> For the complete documentation index, see [llms.txt](https://wong-coupon.gitbook.io/flutter/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://wong-coupon.gitbook.io/flutter/my-ai/coding-agents.md).

# Coding Agents

Organize coding-agent rules, context, and skills; choose workflows and work safely in repositories.

Organize coding-agent rules, context, and skills; choose workflows and work safely in repositories.

## Articles

* [Share Rules Across Coding Agents](/flutter/my-ai/coding-agents/shared-provider-coding-agent-rules.md) — Project a finite rule catalog through thin provider adapters while keeping generated files separate from evidence that an agent actually discovered or followed them.
* [Progressive Context Loading](/flutter/my-ai/coding-agents/progressive-context-loading-token-optimization.md) — Reduce context through catalogs, budgets, and truncation receipts while preserving mandatory QA, freshness, and correctly scoped savings evidence.
* [Save AI Tokens Through Test Structure](/flutter/my-ai/coding-agents/test-splitting-ai-token-savings.md) — Split suites by behavior, keep fixtures independent, and route context; measure reading volume without inferring tokens or cost from line counts.
* [Repo-Native Skills as Executable Playbooks](/flutter/my-ai/coding-agents/repo-native-skill-engineering-playbook.md) — Turn repository guidance into finite manifests, plans, approvals, and receipts while containing paths, resources, capacity, and late results before host execution.
* [Guardrails for AI Feature Generation](/flutter/my-ai/coding-agents/feature-generator-ai-code-guardrails.md) — Wrap generators with immutable plans, no-overwrite checks, approvals, and write-attempt fencing while quarantining targets until ambiguous effects are reconciled.
* [Safe AI in a Dirty Worktree](/flutter/my-ai/coding-agents/safe-ai-dirty-worktree.md) — Bind edits to immutable snapshots, shared file claims, and CAS while keeping approval outside the agent, quarantining ambiguous effects, and never granting implicit commit or push rights.
* [Choosing BMAD or a Direct Agent](/flutter/my-ai/coding-agents/bmad-vs-direct-agent.md) — Use a host-verified risk vector to choose direct execution, BMAD planning, or a human decision without inferring execution authority.

[Buy Me a Coffee](https://buymeacoffee.com/ducmng12g) | [Support Me on Ko-fi](https://ko-fi.com/I2I81AEJG8)