errata

errata is a skill for Claude Code, Codex from donald-ada/workinggenius. It costs 92 tokens per session (1,624 once invoked), scanned A, original, MIT.

A correction workflow for project records and work documents that have been contradicted by later evidence. It changes binding instructions in place and appends the evidence and history to the record that originally captured the claim.

In plain words
What is it for?
Use it when tests, files, a later run, or the user show that an existing project instruction or recorded claim is wrong.
Why use it?
Incorrect decisions, glossary terms, plans, or criteria can mislead every later coding session. This fixes the active guidance while preserving how the correction was discovered.

Skill for Claude CodeCodex

Part of the workinggenius plugin — 19 skills, 1 agent shipped together

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/donald-ada/workinggenius/errata
Any agent
npx skills add donald-ada/workinggenius --skill errata
Clone the repo
git clone --depth 1 https://github.com/donald-ada/workinggenius

Made for: Claude Code, Codex.

Or install workinggenius, the plugin that ships this one along with the rest of its 19 skills, 1 agent.

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 errata

README.md
[![agentmods](https://agentmods.dev/badge/skills/donald-ada/workinggenius/errata.svg)](https://agentmods.dev/skills/donald-ada/workinggenius/errata)
Your own site
<a href="https://agentmods.dev/skills/donald-ada/workinggenius/errata"><img src="https://agentmods.dev/badge/skills/donald-ada/workinggenius/errata.svg" alt="Measured on agentmods" height="20"></a>
Per session 92 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,624 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 $0.00092 $0.01624
Opus 5 $0.00046 $0.00812
Sonnet 5 $0.00018 $0.00325
Haiku 4.5 $0.00009 $0.00162

Measured 3d ago against content hash 49c512cc93e8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

errata 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 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.

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.

skills/errata/SKILL.md · 67 lines

How it starts

The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Errata

Every stage writes at the moment of action — the interview as it ran, the criterion while the output is still on screen. That is the only way a record is ever honest, and it is also why some of what gets written is wrong. A fact gets reported that its instrument could not actually see. A number gets pinned on a machine that later changes. An option gets killed for a reason that does not hold. None of that is the discipline failing; it is what writing at the moment of action costs.

What it cannot be allowed to do is compound. A wrong line in a work file is read by every session after it, believed because the file outranks memory, and built on — and by the third session nobody can tell it apart from the lines that were right.

The concept: the layer that binds gets corrected; the layer that records gets appended to.

The test

One question sorts any line into its layer: would a cold session act differently because of this line?

  • Yes — it binds. The snapshot — its contract with the seams and pinned values, its slice list, its Open section — .genius/DECIDED.md, CONTEXT.md, DESIGN.md, the ## Working Genius section wherever the project keeps it. A wrong line here is an instruction, and it gets rewritten in place.
  • No — it records. Everything in the log (<slug>.log.md, or the <slug>/ folder of an older-format work), and every file at stage: done. It is what was written then, which is its whole worth. Append the correction; never edit the line.

The two are not alternatives. A fact that was wrong is usually in both, and both get handled — the binding copy rewritten, the record it came from annotated.

The three moves

Correct — it was wrong when it was written. Rewrite the binding copy. Append to the record: what was written, what is true, what overturned it, and what made it wrong.

> **Corrected 2026-08-17** — "no headless browser in this project" was wrong:
> Chromium is installed at `/opt/pw-browsers/chromium`, off `PATH`. Slice 3's
> `npx playwright --version` found it. The check was `which chromium`, which
> reads `PATH` and nothing else.

Read the full file on GitHub · 67 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 · 67 lines · 92 tokens per session scan A 49c512cc93e8

Subscribe to this mod's changes

errata is a skill published in the GitHub repository donald-ada/workinggenius (5 stars, last pushed 6d ago), licensed MIT. It adds 92 tokens to every session and 1,624 once invoked, about $0.0005 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.

Related

Other skills, from other repositories