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 rcarmo/piclaw-addons --skill late-night-regretsgit clone --depth 1 https://github.com/rcarmo/piclaw-addonsWrote 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/rcarmo/piclaw-addons/late-night-regrets)<a href="https://agentmods.dev/skills/rcarmo/piclaw-addons/late-night-regrets"><img src="https://agentmods.dev/badge/skills/rcarmo/piclaw-addons/late-night-regrets/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/rcarmo/piclaw-addons/late-night-regrets"><img src="https://agentmods.dev/badge/skills/rcarmo/piclaw-addons/late-night-regrets.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00031 | $0.00871 |
| Opus 5 | $0.00015 | $0.00436 |
| Sonnet 5 | $0.00006 | $0.00174 |
| Haiku 4.5 | $0.00003 | $0.00087 |
Grade A, and why
late-night-regrets 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 11d 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 — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Late Night Regrets
A lightweight, zero-token-cost interaction-quality classifier that runs nightly to identify behavioral patterns worth improving.
What it does
- Trains a Multinomial Naive Bayes classifier on the full chat message history
- Classifies every user message that follows an agent turn into quality categories
- Flags attention-worthy messages (corrections, misinterpretations, under-deliveries)
- Triggers a short agent reflection pass that reads flagged messages, identifies patterns, and writes learnings to notes
Categories
| Label | Meaning |
|---|---|
successful_execution |
Agent fulfilled the request correctly; user approved or moved on |
course_correction |
User had to steer, clarify, or redirect |
misinterpretation |
Agent misread intent; user explicitly corrected |
over_engineering |
Agent did too much; user asked to simplify |
under_delivery |
Agent gave too little; user pushed for more |
context_failure |
Agent forgot/lost context; user had to repeat |
good_proactive |
Agent anticipated a need; user approved |
neutral |
Normal flow, no strong signal |
How classification works (no model tokens)
- Messages are tokenized (lowercased, code/URLs stripped, bigrams, structural features)
- Sequential context features: previous message sender, length, turn-pair detection, self-repetition ratio
- Weak labels derived from pattern matching on user follow-ups (approval words, correction phrases, repetition detection)
- Standard MNB training with 80/20 deterministic split
- Confidence-based filtering: only non-neutral predictions above threshold get flagged
Nightly flow
The scheduled task runs at 02:30 UTC (configurable):
1. bun run <addon>/scripts/train-interaction-quality-bayes.ts
→ retrains on full history, writes weights + predictions + attention file
2. Agent reads attention file, filters last 24h
3. For each flagged message:
- Reads surrounding context from messages DB
- Identifies what the user wanted vs what agent did
4. Writes patterns to notes/memory/interaction-reflections.md
- Date, top patterns, behavioral adjustments
5. Appends new steering cues to notes/memory/feedback.md if warranted
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.
- 11d ago First seen · 97 lines · 31 tokens per session scan A 47c2b5a05773
late-night-regrets is a skill published in the GitHub repository rcarmo/piclaw-addons (23 stars, last pushed today), licensed MIT. It adds 31 tokens to every session and 871 once invoked, about $0.0002 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.
Other skills, from other repositories
pi
Use when needing a minimal, extensible terminal AI coding agent harness in TypeScript with plugin architecture. Pi: minimalist terminal AI coding agent emphasizing simplicity and composability.
OpenShell + Pi: build an isolated coding-agent sandbox
Use when the user wants to install, configure, migrate, harden, inspect, or troubleshoot a Pi coding agent running in NVIDIA OpenShell, especially with rootless Podman, deny-by-default networking, hidden credentials, persistent state, or A2A communication. Do not use for ordinary containerization that does not involve…
ensembl-database
Query Ensembl genome database REST API for 250+ species. Gene lookups, sequence retrieval, variant analysis, comparative genomics, orthologs, VEP predictions, for genomic research.
aatmf-t10-confidentiality-breach
AATMF T10 — Integrity & Confidentiality Breach. System prompt extraction, training-data extraction, model-weight leakage, private-key recovery.
memory
Use when the user asks to remember, recall, forget, update, search, or inspect durable OpenSquilla memory, including profile facts in USER.md and long-term notes in MEMORY.md or memory//.md.
omh-rust
This is a Hermes-native rust workflow skill.