create-goal

A command that turns a clarified objective into a prompt for an autonomous coding loop, along with files for tracking its progress. It uses project documentation to ground the objective in the codebase's terms and constraints.

In plain words
What is it for?
Use it when you want an autonomous agent to pursue one larger objective from start to finish, rather than planning a sequence of smaller tasks for manual review.
Why use it?
It helps define a continuous piece of work with a measurable stopping point before handing it to an automated loop. This keeps the loop focused on a stated outcome and feedback process.

Command for Claude Code

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 commands/ayoubben18/ab-method/create-goal
Clone the repo
git clone --depth 1 https://github.com/ayoubben18/ab-method

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 608 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 $0.00000 $0.00608
Opus 5 $0.00000 $0.00304
Sonnet 5 $0.00000 $0.00122
Haiku 4.5 $0.00000 $0.00061

Measured 2d ago against content hash 3b5eeddb423b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

create-goal 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.

.claude/commands/create-goal.md · 45 lines

How it starts

The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Create Goal

Description

Produce a ready-to-run goal prompt for an autonomous /goal loop (Claude Code or Codex), using the AB Method's grilling + domain-grounding workflow.

Usage

/create-goal

Behavior

Loads and executes the create-goal workflow from .ab-method/core/create-goal.md

This workflow will:

  1. Always invoke grill-with-docs to interview the user and pin down the objective, a verifiable measurable end state, the feedback loops that steer the loop, and constraints (no skip)
  2. Read UBIQ + CONTEXT + tech-stack + patterns + ADRs to ground the goal in canonical terms
  3. Write docs/goals/<goal-name>/goal.md — the prompt you paste into /goal
  4. Write docs/goals/<goal-name>/progress-tracker.md — an empty notes log the loop fills as it runs
  5. Stop. /create-goal produces the files; the /goal loop does the work

/create-goal vs /create-task

  • /create-goal — one continuous objective with a verifiable stop condition, handed to an autonomous loop you walk away from
  • /create-task — broken into missions you review and run through tdd yourself, one at a time

Workflow Details

  • Always grillgrill-with-docs runs on every invocation
  • Verifiable end state required — if "done" can't be checked by a command, the workflow keeps grilling or recommends /create-task
  • Feedback loops steer the run — the grill captures the checks (tests, type-check, build, visual checks) the /goal loop runs every iteration to self-correct
  • The loop is told about the trackergoal.md explicitly instructs the /goal loop to maintain progress-tracker.md, logging only important discovered facts (no generic narration)
  • References docs, doesn't inline them — the loop reads UBIQ/CONTEXT/architecture in-repo
  • Black boxes are briefed, never decided (rare) — a question you genuinely can't answer yet is parked in docs/goals/<goal-name>/unresolved-questions.md, and goal.md gets a Black boxes section telling the loop to build the agreed placeholder behind a TODO(UQ-n) seam and never to answer the question itself. The measurable end state may never depend on an open question — if it does, the goal isn't ready

Read the full file on GitHub · 45 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 · 45 lines · 0 tokens per session scan A 3b5eeddb423b

Subscribe to this mod's changes

create-goal is a command published in the GitHub repository ayoubben18/ab-method (187 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 608 tokens. 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 2026-08-30.