# Coding goal guardrails

Establish the objective, constraints, evidence for completion, and a lightweight record of progress. The parent remains responsible for scope, authorization, goal state, and final acceptance while focused skills handle the actual coding tasks.

Package version: 0.1.0-beta.7.

Included with [Faye Coding Assistant](/plugins/faye-coding-assistant/).

## When to use

Use this when creating a new coding goal, explicitly resuming an interrupted goal, or requesting guardrails directly.

## What to provide

- The goal, scope, constraints, and done condition.
- Current Git state, permitted Git actions, and any existing ledger or resume context.

## Expected results

- Clear boundaries and an evidence-based completion standard.
- A lightweight ledger when needed, with current state, decisions, validation, and safe continuation context.

## Example requests

### Try this request

#### Codex

```
$coding-goal-guardrails For this coding goal: implement and validate the approved local release plan. Preserve existing changes, keep a concise ledger, and do not create commits.
```

#### Grok

```
/coding-goal-guardrails For this coding goal: implement and validate the approved local release plan. Preserve existing changes, keep a concise ledger, and do not create commits.
```

#### Antigravity

```
/coding-goal-guardrails For this coding goal: implement and validate the approved local release plan. Preserve existing changes, keep a concise ledger, and do not create commits.
```

## Requirements

- A new or explicitly resumed coding goal, or a direct guardrails request.
- Any branch or commit authority must be explicit or already established.

## Scope and limitations

- Guardrails do not replace implementation judgment or automatically invoke every coding skill.
- Routine continuation inside an already guarded active goal does not require reinvocation.
- The ledger records authoritative decisions; it does not independently decide scope or completion.

## Related guides and supporting files

- [Read-only coding analyst](/agents/faye-coding-assistant/faye-coding-analyst/) — Defines a bounded analysis role for exploration, diagnosis, review, planning, suite assessment, and handoff.
- [Implementation assistant](/agents/faye-coding-assistant/faye-coding-assistant/) — Defines a write-capable role for explicitly authorized, bounded implementation or repair.
- [Independent goal reviewer](/agents/faye-coding-assistant/faye-goal-fresh-review/) — Defines a read-only reviewer for material risks or conflicting completion evidence in a coding goal.
- [Faye Coding Assistant](/plugins/faye-coding-assistant/) — Understand repositories, plan and implement changes, diagnose failures, and review code with clear evidence.
- [Goal workflow boundaries](/references/faye-coding-assistant/anti-overlap-8af3974e/) — Separates goal guardrails from focused coding, utility, and GitHub workflows.
- [Goal Git policy](/references/faye-coding-assistant/git-handling-policy-ac867766/) — Explains how goal work preserves existing changes and applies explicit Git authorization.
- [Goal ledger template](/references/faye-coding-assistant/goal-ledger-template-9c7283e1/) — Provides a compact record of a coding goal's state, evidence, decisions, and continuation context.
- [Goal completion handoff](/references/faye-coding-assistant/handoff-template-1d3cf775/) — Structures the accepted outcome, validation, Git state, risks, and next action at goal handoff.
- [Bounded coding delegation](/references/faye-coding-assistant/subagent-routing-b82c29b0/) — Guides the use of focused read-only and write-authorized roles within a coding goal.

## Technical source

[Open original source](/marketplace/plugins/faye-coding-assistant/skills/coding-goal-guardrails/SKILL.md)

Source path: plugins/faye-coding-assistant/skills/coding-goal-guardrails