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.
npx agentmods add commands/adityarbhat/feedfwd/learngit clone --depth 1 https://github.com/adityarbhat/feedfwdWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.01474 |
| Opus 5 | $0.00000 | $0.00737 |
| Sonnet 5 | $0.00000 | $0.00295 |
| Haiku 4.5 | $0.00000 | $0.00147 |
Grade A, and why
learn 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/learn Command
================================================================
FeedFwd: Capture knowledge from URLs, pasted text, or screenshots.
This is the primary input mechanism for the plugin.
How it works in Claude Code:
When a user types "/learn ", Claude Code reads this file
and follows the instructions below. The command definition IS
the prompt — Claude interprets and executes it directly.
The command fetches article content in the main session (quick),
then spawns a Task subagent for the distillation work (isolated
from the main context window).
================================================================
name: learn description: "Capture knowledge from a URL, pasted text, or screenshot and distill it into a knowledge card" arguments: "[url1] [url2] [url3] ... or --list"
You are handling the FeedFwd /learn command. The user wants to capture
knowledge and distill it into an actionable knowledge card.
Step 1: Determine the Mode
Look at the arguments the user provided:
- If
--listwas passed: Tell the user to run/knowledgeinstead. Stop here. - If one or more URLs were provided (strings starting with http:// or https://): URL mode. Go to Step 2.
- If no arguments were provided: Paste mode. Go to Step 2b.
- If the argument is not a URL (no http/https prefix): Treat it as pasted text. Go to Step 2b.
Step 2: URL Mode — Fetch Article Content
For each URL provided, fetch the article text:
python scripts/fetch_url.py "<url>"
If the fetch fails (non-zero exit code), report the error to the user:
⚠️ Couldn't fetch <url> — it may be behind a paywall or require authentication.
Paste the content directly instead: run /learn with no args.
Then skip this URL and continue with the next one (if batch).
If the fetch succeeds, save the output text and go to Step 3.
Step 2b: Paste Mode
If no URL was provided, ask the user to paste the article text, screenshot, or content they want to learn from. Once they provide it, use that as the article content and go to Step 3.
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.
- yesterday First seen · 181 lines · 0 tokens per session scan A a74cb4c9b2c9
learn is a command published in the GitHub repository adityarbhat/feedfwd (2 stars, last pushed 6mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,474 tokens. 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.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
constitution
Create or update the project constitution from interactive or provided principle inputs.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.