/references/coverage-audit.md

Coverage Audit

This file maps the required agent-harness knowledge areas to the Markdown files in this skill. Use it to confirm that no major design topic was omitted before applying the skill to a real agent architecture.

Required coverage map

Topic	Covered in	Notes
General-purpose agent harness	SKILL.md, architecture.md	Treats coding as one domain among many.
Coding-agent harness overlay	coding-agents.md, mvp-agent-blueprint.md, tools-and-permissions.md, security-observability.md, evals.md, checklists.md	Optional domain overlay for repository-reading, patching, validating, reviewing, migration, dependency, test, and docs-sync agents.
Agent-legible environment and feedback loops	agent-legibility-feedback-loops.md, architecture.md	Covers source-of-truth knowledge bases, validation signals, mechanical invariants, throughput, and entropy cleanup.
Agentic loop	agentic-loop.md	Includes canonical loop, invariants, budgets, retries, provider-neutral variants, and termination.
Goal-like loop	agentic-loop.md, planning-and-goals.md	Includes objective, done condition, budget, checkpoints, progress log, validation, and stop rules.
Planning mode	planning-and-goals.md	Covers read-only planning, plan artifact, approval, execution after approval, and plan-validate-execute.
Workflow orchestration	workflow-orchestration.md, architecture.md, planning-and-goals.md, checklists.md	Covers planner-generated workflows, work packets, worker and verifier contexts, integration, durable workflow state, budgets, approvals, and anti-patterns.
Auto context and compaction	context-memory-compaction.md	Covers context tiers, scoped instruction loading, retrieval, compaction triggers, handoff summaries, and rehydration.
Prompt caching and cost control	prompt-caching-and-cost.md, context-memory-compaction.md, provider-api-patterns.md	Covers stable-prefix design, deterministic serialization, provider cache fields, TTL/retention notes, compaction/cache tradeoffs, and monitoring.
Skills attachment	skills-and-connectors.md, SKILL.md	Covers Agent Skills structure, progressive disclosure, trigger descriptions, governance, and evals.
MCP and external connectors	skills-and-connectors.md	Covers resources/prompts/tools, staged loading, namespacing, authorization, deferred tool loading, and code-execution patterns.
System prompts and instructions	system-prompts-instructions.md	Covers authority hierarchy, runtime reminders, injection boundaries, and prompt templates.
Tool design	tools-and-permissions.md	Covers schemas, risk taxonomy, structured outputs, result limits, errors, sandboxing, secrets, and tool visibility.
Permissions and approvals	tools-and-permissions.md, security-observability.md	Covers permission matrix, draft/commit split, approval records, and policy enforcement.
Provider API differences	provider-api-patterns.md	Covers OpenAI Responses-style APIs, Chat Completions-style/OpenAI-compatible APIs, Anthropic APIs, hosted tools, adapters, streaming, and state.
Security	security-observability.md	Covers threat model, guardrails, prompt injection, approvals, launch gates, and incidents.
Observability	security-observability.md	Covers traces, events, token/cost/latency, and replay.
Evals	evals.md, coding-agents.md, checklists.md	Covers task success, tool precision, adversarial tests, trace grading, regression tests, and eval-driven launch criteria.
Implementation checklist	checklists.md	Includes design, tool, permission, context, planning, goal, skill, connector, eval, and rollout checklists.

Required language and scope checks

• The skill is provider-neutral.
• The skill refers to OpenAI, Anthropic, and OpenAI-compatible APIs only where provider-specific API patterns matter.
• The skill does not depend on coding-agent-specific assumptions; the coding-agent MVP profile is an optional domain overlay.
• The skill contains only Markdown files.
• The skill includes prompt-cache architecture and cache-hit monitoring.
• The skill includes agent-legibility, knowledge-base, feedback-loop, and entropy-management practices.
• The skill includes workflow orchestration as a generic harness pattern without depending on a vendor-specific runtime.
• The skill uses progressive disclosure: SKILL.md is the entry point; detailed guidance is in focused reference files.

Minimum file set

agents-best-practices/
  SKILL.md
  references/
    architecture.md
    agent-legibility-feedback-loops.md
    coding-agents.md
    agentic-loop.md
    tools-and-permissions.md
    workflow-orchestration.md
    context-memory-compaction.md
    prompt-caching-and-cost.md
    planning-and-goals.md
    skills-and-connectors.md
    system-prompts-instructions.md
    provider-api-patterns.md
    security-observability.md
    evals.md
    checklists.md
    coverage-audit.md

MVP agent blueprint generation

Covered in SKILL.md, references/mvp-agent-blueprint.md, references/architecture.md, and references/checklists.md. The skill now explicitly instructs assistants to produce a domain-specific MVP harness blueprint when the user asks to make or build an agent. The blueprint includes agentic loop, tool registry, permissions, context and memory, auto-compaction, planning mode, goal-like loop criteria, skills, MCP/external connectors, prompt caching, cost-aware context, observability, dedicated eval strategy, and launch path.