fix: guard constitution command against feature execution

[BenBtg]

Description

Mixed governance and feature directives can cause /speckit.constitution to modify application code while updating project governance. This adds an explicit scope guard to both the default and lean constitution prompts so non-governance intents are deferred to a conditional Next Actions section instead of executed.

The guard preserves constitution-driven dependent artifact updates and the existing user-controlled handoff. Native per-agent tool restrictions and guards for other commands remain separate follow-up work.

Fixes: #896

Testing

• Tested locally with uv run specify --help
• Ran existing tests with uv sync && uv run pytest
• Tested with a sample project (if applicable)
• Ran targeted constitution and lean preset tests with .venv/bin/python -m pytest tests/test_presets.py -q -k 'constitution_commands_guard or TestLeanPreset'

AI Disclosure

• I did not use AI assistance for this contribution
• I did use AI assistance (describe below)

Implemented and prepared on behalf of @BenBtg by GitHub Copilot (model: GPT-5.6 Sol, autonomous). Copilot reviewed issue #896, updated the prompts and regression tests, and ran the targeted test suite.

[Copilot]

Pull request overview

Adds governance-only scope guards to constitution commands, deferring feature requests to follow-up actions.

Changes:

• Guards core and lean constitution prompts against non-governance work.
• Adds regression assertions for guard wording.
• Preserves constitution-dependent artifact synchronization.

Show a summary per file

File	Description
templates/commands/constitution.md	Adds the core scope guard and deferred actions.
presets/lean/commands/speckit.constitution.md	Adds equivalent lean-preset protection.
tests/test_presets.py	Verifies both prompts contain guard directives.

Review details

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

• Files reviewed: 3/3 changed files
• Comments generated: 3
• Review effort level: Medium

[BenBtg]

Addressed the review round in commit 02fd60c: deferred follow-up commands now use __SPECKIT_COMMAND_SPECIFY__ in both prompt sources, and the regression test requires the placeholder while remaining whitespace-tolerant. Targeted constitution and lean preset tests pass.

Posted on behalf of @BenBtg by GitHub Copilot (model: GPT-5.6 Sol, autonomous).

[Copilot]

Review details

• Files reviewed: 3/3 changed files
• Comments generated: 1
• Review effort level: Medium

[BenBtg]

Addressed the latest review in commit 9fd5b39: the lean guard now classifies every input part and defers every non-governance intent, while retaining feature and build categories as examples. The shared regression assertion now enforces this rule for both prompts.

Posted on behalf of @BenBtg by GitHub Copilot (model: GPT-5.6 Sol, autonomous).

[Copilot]

Review details

• Files reviewed: 3/3 changed files
• Comments generated: 0 new
• Review effort level: Medium