A command set for reviewing and managing small records of useful user feedback from coding-agent sessions. These records can capture corrections, preferences, approvals, decisions, or manually added notes before they are added to a topic wiki.

Use it to list or inspect feedback candidates, capture a note manually, and promote an approved candidate into a topic's notes and activity log.

It keeps important feedback from being lost in long conversations and separates captured suggestions from changes that have actually been approved for reuse. It also supports showing a redacted preview before promotion.

feedback — Command by nvk

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 commands/nvk/llm-wiki/feedback

Clone the repo

git clone --depth 1 https://github.com/nvk/llm-wiki

Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 553 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.

Model	Per session	Once invoked
Fable 5	$0.00015	$0.00553
Opus 5	$0.00008	$0.00277
Sonnet 5	$0.00003	$0.00111
Haiku 4.5	$0.00002	$0.00055

Measured 2d ago against content hash 473b04713fc3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

feedback 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.

claude-plugin/commands/feedback.md · 48 lines

How it starts

The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Your task

Manage the feedback-curator layer for llm-wiki sessions. Feedback candidates are small distilled records of user corrections, preferences, approvals, and plan acceptance signals. They live under HUB/.sessions/feedback/ until explicitly promoted into a topic wiki.

First read references/feedback.md and references/sessions.md, then resolve HUB by the standard hub-resolution protocol from references/hub-resolution.md. Prefer the deterministic helper when available:

scripts/llm-wiki-session --hub "$HUB" feedback <subcommand>

Parse $ARGUMENTS

list [--unpromoted] [--type correction|preference|approval|decision|manual] [--min-confidence low|medium|high]: show captured candidates.
show <candidate-id>: display one candidate with its distilled lesson and redacted user-feedback preview.
capture --text "..." [--session-id <id>]: manually add a feedback candidate.
promote <candidate-id> --topic <slug>: write the candidate into the target topic's raw/notes/ layer and append the topic log.

Policy

Candidate capture is allowed from trusted hooks, but promotion is explicit.
Do not treat generic acknowledgements (ok, thanks, cool) as durable lessons.
High-value feedback includes corrections, preferences, always/never rules, and explicit statements that the agent did the right thing.
Approval/decision candidates are validation of the immediately preceding context unless corroborated by a session digest.
Store redacted previews and hashes, not full transcripts.

Examples

scripts/llm-wiki-session --hub "$HUB" feedback list --unpromoted
scripts/llm-wiki-session --hub "$HUB" feedback show fb-abc123
scripts/llm-wiki-session --hub "$HUB" feedback capture --text "next time use the release checklist before tagging"
scripts/llm-wiki-session --hub "$HUB" feedback promote fb-abc123 --topic meta-llm-wiki

Report absolute paths for every promoted note. Append topic log.md when promoting into a topic wiki.

Read the full file on GitHub · 48 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.

2d ago First seen · 48 lines · 15 tokens per session scan A 473b04713fc3

Subscribe to this mod's changes

feedback is a command published in the GitHub repository nvk/llm-wiki (1,175 stars, last pushed 5d ago), licensed MIT. It adds 15 tokens to every session and 553 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 2026-08-30.

Code review Data and AI LLM Productivity agentic-ai agentic-skills agentic-workflow claude-code codex

feedback

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

Your task

Parse $ARGUMENTS

Policy

Examples

What this file has done since we first saw it

Other commands, from other repositories

implement

workflow-claude-commands

weather-orchestrator

project

review

research-verify