Instruction file imported from HamidRezaRezaeiGitHub/man-agent-ment (
.github/instructions/ai-assets.instructions.md). Copyright stays with the author.
AI Asset Instructions
- Treat instruction files as product surface area: concise, deterministic, and easy to audit.
- Avoid duplicating long guidance across agent-specific files. Prefer short adapters that point to the shared contract.
- When a tool cannot import or follow references reliably, include the critical behavior directly in that tool's native file.
- Include source links in docs when describing external agent behavior.
- For dispatcher and planning-workflow changes, keep the canonical installable behavior in
packs/default/ai/workflows/workflow-dispatch.mdandpacks/default/ai/workflows/requirement-planning.md, and keep native adapters short. - For skill changes, keep canonical portable behavior in
packs/default/ai/skills/<skill>/SKILL.md; keep agent-native skill or instruction files thin and pointed at the canonical skill. - For wiki-documentation changes, keep the canonical installable behavior in
packs/default/ai/workflows/wiki-documentation.md. - For command, CI, and testing guidance, keep project-agnostic behavior in
packs/default/ai/workflows/and avoid technology-specific commands. - For vibe-coding translation, keep behavior project-agnostic and focused on clarifying intent before implementation.
- For architecture guidance, keep behavior focused on current system boundaries, data flow, contracts, decisions, and wiki documentation.
- For code review guidance, prioritize behavioral defects, architecture fit, tests, CI, security/privacy, operations, and documentation drift over style.
- For requirement handoff changes, keep
PLAN.mdfocused on control flow andFINDINGS.mdfocused on reusable requirement-specific context.