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 instructions/savedpixel/ai-agent-rules-generator/agent-observationsgit clone --depth 1 https://github.com/savedpixel/ai-agent-rules-generatorWrote 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/instructions/savedpixel/ai-agent-rules-generator/agent-observations)<a href="https://agentmods.dev/instructions/savedpixel/ai-agent-rules-generator/agent-observations"><img src="https://agentmods.dev/badge/instructions/savedpixel/ai-agent-rules-generator/agent-observations.svg" alt="Measured on agentmods" 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 | $0.01787 | $0.01787 |
| Opus 5 | $0.00894 | $0.00894 |
| Sonnet 5 | $0.00357 | $0.00357 |
| Haiku 4.5 | $0.00179 | $0.00179 |
Grade A, and why
ai-agent-rules-generator agent-observations.instructions.md 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 4d 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 — 180 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Observation Logs — Mandatory Disclosure System
Purpose
During any task — planning, implementation, auditing, reporting, or debugging — you will encounter information the user needs to know: anomalies in data, inconsistencies between code and documentation, improvement opportunities, potential risks, or broken assumptions. You MUST proactively disclose these findings. Silence is a violation.
This system exists because agents have historically buried important findings, only surfacing them when explicitly asked for a report. That is unacceptable. The observation logs are the user's primary way to stay informed about the health of their system without having to ask.
Directory Structure
docs/agent-observations/
├── critical.md ← Open critical issues only
├── recommendations.md ← Open recommendations only
├── anomalies.md ← Open anomalies only
└── closed/
├── critical.md ← Resolved critical issues (archive)
├── recommendations.md ← Resolved recommendations (archive)
└── anomalies.md ← Resolved anomalies (archive)
- Root files (
docs/agent-observations/*.md) contain only open/in-progress items. This gives the user a clean view of what actually needs attention. - Closed files (
docs/agent-observations/closed/*.md) are append-only archives of resolved items. Never delete entries from closed files.
The Three Observation Logs
1. critical.md — Blocking Issues
Things that are broken, dangerous, or require immediate attention. If you find something critical and do NOT log it, you are hiding a problem from the user.
Examples:
- Data inconsistencies that affect production output
- Security misconfigurations (exposed credentials, missing auth checks, unsafe defaults)
- Broken assumptions that code relies on
- Regression risks from the current task
- Data corruption or loss
- Webhook signature verification bypass
- Unescaped user input in payment forms
2. recommendations.md — Improvement Suggestions
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.
- 4d ago First seen · 180 lines · 1,787 tokens per session scan A f06e3642c9a1
ai-agent-rules-generator agent-observations.instructions.md is an instructions file published in the GitHub repository savedpixel/ai-agent-rules-generator (7 stars, last pushed 4mo ago), licensed MIT. It adds 1,787 tokens to every session, about $0.0089 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 instructions, from other repositories
copilot-setting AGENTS.md
AGENTS.md instructions for zexion7873/copilot-setting, covering copilot-setting — agent guide, communication language, repository purpose, validation commands and architecture.
copilot-setting java.instructions.md
Load when writing or reviewing .java code — Java 8 version-lock floor: syntax, exceptions, logging, concurrency. Triggers on: var, records, text blocks, List.of/Map.of, pattern matching, java.time, BigDecimal, SLF4J. Forces Java 8, not modern Java. Defer SQL/Spring/JSP to their files.
copilot-setting spring-hibernate.instructions.md
Load when writing or reviewing Java or hbm.xml on a Spring 3.2 + Hibernate 4.2 stack — native Session API, XML mappings/tx, Spring MVC. Triggers on: getCurrentSession (not openSession), hbm.xml (not JPA @Entity), (not @Transactional), @Controller/@RequestMapping (not @GetMapping). No Spring Boot/JPA. Defer SQL to…
copilot-setting sql-ddl.instructions.md
Load when writing or reviewing MySQL DDL or migration scripts — CREATE/ALTER TABLE, indexes, stored procedures, backfills. Triggers on: ALTER TABLE, ALGORITHM=INSTANT/INPLACE, online schema change (pt-osc, gh-ost), rollback scripts, chunked backfill, DROP COLUMN, sp stored procedures. MySQL 8.0 / InnoDB. Query and…
copilot-setting xml-config.instructions.md
Load when writing or reviewing stack XML — Spring applicationContext, hbm.xml, web.xml, or Maven pom.xml. Triggers on: (not ), spring-beans-3.2.xsd (not 4.0), , DispatcherServlet, , scope test/provided, source/target 1.8, no -SNAPSHOT/LATEST. No Spring Boot. Defer SQL/Java to their files.
copilot-setting sql.instructions.md
Load when writing or reviewing SQL in Java code — JDBC DAOs, HQL, query tuning. Triggers on: PreparedStatement/? (no concat), :paramName, no SELECT , WHERE on UPDATE/DELETE, indexes, EXPLAIN. Raw JDBC, not Spring Boot. DDL/migrations: sql-ddl.instructions.md. Defer Hibernate queries to spring-hibernate.instructions.md.