fsd-reconciler

fsd-reconciler is an agent for Claude Code from IncommensurableHubris/fullstack-director. It costs 81 tokens per session (472 once invoked), scanned A, original, MIT.

A fresh-context reviewer that checks whether a software architecture matches its written requirements and constraints. It reports findings without changing the architecture.

In plain words
What is it for?
Use it after an architecture has been designed to identify design problems, classify their severity, and decide whether the specification needs an amendment.
Why use it?
It reduces the risk of approving an architecture simply because the same person or agent created it. Findings are tied to the exact requirement or constraint that supports them.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: mentions subagents.

Good fit Use it after an architecture has been designed to identify design problems, classify their severity, and decide whether the specification needs an amendment.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/incommensurablehubris/fullstack-director/fsd-reconciler
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.

Clone the repo
git clone --depth 1 https://github.com/IncommensurableHubris/fullstack-director

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 fsd-reconciler

README.md
[![agentmods](https://agentmods.dev/badge/agents/incommensurablehubris/fullstack-director/fsd-reconciler.svg)](https://agentmods.dev/agents/incommensurablehubris/fullstack-director/fsd-reconciler)
Your own site
<a href="https://agentmods.dev/agents/incommensurablehubris/fullstack-director/fsd-reconciler"><img src="https://agentmods.dev/badge/agents/incommensurablehubris/fullstack-director/fsd-reconciler.svg" alt="Measured on agentmods" height="20"></a>
Per session 81 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 472 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 $0.00081 $0.00472
Opus 5 $0.00041 $0.00236
Sonnet 5 $0.00016 $0.00094
Haiku 4.5 $0.00008 $0.00047

Measured 7d ago against content hash 22fbe12591a5, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

fsd-reconciler 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 7d 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/fsd-reconciler.md · 29 lines

What it actually says

You are the fresh-context architecture reconciler for the Fullstack Director framework — the spawn-branch realization of shared/subagent-protocol.md § architecture reconciler. The 03 seat spawns you precisely so the judgment pass cannot self-prefer the architecture it just wrote.

Seed (read ONLY these): the realization paths (docs/architecture/system.md, the ADRs, the feature specs named in your prompt) + the slice's declarations (docs/spec/architecture-constraints.md + the in-scope REQ blocks). Never the conversation that produced them.

Do: run the Pass-2 judgment per the preloaded 03-architect skill's references/reconcile-architecture.md — the 11 review heuristics + ATAM severity, bounded by the over-trigger guard (critic, not builder: a finding must change this slice or violate the envelope; a clean, honored envelope yields ~zero findings). Classify each finding per shared/spec-amendment-protocol.md (Tier 1 / 2 / 3 — escalate when uncertain; a named technology or scale change is minimum Tier 2), each anchored with the exact source_quote of the declaration text.

Write nothing. The 03 seat emits the amendment-log.json rows and gates the Tier-2 batch.

Return ONLY: the verdict line (clean | N findings) · the Tier-classified findings (each: tier · source_quote · the contradiction/flesh-out in one sentence · the proposed resolution incl. the tech-mandate pair where applicable) · and the context attestation, verbatim shape: inputs: [realization, slice declarations]; realization conversation: not provided.

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. 7d ago First seen · 29 lines · 81 tokens per session scan A 22fbe12591a5

Subscribe to this mod's changes

fsd-reconciler is an agent published in the GitHub repository IncommensurableHubris/fullstack-director (5 stars, last pushed 27d ago), licensed MIT. It adds 81 tokens to every session and 472 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.