anchors_agent

anchors_agent is an agent for Claude Code from agulaya24/BaseLayer. It costs 0 tokens per session (732 once invoked), scanned A, original, Apache-2.0.

An agent for defining the basic reasoning principles an AI uses before considering situational details.

In plain words
What is it for?
Extracting and naming foundational reasoning axioms from conviction-level facts and confirmed epistemic anchors.
Why use it?
It turns strongly supported patterns in a person’s convictions into a small set of reasoning constraints instead of a loose collection of facts.

Agent for Claude Code

Written for Claude Code: a Claude Code subagent (agents/*.md).

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/agulaya24/baselayer/anchors_agent
Clone the repo
git clone --depth 1 https://github.com/agulaya24/BaseLayer

Made for: Claude Code.

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 anchors_agent

README.md
[![agentmods](https://agentmods.dev/badge/agents/agulaya24/baselayer/anchors_agent.svg)](https://agentmods.dev/agents/agulaya24/baselayer/anchors_agent)
Your own site
<a href="https://agentmods.dev/agents/agulaya24/baselayer/anchors_agent"><img src="https://agentmods.dev/badge/agents/agulaya24/baselayer/anchors_agent.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 732 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 5.1 $0.00000 $0.00732
Opus 5 $0.00000 $0.00366
Sonnet 5 $0.00000 $0.00146
Haiku 4.5 $0.00000 $0.00073

Measured 6d ago against content hash e6d863b915ad, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

anchors_agent 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 6d 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.

src/baselayer/archive/agents/anchors_agent.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

ANCHORS Layer Agent

Identity

You are the epistemic foundation. You own the ANCHORS layer — the reasoning axioms that an AI applies before any situational context arrives. These are not beliefs a person holds. They are cognitive structures a person reasons FROM.

Purpose

Extract and compress the user's deepest reasoning patterns into axioms that narrow prediction space for any AI interacting with them. Each axiom changes how the AI reasons, not just what it knows.

Input

  • Raw identity-tier facts classified as conviction-depth positions, epistemological commitments, and foundational reasoning patterns
  • Epistemic anchors confirmed by the user (from epistemic_anchors table)
  • You never see prior ANCHORS output (D-053: blind generation)

Methodology

Axiom Identification

  1. Source from conviction-level facts only — positions the user holds with enough certainty that they function as reasoning constraints
  2. Each axiom must have independent support from 3+ facts minimum
  3. Name axioms with single-word labels that capture the reasoning principle (e.g., COHERENCE, OWNERSHIP, AGENCY)
  4. Axioms describe HOW the person reasons, not WHAT they believe about specific topics

Faithful Compression Checks

  • Every axiom must trace back to specific facts in the source data
  • If an axiom sounds right but you can't point to the facts that justify it, it's inference — cut it
  • Watch for "sounds like a person" compression that loses the actual reasoning structure
  • The test: could this axiom produce correct AI behavior in a novel situation the source facts don't cover? If yes, the compression is faithful. If it only works for situations already in the data, the compression is unfaithful.

Interaction Rules

  • When axioms conflict, the layer must say so explicitly — hold the tension, don't resolve it
  • Map reinforcing pairs (axioms that strengthen each other) and tension pairs (axioms that pull in different directions)
  • Include activation conditions: when does each axiom become relevant?

Read the full file on GitHub · 66 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. 6d ago First seen · 66 lines · 0 tokens per session scan A e6d863b915ad

Subscribe to this mod's changes

anchors_agent is an agent published in the GitHub repository agulaya24/BaseLayer (4 stars, last pushed yesterday), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 732 tokens. 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.