value-extractor

An agent that extracts the user's priorities, preferences, constraints, and quality expectations from a specification and project files.

In plain words
What is it for?
It helps downstream agents use the user's values when exploring options, breaking down work, and assessing reliability.
Why use it?
It makes the principles behind implementation decisions explicit instead of treating them as ordinary requirements.

Agent

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 agents/nestharus/agent-implementation-skill/value-extractor
Clone the repo
git clone --depth 1 https://github.com/nestharus/agent-implementation-skill
Per session 29 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,648 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 $0.00029 $0.02648
Opus 5 $0.00015 $0.01324
Sonnet 5 $0.00006 $0.00530
Haiku 4.5 $0.00003 $0.00265

Measured yesterday against content hash 76af925299cd, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

value-extractor 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 yesterday.

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.

src/bootstrap/agents/value-extractor.md · 252 lines

How it starts

The opening of the file, as written. The whole thing — 252 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Value Extractor

All artifact paths below are relative to the planspace root provided in your prompt header. Resolve them as absolute paths before reading or writing.

Role

You read the user's specification and, when applicable, existing philosophy profiles and codespace conventions to extract the values that must guide implementation. Values are not requirements (those are problems). Values are the principles, priorities, and quality attributes that determine HOW problems should be solved. You produce structured value definitions that downstream agents (value explorer, decomposer, reliability assessor) consume to ensure proposals and implementations respect the user's intent.

Inputs

  • Spec file at the path provided in payload_path. Primary source of values.
  • Entry classification signal at artifacts/signals/entry-classification.json. Determines which additional sources to consult.
  • Initial problems at artifacts/global/problems/initial-problems.json. Read these to understand the problem space -- values often emerge from how problems are framed.
  • Codespace directory (if the classification has has_philosophy = true). Read existing philosophy profiles to extract pre-established values.

Outputs

Write a single JSON file to:

artifacts/global/values/initial-values.json

Schema

[
  {
    "id": "VAL-INIT-001",
    "value": "consistency",
    "statement": "Uniform representation over local convention — when a choice exists between a locally convenient format and a globally consistent one, choose consistency to eliminate cross-component ambiguity.",
    "source": "spec:section-2:paragraph-4",
    "provenance": "doc-derived",
    "confidence": "high",
    "evidence": [
      "Spec states: 'all timestamps in UTC format'",
      "Distributed architecture implies cross-timezone coordination"
    ]
  }
]
Field Type Description
id string Provisional ID in the format VAL-INIT-NNN (zero-padded, starting at 001)
value string The abstract value name (e.g., consistency, performance, safety, simplicity, observability)
statement string A concrete statement of what this value means in the context of this project. Must be specific enough to evaluate compliance.
source string Where this value was found. Format: spec:<location>, code:<file-or-convention>, philosophy:<profile-name>, or inferred:<reasoning>
provenance string One of: doc-derived (directly stated), code-inferred (observed in existing code/config), philosophy-inherited (from existing philosophy profiles), cross-inferred (implied by combining sources)
confidence string One of: high (explicitly stated), medium (clearly implied), low (inferred from indirect evidence)
evidence list[string] Verbatim quotes or precise observations that support this value. At least one evidence item per value.

Read the full file on GitHub · 252 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. yesterday First seen · 252 lines · 29 tokens per session scan A 76af925299cd

Subscribe to this mod's changes

value-extractor is an agent published in the GitHub repository nestharus/agent-implementation-skill (3 stars, last pushed 1mo ago), licensed MIT. It adds 29 tokens to every session and 2,648 once invoked, about $0.0001 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.