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 skills/baskb/feedback-studio/demonpx skills add baskb/feedback-studio --skill demogit clone --depth 1 https://github.com/baskb/feedback-studioWhat 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 | $0.00081 | $0.00997 |
| Opus 5 | $0.00041 | $0.00498 |
| Sonnet 5 | $0.00016 | $0.00199 |
| Haiku 4.5 | $0.00008 | $0.00100 |
Grade A, and why
demo scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Feedback Studio — demo playground
Serves a bundled, deliberately imperfect landing page ("Roastly") from a throwaway
temp-dir copy. By default it starts empty — no comments — so the user adds their own
(ideal for a clean walkthrough or recording). The page hides a couple of flaws (a bland
headline, a stale footer year) as ready targets. If the user wants worked examples instead,
the --seeded option loads three open comments — one change (a button label), one
improve (a lorem-ipsum paragraph), one fix (a typo). Either way, processing edits the
throwaway copy, never the user's project.
Start it
Run in the background, then open the printed URL yourself. Pass --no-open so the server
doesn't also auto-open a tab; two opens cause a visible page flash (bad on a demo video). The
demo starts empty by default (--no-seed):
node "${CLAUDE_PLUGIN_ROOT}/bin/feedback-studio.mjs" --demo --no-seed --no-open
Only if the user wants the three worked example comments (they said --seeded, "with
examples", or "seed it") drop --no-seed:
node "${CLAUDE_PLUGIN_ROOT}/bin/feedback-studio.mjs" --demo --no-open
Pass through any extra flags the user asked for (--tunnel for phone voice, --port).
Capture two paths from the server output — you need them for processing:
Source -> demo site (throwaway copy: <DEMO_DIR>)Comments -> <DEMO_DIR>/.feedback/comments.json
Open the URL (start <url> Windows / open <url> macOS / xdg-open <url> Linux), then
tell the user, briefly:
- The page starts empty — press P (or the Point button), then click any element or
select text to leave a comment; the mic dictates. (If they ran
--seeded, three example pins are already there to click and read.) - Two flaws are hidden on the page — a bland say-nothing headline and a stale footer year — good targets to find and pin. Try one of each type (fix / change / improve) to show the model.
- When they're ready, the cue is PPF — Please Process Feedback (or just "process the feedback"). The please is intentional: we're nice to our agents. ;-)
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.
- 3d ago First seen · 72 lines · 81 tokens per session scan A 96b23b5e3010
demo is a skill published in the GitHub repository baskb/feedback-studio (2 stars, last pushed 6d ago), licensed MIT. It adds 81 tokens to every session and 997 once invoked, about $0.0004 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.
Other skills, from other repositories
md-review
Run the standard markdown review loop end-to-end with the mdxg-redline CLI: send a markdown document to a human reviewer, wait (via Bash polling) for the feedback.json they export, then apply their location-aware inline comments back into the markdown and revise it. Always use this skill when the user wants…
markstream-react
Integrate the beta markstream-react package into a React 18+ or Next app. Use when Codex needs to add the React renderer, choose the root, next, or server entrypoint, import CSS correctly, choose between content and nodes, keep client boundaries safe, add renderer-local streamingComponents or htmlComponents, use…
decided-import
Reformat ONE existing document (a decision, requirement, design, roadmap, or prompt) into ONE valid RAC (requirements-as-code) artifact, with a mandatory human-review step before any file is written and decided validate as the deterministic close. Use when a user wants to add or import a single existing decision or…
nba
Read the live cycle state and return the single highest-leverage next best action, not a menu. Use when a project is between phases, the author asks what to do next, too many valid threads are open, or the work needs re-entry into frame, build, drive, re0-memo, hate, re0-work, or ship.
dependabot-review
Analyzes a Dependabot PR to determine what actually changed in each bumped package and whether those changes affect this repo. Reports changed APIs/methods, which doc pages use them, and the realistic probability of any visible impact on the docs site.
review-validation
Validate review findings from specialist agents by reading the actual repo file content and checking each finding for accuracy, applicability, and false positives.