Repository navigation
feat: Enforce mandatory 4-skill initialization protocol - #3
Merged
Merged
Conversation
## Changes - Added AI_INITIALIZATION.md defining non-skippable 4-phase execution sequence - Updated AI_CONSTITUTION.md to bind initialization protocol as layer 2 authority - Updated copilot-instructions.md with verifiable output markers and pre-send gate - Enhanced _execution-precheck skill with 6-step self-check and truthfulness test - Enhanced _context-ack skill with diff-formatted footer and anti-fake-list rules - Removed deprecated EXECUTION_RULES.md (replaced by initialization protocol) ## Technical Details The initialization protocol enforces: 1. Phase 1 (_instruction-guard): Explicit file reading before every response 2. Phase 2 (_execution-precheck): 6-step pre-flight check including honesty verification 3. Phase 3 (_context-ack): Standardized format with required markers 4. Phase 4 (_file-output-guard): File operation validation Every response must now include: - First line: `✨ 已启用上下文校验` - Standalone salutation: `尊敬的主人：` - Footer: `已读指令` (only actual reads) - Footer: `已启用技能` (one per line, with evidence) - Footer: Repository status in diff format ## Testing - Verified all 4 phases execute in correct order - Confirmed all output markers present - Validated no fake list items in footer - Tested format compliance in new session ## Rationale Previous system had skill definitions and documentation but no enforcement mechanism. This creates a fail-closed protocol that cannot be bypassed, ensuring: - Consistent quality across all responses - Traceability of instruction compliance - Prevention of hallucinated skill usage - Honest self-assessment before every response --- 🤖 Generated by _git-commit skill
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
feat: Enforce mandatory 4-skill initialization protocol
Summary
This PR introduces a fail-closed initialization protocol that enforces 4 system skills on every AI response, eliminating the gap between skill documentation and actual execution.
Problem
The previous system had well-defined Tier 2 skills (_instruction-guard, _execution-precheck, _context-ack, _file-output-guard) but lacked an enforcement mechanism. Skills were loaded into context and documented, but weren't automatically executed, leading to:
Solution
Created a new AI_INITIALIZATION.md protocol that defines:
Non-skippable 4-Phase Sequence
.github/copilot-instructions.mdVerifiable Output Markers
Every response must include:
✨ 已启用上下文校验尊敬的主人：已读指令(only files actually read)已启用技能(one per line with evidence)仓库状态(branch, uncommitted, untracked)Pre-Send Gate
Mandatory checklist before every response:
.github/copilot-instructions.mdexplicitly read this turn_execution-precheck6-step self-check completed_context-ackformat fully applied_file-output-guardconstraints satisfiedChanges
.github/AI_INITIALIZATION.md- Mandatory initialization protocol.github/AI_CONSTITUTION.md- Bind initialization as layer 2 authority.github/copilot-instructions.md- Add enforcement rules and verifiable markers.agent/skills/_execution-precheck/SKILL.md- Add 6-step self-check with truthfulness test.agent/skills/_context-ack/SKILL.md- Add diff format, anti-fake-list rules.github/EXECUTION_RULES.md- Replaced by stronger initialization protocolTesting
Impact
Before: Skills existed but weren't enforced → inconsistent quality
After: Skills execute on every response → guaranteed consistency
Checklist
🤖 Generated by _pr-creator skill