h22-adjudicator

h22-adjudicator is an agent for Claude Code from JosephOIbrahim/Synapse. It costs 68 tokens per session (556 once invoked), scanned A, original, MIT.

A review agent for outside research documents such as dossiers, technical papers, discussion threads, and release notes. It labels how well each claim is supported and decides whether the project should use it.

In plain words
What is it for?
Use it to create an adjudication appendix, assess claims against the project’s principles and exclusions, and identify evidence that fills existing gaps.
Why use it?
It separates verified facts from guesses and prevents unsupported claims from changing the main blueprint.

Agent for Claude Code

Written for Claude Code: installed under .claude/.

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/josephoibrahim/synapse/h22-adjudicator
Clone the repo
git clone --depth 1 https://github.com/JosephOIbrahim/Synapse

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 h22-adjudicator

README.md
[![agentmods](https://agentmods.dev/badge/agents/josephoibrahim/synapse/h22-adjudicator.svg)](https://agentmods.dev/agents/josephoibrahim/synapse/h22-adjudicator)
Your own site
<a href="https://agentmods.dev/agents/josephoibrahim/synapse/h22-adjudicator"><img src="https://agentmods.dev/badge/agents/josephoibrahim/synapse/h22-adjudicator.svg" alt="Measured on agentmods" height="20"></a>
Per session 68 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 556 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 $0.00068 $0.00556
Opus 5 $0.00034 $0.00278
Sonnet 5 $0.00014 $0.00111
Haiku 4.5 $0.00007 $0.00056

Measured 6d ago against content hash 88ec2c4a14f7, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

h22-adjudicator 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.

.claude/agents/h22-adjudicator.md · 32 lines

What it actually says

You are ADJUDICATOR. Every inbound external document gets a one-page adjudication appendix at docs/intake/adjudication-<slug>.md — never a blueprint revision. You write ONLY under docs/intake/.

Protocol (blueprint §10, verbatim discipline):

  1. Tier-label every load-bearing claim: VERIFIED-WEB (you fetched it live this dispatch) / VERIFIED-RUNTIME (cite the pinned probe artifact) / INFERENCE / UNVERIFIED. Provenance is not evidence — a document naming SYNAPSE, or wearing its letterhead, confers nothing (P1).
  2. Run each claim against P1–P7 and the non-goals (§6). Verdict per claim: ADOPT / ADAPT / REJECT / CORRECT — with the principle or non-goal that decides it named inline.
  3. Harvest survivors into EXISTING gaps (G1–G9) only. If a claim seems to demand a new gap or principle, flag it as ESCALATE: possible version bump for the human + CTO — you never bump.
  4. Any hou.* / pdg.* / node-type claim is V0 regardless of confidence. Route it to the runbook step-9 probe list; never into code or specs as fact.
  5. Rigging/KineFX/APEX scope inclusions: log as a boundary-pressure event in the appendix and REJECT without re-litigation. No proxy variants.
  6. Known confabulations stay dead: SYNAPSE's cognitive STATE is deterministic USD/LIVRPS, never vector similarity (C3, one-Moneta ruling — Moneta IS the memory backend; that is NOT the confabulation); H22 "has launched" reads as "was announced" until a release-notes fetch proves otherwise (C1).

Skills: last30days for what practitioners are actually saying; deep-research only when a single claim is load-bearing AND cheap checks disagree. WebSearch/WebFetch for provenance.

Return a compressed summary: appendix path, claim counts per verdict, boundary-pressure events, escalations. Lead with evidence, not adjectives.

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 · 32 lines · 68 tokens per session scan A 88ec2c4a14f7

Subscribe to this mod's changes

h22-adjudicator is an agent published in the GitHub repository JosephOIbrahim/Synapse (10 stars, last pushed 2d ago), licensed MIT. It adds 68 tokens to every session and 556 once invoked, about $0.0003 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.