Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add karlkfi/claude-bouncer/plugin install foreground-guardWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/karlkfi/claude-bouncer/reduce-foreground-guard-prompts)<a href="https://agentmods.dev/skills/karlkfi/claude-bouncer/reduce-foreground-guard-prompts"><img src="https://agentmods.dev/badge/skills/karlkfi/claude-bouncer/reduce-foreground-guard-prompts/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/karlkfi/claude-bouncer/reduce-foreground-guard-prompts"><img src="https://agentmods.dev/badge/skills/karlkfi/claude-bouncer/reduce-foreground-guard-prompts.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 5.1 | $0.00074 | $0.03235 |
| Opus 5 | $0.00037 | $0.01618 |
| Sonnet 5 | $0.00015 | $0.00647 |
| Haiku 4.5 | $0.00007 | $0.00324 |
Grade A, and why
reduce-foreground-guard-prompts scanned grade A with 1 finding 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 12d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
task for…"). Skip the wait and do the follow-up check now (`sleep 300 && curl How it starts
The opening of the file, as written. The whole thing — 213 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reducing foreground-guard prompts
foreground-guard is a PreToolUse hook for Bash that guards the session's
main-thread time. It blocks in two classes and defers silently on
everything else. The verdict is deny by default: the resource at stake is the
session's own time, so every finding carries a rewrite the agent can apply and
nobody has to be woken. "action": "ask" on either class turns the denies back
into prompts for someone who wants to watch the guard work.
- Class A — foreground poll/watch: a command that parks the main thread on a
live view — watch/follow modes (
gh run watch,gh pr checks --watch,kubectl logs -f,kubectl get -w,tail -f,journalctl -f,docker logs -f,watch ...), awhile/until/forloop that polls withsleep, a chained repeat-with-sleep (cmd; sleep N; cmd), or a baresleep Nat/above the floor (default 10s).run_in_background: truedoes not exempt Class A: a detached poll still holds a task slot for the whole wait and hands back output whose freshness the agent can't judge, so it still blocks. - Class B — slow command with an inadequate timeout: a command the repo
registered as slow that is about to run in the foreground with the Bash call's
timeoutbelow the registered minimum — it would be killed mid-run.run_in_background: trueis the fix here, and exempts it.
So a flood of prompts almost always means the agent keeps waiting on a poll instead of snapshotting and re-checking next turn, or keeps under-timing a known-slow command — both fixable habits — not that the work genuinely needs to block.
Diagnose
Don't guess about past friction — measure it. The plugin ships an analyzer,
scripts/friction-report.py, that re-reads the hook decisions Claude Code
already recorded in the local session transcripts and ranks them by category,
flagged tool, and triggering command (no telemetry — see PRIVACY.md). Run it
first so the diagnosis is grounded in the user's real prompt history:
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.
- 12d ago First seen · 213 lines · 74 tokens per session scan A 55c1458846d6
reduce-foreground-guard-prompts is a skill published in the GitHub repository karlkfi/claude-bouncer (3 stars, last pushed yesterday), licensed MIT. It adds 74 tokens to every session and 3,235 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
critique
Dispatch to pi behind a critic gate — a second, independent pi session judges the work against a contract written up front, and rejections go back for a bounded number of rounds.
review
Have pi review code as a second opinion, then adjudicate its findings.
discuss
Think a problem through with pi as a second opinion, over as many turns as it takes.
nft-standards
Implement NFT standards (ERC-721, ERC-1155) with proper metadata handling, minting strategies, and marketplace integration. Use when creating NFT contracts, building NFT marketplaces, or implementing digital asset systems.
istio-traffic-management
Configure Istio traffic management including routing, load balancing, circuit breakers, and canary deployments. Use when implementing service mesh traffic policies, progressive delivery, or resilience patterns.
postgresql-table-design
Use this skill when designing or reviewing a PostgreSQL-specific schema. Covers best-practices, data types, indexing, constraints, performance patterns, and advanced features.