backpressure

backpressure is a skill for Claude Code, Codex from mcwalrus/oppertunity-party. It costs 138 tokens per session (2,581 once invoked), scanned C, original, MIT.

Agent backpressure is a feedback system that tells a coding agent when its output is wrong and how to correct it. The idea comes from engineering systems where resistance controls a flow.

In plain words
What is it for?
Use it to evaluate or design the checks around an agent workflow. It helps diagnose why an agent is not correcting itself and decide which tools or feedback signals it needs.
Why use it?
It reduces the need for a developer to repeatedly point out missed checks, broken code, or other predictable mistakes. Good feedback is automated, immediate, specific, and hard to ignore.

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/mcwalrus/oppertunity-party/backpressure
Any agent
npx skills add mcwalrus/oppertunity-party --skill backpressure
Clone the repo
git clone --depth 1 https://github.com/mcwalrus/oppertunity-party

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 backpressure

README.md
[![agentmods](https://agentmods.dev/badge/skills/mcwalrus/oppertunity-party/backpressure.svg)](https://agentmods.dev/skills/mcwalrus/oppertunity-party/backpressure)
Your own site
<a href="https://agentmods.dev/skills/mcwalrus/oppertunity-party/backpressure"><img src="https://agentmods.dev/badge/skills/mcwalrus/oppertunity-party/backpressure.svg" alt="Measured on agentmods" height="20"></a>
Per session 138 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,581 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 1 finding. 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 $0.00138 $0.02581
Opus 5 $0.00069 $0.01290
Sonnet 5 $0.00028 $0.00516
Haiku 4.5 $0.00014 $0.00258

Measured 3d ago against content hash 7fca94d24a28, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

backpressure scanned grade C with 1 finding 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 3d 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

(`rm -rf .docker/volumes/`) and isolated test databases serve this purpose.
.agents/skills/backpressure/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 Backpressure

What Backpressure Means Here

In agent-driven development, backpressure is the resistance signal that tells an agent its output was wrong and exactly how to fix it. The term is borrowed deliberately: just as hydraulic backpressure resists flow until upstream pressure builds to push through, a well-designed feedback system resists bad output until the agent corrects it.

Your backpressure as an engineer is finite. Every time you type a message telling an agent it missed an import, forgot a semicolon, or broke a layout, you are spending it. That is time taken away from architecture, judgment, and problems machines cannot solve. Any feedback a machine can generate automatically is feedback you should never be giving manually.

The central question to ask about any agent workflow: am I wasting my backpressure?


The Four Properties of Good Backpressure

Feedback signals that work share these properties. Use them to evaluate any check in your pipeline:

Property What it means What bad looks like
Automated Generated by a machine, not you You are typing the same correction more than once
Immediate Fires at the moment of the mistake, not later Agent commits broken code; you catch it on review
Specific Names the file, line, and reason "Build failed" with no further detail
Actionable The agent can fix it without asking you Vague lint warnings the agent ignores

When a check fails all four, remove it or improve it. A check that is automated but not specific (e.g. "tests failed" with no output) is worse than useless — it consumes the agent's context without providing a path forward.


Backpressure Layers

Think in layers. Each layer catches a different class of mistake. The goal is full coverage with no gaps that require you to fill in manually.

Layer 1: Compiler / Type System

The fastest and cheapest feedback. Fires before any code runs.

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. 3d ago First seen · 231 lines · 138 tokens per session scan C 7fca94d24a28

Subscribe to this mod's changes

backpressure is a skill published in the GitHub repository mcwalrus/oppertunity-party (0 stars, last pushed 11d ago), licensed MIT. It adds 138 tokens to every session and 2,581 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

emdash-cli

Use the EmDash CLI to manage content, schema, media, and more. Use this skill when you need to interact with a running EmDash instance from the command line — creating content, managing collections, uploading media, generating types, or scripting CMS operations.

emdash-cms/emdash · 56 tokens

adversarial-reviewer

Adversarial code review that assumes bugs exist and hunts for them. Use when asked to review code, find bugs, audit for correctness, stress-test a PR, or when someone says "tear this apart" or "what's wrong with this". Give no benefit of the doubt — every line is guilty until proven innocent.

emdash-cms/emdash · 71 tokens

cocoindex-diagrams

This skill should be used when creating, editing, or reviewing inline SVG diagrams for the CocoIndex docs site (anything under docs/src/content/docs/). It encodes the component-based primitive system under docs/src/components/diagrams/, the palette + shape semantics, the preview-and-verify loop using headless Chrome…

cocoindex-io/cocoindex · 124 tokens

investigate

Investigate a single EmDash issue end to end -- classify, choose a repro path, reproduce, diagnose, verify, and (only on an explicit maintainer fix directive) fix and confirm. Every verdict carries its evidence.

emdash-cms/emdash · 48 tokens

dagster-docs

Expert guidance for writing documentation for the Dagster docs website. ALWAYS use before creating or updating documentation in the docs directory.

dagster-io/dagster · 29 tokens

repro-public

Reproduce a bug in the public-facing rendered site (not the admin). Attach a container, start the demo dev server, drive public routes with agent-browser, and capture the reproduction as screenshots plus a replayable transcript.

emdash-cms/emdash · 48 tokens