task-execution

A method for carrying out an approved, pre-defined task list without expanding its scope.

In plain words
What is it for?
Use it to read and verify each task, implement only that task, run its checks, mark it complete in tasks.md, and commit the task group.
Why use it?
It prevents implementation work from drifting into unrelated improvements or changing decisions already made in the specification.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/orlando-japan/claude-code-setting/task-execution
Any agent
npx skills add orlando-japan/claude-code-setting --skill task-execution
Clone the repo
git clone --depth 1 https://github.com/orlando-japan/claude-code-setting

Made for: Claude Code, Codex.

Per session 32 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 788 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00032 $0.00788
Opus 5 $0.00016 $0.00394
Sonnet 5 $0.00006 $0.00158
Haiku 4.5 $0.00003 $0.00079

Measured 2d ago against content hash b80e9a1d1559, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

task-execution scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

templates/extra/skills/task-execution/SKILL.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Task execution

You have a task list. Your job is to execute it faithfully — not to expand it, not to improve it, not to second-guess the spec while implementing.

Protocol

  1. Read the next task fully before writing code. Understand the verification before the implementation.
  2. Verify understanding — if anything in the task is ambiguous, stop and ask before coding. Asking after implementation wastes work.
  3. Implement just that task. Not the next one, not related cleanup, not "while I'm here" improvements.
  4. Run the task's verification. It must pass for this task to be done.
  5. Mark the task done in tasks.md — edit the file, don't just say it in chat.
  6. Commit. One task group per commit, referencing the task IDs or spec slug.
  7. Next task.

Golden rules

Don't expand scope

If task 3 says "add a redirect for logged-out users hitting /settings," the task is that redirect. Not redirects for all protected routes. Not a refactor of the middleware. Not better error messages. Just the redirect.

Don't re-plan

The spec was approved. Re-litigating design decisions during implementation wastes time and destabilizes the work. If you spot a real problem with the design, stop and surface it — don't silently change course.

Don't silently fix the spec

If a task reveals the spec is wrong — a requirement that can't be met, a scenario that's self-contradictory, a design that breaks an existing feature — stop. Report the discrepancy. Let the user decide whether to update the spec or adjust the implementation. Don't patch it invisibly.

Don't batch verification

Run each task's verification immediately after implementing it. If you batch and verify at the end, you can't tell which change broke what.

Don't skip tasks

Tasks are ordered for a reason. Skipping ahead risks building on foundations that aren't in place yet. If a task is genuinely impossible to do now, surface it.

When to stop

  • A task's verification fails and you can't diagnose it in ~30 min → stop, ask the user or the original spec author.
  • A task reveals a spec contradiction → stop, surface, wait for decision.
  • The task takes longer than estimated by >2x → stop, re-estimate, let the user decide whether to continue.
  • You're about to touch a file the spec didn't anticipate → stop, ask.

Read the full file on GitHub · 66 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 66 lines · 32 tokens per session scan A b80e9a1d1559

Subscribe to this mod's changes

task-execution is a skill published in the GitHub repository orlando-japan/claude-code-setting (2 stars, last pushed 3mo ago), licensed MIT. It adds 32 tokens to every session and 788 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories