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 skills add sara-star-quant/presence --skill confidence-gategit clone --depth 1 https://github.com/sara-star-quant/presenceWrote 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/sara-star-quant/presence/confidence-gate)<a href="https://agentmods.dev/skills/sara-star-quant/presence/confidence-gate"><img src="https://agentmods.dev/badge/skills/sara-star-quant/presence/confidence-gate/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/sara-star-quant/presence/confidence-gate"><img src="https://agentmods.dev/badge/skills/sara-star-quant/presence/confidence-gate.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 | $0.00078 | $0.00614 |
| Opus 5 | $0.00039 | $0.00307 |
| Sonnet 5 | $0.00016 | $0.00123 |
| Haiku 4.5 | $0.00008 | $0.00061 |
Grade A, and why
confidence-gate 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 9d 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 — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Confidence gate: claim what you can verify
presence's Stop hook parses your final assistant message for unhedged success language. If it finds words like "fixed", "done", "works", "passing tests", and there's been an edit but no passing test/build event since that edit, it logs the discrepancy. In strict presets it also re-prompts you to verify before stopping.
How to avoid the warning
Three options, in order of preference:
1. Actually verify
Run the test suite, build, lint, or whatever the project uses for verification, before declaring success. The presence event log captures npm test, pytest, cargo test, tsc, next build, and ~15 other common commands automatically. Running any of them creates a test_pass / build_pass event that satisfies the gate.
2. Hedge explicitly
If you can't run the verification (no tests in the project, network-isolated environment, user wants quick scaffold without running CI), hedge the claim:
- "I think this fixes the issue, but I haven't run the tests."
- "This should work; needs verification."
- "Untested: ..."
These hedges are recognized by the gate and disable the warning.
3. Don't claim what you didn't do
The cheapest fix: don't say "fixed" if you only edited the file. Say "made the change to X; please run the tests to confirm."
Why this matters
The single most common mode of agent failure is asserting completion when the work isn't actually verified. The user trusts the assertion, ships the change, and discovers the regression later when it's expensive to roll back. The gate exists to make this failure mode visible before the session ends.
Settings
The gate intensity is set by the active preset (see /presence-preset):
| Preset | Commit gate (PreToolUse) | Stop gate |
|---|---|---|
solo-dev (default) |
off | silent (logged to confidence.jsonl, surfaced via /presence-doctor) |
team-oss |
warn (advisory message into context, no interruption) | silent |
enterprise-strict |
block (refuses commit until verified) | block (re-prompts on unverified success) |
zerotrust |
block | block |
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.
- 9d ago First seen · 48 lines · 78 tokens per session scan A e3a0a791e1d2
confidence-gate is a skill published in the GitHub repository sara-star-quant/presence (7 stars, last pushed 2d ago), licensed Apache-2.0. It adds 78 tokens to every session and 614 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
media-gen
Photoreal image + video generation and upscaling via fal.ai (pay-per-use, one API key, always-current models: Kling, Seedance, FLUX, nano-banana Pro, Topaz). Use for image-to-video ("animate this image", "turn this photo into a video"), photoreal video clips for websites/demos, hero video loops, upscaling…
research-notebook
Deep research on any topic via the NotebookLM CLI. Use whenever the user requests research on a topic, in any project, in any language. Triggers on phrases like "research this", "deep dive on", "do research on", "dig into", "find out about", "investigate", "look into", "research X for me". Creates or reuses a…
learn
Use when the user shares a conclusion, note, or takeaway from something they're studying — a book, course, or article — and wants it captured into their personal learning archive. Trigger phrases include "what I learned today", "notes from the book", "conclusions from chapter X", plus the explicit /learn command. Also…
wiki-article
Write a wiki article capturing structure you learned about an active project in the user's hub. Use when you grepped or spawned Explore agents to understand a project and must record that knowledge before /done, or when the user asks to "write a wiki article", "document this in the wiki", or "add this to the wiki".
are-you-done
A completion checker for coding work that asks for evidence before allowing an assistant to say a task is finished.
did-you-forget-me
A requirement-checking skill that compares the current work with what the user explicitly asked for. It is written in Korean and includes a fixed Korean report format.