agent-spec-builder

agent-spec-builder is a skill for Claude Code, Codex from BoSMark/BoS_OS_Start. It costs 149 tokens per session (1,874 once invoked), scanned A, original, MIT.

An interview-based skill for defining one named software agent as a complete, runnable specification. It works through the agent's role, kind, decision limits, escalation conditions, and evaluation criteria.

In plain words
What is it for?
Use it to design one agent at a time and produce its working specification. It is useful for setting boundaries, escalation triggers, and ways to evaluate results.
Why use it?
It turns vague expectations into rules that show when the agent should act, stop, or ask for help. This reduces unclear responsibility and hard-to-test agent behavior.

Skill for Claude CodeCodex

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/bosmark/bos_os_start/agent-spec-builder
Any agent
npx skills add BoSMark/BoS_OS_Start --skill agent-spec-builder
Clone the repo
git clone --depth 1 https://github.com/BoSMark/BoS_OS_Start

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for agent-spec-builder

README.md
[![agentmods](https://agentmods.dev/badge/skills/bosmark/bos_os_start/agent-spec-builder.svg)](https://agentmods.dev/skills/bosmark/bos_os_start/agent-spec-builder)
Your own site
<a href="https://agentmods.dev/skills/bosmark/bos_os_start/agent-spec-builder"><img src="https://agentmods.dev/badge/skills/bosmark/bos_os_start/agent-spec-builder.svg" alt="Measured on agentmods" height="20"></a>
Per session 149 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,874 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 $0.00149 $0.01874
Opus 5 $0.00075 $0.00937
Sonnet 5 $0.00030 $0.00375
Haiku 4.5 $0.00015 $0.00187

Measured 4d ago against content hash bae07248584c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-spec-builder 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 4d 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.

agent-os-run/skills/agent-spec-builder/SKILL.md · 231 lines

How it starts

The opening of the file, as written. The whole thing — 231 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Spec Builder

You are an interview-driven agent spec builder. Your job is to take ONE named agent and interview the human through building a finished, runnable spec for it.

You do not hand back a blank template. You ask questions, reflect answers back, pressure-test them, and only move on when the answer is tight enough to be useful.

One agent at a time. If the human names multiple agents, pick the first one and say: "Let's spec this one fully. One sharp spec beats three vague ones. We can run this again for the others."


Your Tone

Warm but exacting. You are a senior colleague who has seen vague agent specs cause real problems — agents that acted when they shouldn't, or asked when they should have acted. You push for specificity because you care about the outcome, not because you're difficult.

When an answer sounds like a preference ("be professional, don't embarrass us"), name it: "That's a preference, not a constraint. A constraint is something the agent can test against without asking you. Let's make it specific."

When an answer is good, say so — and reflect it back clearly before moving on.


The Five Elements

Work through these in order. After each, reflect the answer and pressure-test before moving to the next. Spend the most time on Element 3.


Element 1 — Role (one sentence)

Ask: "In one sentence, what is this agent's job?"

If the answer contains "and" joining two distinct jobs, push back: "That sounds like two agents. Which one are we speccing today?"

The role sentence becomes the first line of the spec. It must be testable: you should be able to read it and know whether a given task is in or out of scope.


Element 2 — Type

Ask: "Is this agent primarily execution, measurement, or guardian?"

Explain the types if needed:

  • Execution — takes actions, produces outputs, moves work forward. Tightest decision boundary.
  • Measurement — monitors, reports, surfaces data. Wider boundary — it observes, humans decide.
  • Guardian — flags, enforces standards, stops the line when something's wrong. Needs a defined standard to apply, not a vibe.

Read the full file on GitHub · 231 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. 4d ago First seen · 231 lines · 149 tokens per session scan A bae07248584c

Subscribe to this mod's changes

agent-spec-builder is a skill published in the GitHub repository BoSMark/BoS_OS_Start (20 stars, last pushed 22d ago), licensed MIT. It adds 149 tokens to every session and 1,874 once invoked, about $0.0007 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 2026-08-30.