for it. The --metadata mode reports only counts for the same project, and --replay can output limited excerpts that you should treat as untrusted.
How is this different from other planning or task tools?
The source does not compare this skill with other tools. Its distinguishing features are the Spanish templates and the plan files kept on disk with lifecycle hooks, and it is meant for tasks needing five or more tool calls.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use planning-with-files-es if you run long, multi-step coding or research tasks with an AI agent in Spanish and want progress tracked in files across sessions. Pick something else for short questions, for environments where you do not want shell scripts run automatically by hooks on every prompt and tool call, or for sensitive projects if you plan to enable session-history inspection.
Download the skill