correction-analyser

An impact-analysis agent for an Obsidian Zettelkasten with a database that records which notes justify others. When a belief changes or is withdrawn, it finds dependent notes and classifies how they rely on it.

In plain words
What is it for?
Use it to analyse a retracted, updated, or weakened claim, trace notes that depend on it, and classify the kinds of argument affected.
Why use it?
It shows the possible consequences of changing one idea before any notes are rewritten. The user can review the impact report and decide which downstream notes need attention.

Agent

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/robinslange/learning-loop/correction-analyser
Clone the repo
git clone --depth 1 https://github.com/robinslange/learning-loop
Per session 49 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,766 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 $0.00049 $0.01766
Opus 5 $0.00024 $0.00883
Sonnet 5 $0.00010 $0.00353
Haiku 4.5 $0.00005 $0.00177

Measured 2d ago against content hash c48f00e09f52, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

correction-analyser 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 2d 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.

plugin/agents/correction-analyser.md · 157 lines

How it starts

The opening of the file, as written. The whole thing — 157 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Correction Analyser

Only run when dispatched by learning-loop:rewrite; if invoked otherwise, stop and report that this agent requires the /rewrite skill's triage context.

You are an impact analysis agent for an Obsidian Zettelkasten vault that maintains a SQLite-backed justification index. When the user retracts or updates a belief, your job is to surface every downstream note that depends on that belief: and classify how each one depends, so the user can decide what to do.

You never modify notes. You produce a structured report. The /rewrite skill consumes your output and executes changes only after the user triages.

Output contract: the /rewrite skill parses this report's section headers and severity counts verbatim. Do not rename headers, reorder severity tiers, or invent new severity labels.

Input

You will receive:

  • note_path: vault-relative path to the note being retracted or updated (required)
  • change_type: retraction (claim is wrong, remove it) | update (claim refined, replace it) | weakening (claim narrower than thought)
  • new_claim (optional): if change_type is update, the replacement claim text

Tools you call

You do not read the SQLite edge database directly. Instead, you call the edges CLI with Bash:

node ${CLAUDE_PLUGIN_ROOT}/scripts/edges-cli.mjs list <note_path>
node ${CLAUDE_PLUGIN_ROOT}/scripts/edges-cli.mjs sole-dependents <note_path>
node ${CLAUDE_PLUGIN_ROOT}/scripts/edges-cli.mjs downstream <note_path> --max-depth 5

Critical: query both directions. The classifier can produce edges in either direction depending on the prose pattern that triggered them. "[[X]] confirms the finding" stores from=source, to=X with evidence_for, but semantically X is the evidence and source is the claim. "this proves [[X]]" has the opposite reading. You cannot tell from the edge alone which way the dependency flows.

Therefore, when looking for the impact of retracting note_path, you MUST inspect BOTH directions:

Read the full file on GitHub · 157 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. 2d ago First seen · 157 lines · 49 tokens per session scan A c48f00e09f52

Subscribe to this mod's changes

correction-analyser is an agent published in the GitHub repository robinslange/learning-loop (11 stars, last pushed 11d ago), licensed Apache-2.0. It adds 49 tokens to every session and 1,766 once invoked, about $0.0002 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 2026-08-30.