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.
git clone --depth 1 https://github.com/sijeeshmiziha/visionagentWrote 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/rules/sijeeshmiziha/visionagent/long-running-agents)<a href="https://agentmods.dev/rules/sijeeshmiziha/visionagent/long-running-agents"><img src="https://agentmods.dev/badge/rules/sijeeshmiziha/visionagent/long-running-agents.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 5.1 | $0.00421 | $0.00421 |
| Opus 5 | $0.00211 | $0.00211 |
| Sonnet 5 | $0.00084 | $0.00084 |
| Haiku 4.5 | $0.00042 | $0.00042 |
Grade A, and why
long-running-agents 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 6d 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.
What it actually says
Long-Running Agent Harness
Two-Agent Pattern
- Initializer agent (first run): Set up the environment—feature list (JSON), init script (e.g.
init.sh), progress file (e.g.*-progress.txt), and an initial git commit. - Coding agent (every subsequent run): Make incremental progress only; leave structured updates (git commit + progress file) so the next session can resume.
Incremental Progress
- Work on one feature at a time. Never attempt to one-shot complex tasks; this leads to half-implemented, undocumented work when context resets.
- Choose the next feature from the feature list; do not skip or batch unrelated work.
Progress and Clean State
- Maintain a structured progress file (e.g.
*-progress.txtor JSON) and write descriptive git commits at the end of each session. - End every session with clean state: code that could be merged to main—no major bugs, orderly and well-documented. Use git to revert bad changes if needed.
Feature List
- Use structured JSON (not Markdown) for feature tracking with pass/fail status.
- Do not remove or edit feature definitions; only update the
passes(or equivalent) field. Removing tests leads to missing or buggy functionality.
Getting Up to Speed
At the start of each new session:
- Run
pwd; read progress file and git log to understand recent work. - Read the feature list and pick the highest-priority incomplete feature.
- Run init script / dev server and run a basic smoke test before implementing anything new. Fix existing bugs first.
End-to-End Testing
- Verify features as a user would (e.g. browser automation for web apps).
- Only mark a feature as passing after careful testing. Do not mark features done without proper verification.
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.
- 6d ago First seen · 40 lines · 421 tokens per session scan A 325039adcadb
long-running-agents is a cursor rule published in the GitHub repository sijeeshmiziha/visionagent (2 stars, last pushed 5mo ago), licensed MIT. It adds 421 tokens to every session, about $0.0021 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 cursor rules, from other repositories
database
Cursor rule "database" from ItamarZand88/awesome-agent-conventions, covering database best practices, prisma setup, prisma models, prisma queries and supabase setup.
grug-brained-dev_simple-tools
Choose tools and libraries that are simple and well-understood.
java-springboot-jpa-cursorrules-prompt-file
description: "Cursor rules for Java development with Springboot and JPA integration." globs: / alwaysApply: false.
angular-typescript-cursorrules-prompt-file
description: "Cursor rules for Angular development with TypeScript integration." globs: / alwaysApply: false.
fortran
Cursor rule "fortran" from ItamarZand88/awesome-agent-conventions, covering fortran programming guidelines, basic principles, kinds and types, naming and style and procedures.
160-workflow-orchestration
WHEN executing tasks MANAGE workflow processes for consistent memory operations.