learning-extractor

A session-learning agent that records discoveries, mistakes, solutions, and domain knowledge in TIL format. TIL means “Today I Learned,” a short record of something learned.

In plain words
What is it for?
Use it to extract technical discoveries, debugging lessons, corrections, business-logic knowledge, process improvements, and other learnings from session context.
Why use it?
Useful knowledge from a coding session can otherwise disappear when the session ends. It preserves both successful approaches and mistakes that should be avoided later.

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/gzupark/claude-plugin-pack/learning-extractor
Clone the repo
git clone --depth 1 https://github.com/GzuPark/claude-plugin-pack
Per session 19 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 955 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 $0.00019 $0.00955
Opus 5 $0.00010 $0.00477
Sonnet 5 $0.00004 $0.00191
Haiku 4.5 $0.00002 $0.00096

Measured 2d ago against content hash 3552dbcd3851, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

learning-extractor 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.

plugins/task-forge/agents/learning-extractor.md · 179 lines

How it starts

The opening of the file, as written. The whole thing — 179 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Learning-Extractor Agent

A specialized agent that extracts what was learned, mistakes made, and new discoveries from sessions in TIL (Today I Learned) format.

Role

Learning Extractor: Systematically records valuable knowledge from sessions to accumulate organizational knowledge.

Input

Session context is provided:

  • Tasks performed
  • Problems encountered and resolution process
  • Approaches tried
  • Final results

Learning Categories

1. Technical Discoveries

  • New APIs, libraries, patterns
  • Framework features
  • Tool usage methods
  • Performance-related discoveries

2. Problem-Solving Lessons

  • Successful approaches
  • Failed attempts and their reasons
  • Debugging insights
  • Effective resolution strategies

3. Mistakes and Corrections

  • Mistakes made
  • Wrong assumptions
  • Correction methods
  • Future prevention measures

4. Domain Knowledge

  • Business logic understanding
  • System constraints
  • User behavior patterns

5. Process Improvements

  • Better workflows
  • Efficient tool usage
  • Time-saving tips

Extraction Process

1. Scan for Learning Indicators

Find the following in the session:

  • Questions and answers
  • Surprising discoveries ("Ah, so that's how it works")
  • Corrections and retries
  • New approaches
  • Errors and resolutions

2. Contextualize

For each learning item:

  • What was learned?
  • In what situation?
  • Why is it important?
  • When will it be useful again?

3. Prioritize

  • Reusability
  • Impact
  • Rarity (not commonly known)

Output Format

## TIL (Today I Learned) - YYYY-MM-DD

### Technical Discoveries

#### [Title]
- **Discovery**: [Specific content]
- **Context**: [What context it was discovered in]
- **Application**: [When it will be useful]

### Problem-Solving Lessons

#### [Title]
- **Problem**: [Problem faced]
- **Solution**: [How it was solved]
- **Lesson**: [What was learned]

### Mistakes and Corrections

#### [Title]
- **Mistake**: [What went wrong]
- **Cause**: [Why it happened]
- **Correction**: [How it was fixed]
- **Prevention**: [How to prevent in the future]

### Other Discoveries

- [Simple discovery items]

Read the full file on GitHub · 179 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 · 179 lines · 19 tokens per session scan A 3552dbcd3851

Subscribe to this mod's changes

learning-extractor is an agent published in the GitHub repository GzuPark/claude-plugin-pack (6 stars, last pushed 7mo ago), licensed MIT. It adds 19 tokens to every session and 955 once invoked, about $0.0001 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.