flockion_agent_design

A guide for designing Flockion agents and agent teams with defined jobs, inputs, outputs, tools, permissions, memory, safeguards, and failure handling.

In plain words
What is it for?
Use it when creating or improving agents, teams, prompts, tools, orchestration, handoffs, approval points, and run monitoring.
Why use it?
It helps turn loosely defined prompt-based assistants into systems whose behavior, side effects, costs, and risky actions can be reviewed.

Skill for Claude CodeCodex

Part of the flockion plugin — 15 skills shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/error505/flockion_ai_engineering/agent-design
Any agent
npx skills add error505/Flockion_AI_Engineering --skill agent-design
Clone the repo
git clone --depth 1 https://github.com/error505/Flockion_AI_Engineering

Made for: Claude Code, Codex.

Or install flockion, the plugin that ships this one along with the rest of its 15 skills.

Per session 78 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 724 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00078 $0.00724
Opus 5 $0.00039 $0.00362
Sonnet 5 $0.00016 $0.00145
Haiku 4.5 $0.00008 $0.00072

Measured 2d ago against content hash ab87f33e5cdf, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

flockion_agent_design scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/agent-design/SKILL.md · 170 lines

How it starts

The opening of the file, as written. The whole thing — 170 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Flockion Agent Design

You design agents that can be trusted, reused, forked, and operated.

Agents are products, not prompt demos.

An agent must have:

  • one clear job
  • clear input
  • clear output
  • narrow tools
  • safe permissions
  • explicit failure behavior
  • observable runs
  • clear value to the user

Agent Design Output

For a single agent, use:

agent name:
purpose:
when to use:
input:
output:
tools:
memory:
knowledge:
guardrails:
human approval:
failure behavior:
observability:
cost control:
example input:
example output:

For an agent team, use:

team name:
purpose:
user input:
final output:
agents:
orchestration:
handoffs:
shared memory:
tools:
approval points:
failure behavior:
audit trail:
cost controls:
marketplace positioning:
example run:

Agent Rules

  • One agent, one job.
  • Do not create an agent for simple deterministic logic.
  • Use code for rules, LLM for reasoning, language, extraction, ranking, and summarization.
  • Tools must be narrow.
  • Tool permissions must match the agent role.
  • Agents must not hide side effects.
  • Risky actions need human approval.
  • Final output must be structured.
  • Marketplace agents need clear sample input and output.
  • Every production agent needs observability.

Tool Rules

Each tool must define:

tool name:
purpose:
allowed actions:
blocked actions:
required input:
output:
risk level:
approval required:

Memory Rules

Use memory only when it improves future runs.

Do not store:

  • secrets
  • unnecessary personal data
  • temporary execution data
  • sensitive data without clear need

Memory must have a reason.

Orchestration Patterns

Choose the simplest pattern:

  1. Single agent
  2. Router
  3. Sequential team
  4. Parallel review team
  5. Human approval step
  6. Multi-agent debate
  7. Long-running workflow

Do not use multi-agent orchestration when one agent is enough.

Guardrails

Always define:

  • what the agent must do
  • what it must not do
  • when it must ask for clarification
  • when it must refuse
  • when it must escalate to a human
  • what format it must return

Read the full file on GitHub · 170 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 170 lines · 78 tokens per session scan A ab87f33e5cdf

Subscribe to this mod's changes

flockion_agent_design is a skill published in the GitHub repository error505/Flockion_AI_Engineering (5 stars, last pushed 2mo ago), licensed MIT. It adds 78 tokens to every session and 724 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

frontier

Execute any task at frontier quality. Three layers; checkable domain standards for all 21 crafts that lift even a single response (quick), best-of-N candidates for creative work, and a convergence loop with a strong-model taste gate for work that must be right (full). Self-contained; bundles the protocol, every craft…

apoorvjain25/frontier · 187 tokens

gold-standard

World-class completeness audit — score a project's rules/standards/features against best-in-class exemplars, name the gaps, fill missing rules, adopt as binding, then offer to conform existing code. Triggers on keywords: "/gold-standard", "gold-standard", "audit rules", "are we world-class", "fill gaps", "complete our…

HetCreep/CoalMine · 80 tokens

drift-canary

Compatibility and schema drift canary — checks for database schema migration safety, breaking API contract changes, serializable payload mismatches, and backward compatibility drift. Triggers on keywords: "/drift-canary", "drift-canary", "contract drift", "breaking changes". Use when changing DB schemas, API…

HetCreep/CoalMine · 78 tokens

resilience-audit

Failure-mode audit (FMEA for software) — for each way the system can fail (network, storage, partial completion, crash, concurrency, bad input), check whether code DETECTS, HANDLES, RECOVERS, and COMMUNICATES it. Triggers on: "/resilience-audit", "resilience-audit", "FMEA audit". Use when touching network, storage…

HetCreep/CoalMine · 119 tokens

rot-canary

Code-health scan — dead code, bug-prone logic, resource leaks, concurrency bugs, silent failures, input-boundary issues, doc rot. Triggers on: "/rot-canary", "rot-canary", "code-health" (legacy aliases: "/rotcanary", "rotcanary"). Auto-runs at session end on touched files (QUICK, report only) via platform hooks …

HetCreep/CoalMine · 114 tokens

source-grounding

Verify version-sensitive facts against live authoritative sources before asserting them in code or answers. Triggers on: "/source-grounding", "source-grounding", "sourcing". Standing rule — always active via CLAUDE.md. Invoke for deep verification work (API signatures, CVEs, model IDs, auth flows, deprecated patterns…

HetCreep/CoalMine · 74 tokens