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/neuromechanist/research-skillsWrote 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/neuromechanist/research-skills/self_improve)<a href="https://agentmods.dev/rules/neuromechanist/research-skills/self_improve"><img src="https://agentmods.dev/badge/rules/neuromechanist/research-skills/self_improve/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/rules/neuromechanist/research-skills/self_improve"><img src="https://agentmods.dev/badge/rules/neuromechanist/research-skills/self_improve.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 | $0.00004 | $0.00628 |
| Opus 5 | $0.00002 | $0.00314 |
| Sonnet 5 | $0.00001 | $0.00126 |
| Haiku 4.5 | $0.00000 | $0.00063 |
Grade A, and why
self_improve 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 5d 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 — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Continuous Rule Improvement
Philosophy: Rules Grow from Understanding
Think deeply: Why did this pattern emerge? What problem does it solve? Learn actively: Every project teaches something - capture it. Evolve thoughtfully: Rules should guide, not constrain creativity.
Improvement Triggers
- Pattern used 3+ times → Create rule
- Common failures in .context/scratch_history.md → Add prevention rule
- Successful .context/research.md solutions → Standardize
- Mature .context/ideas.md concepts → Formalize
- Repeated PR feedback → Document standard
Analysis Sources
- .context/scratch_history.md: Mine for anti-patterns
- .context/research.md: Extract proven solutions
- .context/ideas.md: Promote design principles
- .context/plan.md: Identify workflow patterns
- Code reviews: Track common feedback
Rule Updates
Add Rules When:
- New pattern appears 3+ times
- Common bug could be prevented
- Better approach discovered
- Security/performance pattern emerges
Modify Rules When:
- Better examples found in codebase
- Edge cases discovered
- Implementation changed
- Related rules updated
Remove Rules When:
- Tech stack changed
- Pattern deprecated
- No longer applicable
Quality Checks
- Actionable: Clear what to do
- Specific: No ambiguity
- Examples: From actual code
- Cross-referenced: Link related rules
Learning-Driven Creation
Extract wisdom, not just fixes:
# From .context/scratch_history.md failure:
# "Database connections leaked after 24hrs"
# THINK: Why? Resource management issue.
# LESSON: Explicit cleanup isn't reliable.
# → Rule: Always use context managers
with get_db() as db: # Guaranteed cleanup
process(db)
# Not: db = get_db(); process(db); db.close() # Risky
Ask yourself:
- What's the root cause?
- Will this prevent future issues?
- Is this a symptom of a bigger pattern?
Thoughtful Maintenance Process
- Weekly: Review .context/scratch_history.md - What patterns emerged?
- After features: Mine .context/research.md - What worked well?
- Post-refactor: Update rules - What changed fundamentally?
- Quarterly: Audit all - Are rules still serving us?
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.
- 5d ago First seen · 90 lines · 4 tokens per session scan A 373e6c9dec0e
self_improve is a cursor rule published in the GitHub repository neuromechanist/research-skills (45 stars, last pushed 6d ago), licensed BSD-3-Clause. It adds 4 tokens to every session and 628 once invoked, about $0.0000 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 2026-09-04.
Other cursor rules, from other repositories
n8n-workflow-format
Use n8n-as-code TypeScript workflow files and explicit sync discipline for n8n workflow work.
retention-predictor
Skill — retention-predictor. Predicts retention potential via usage frequency, habit formation mechanics, and churn risk factors.
controllers
Remember: Controllers are just the translation layer between HTTP and your application. Keep them simple and let other layers handle complexity.
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.