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 skills/samibs/skillfoundry/improvenpx skills add samibs/skillfoundry --skill improvegit clone --depth 1 https://github.com/samibs/skillfoundryWhat 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 | $0.00009 | $0.03216 |
| Opus 5 | $0.00005 | $0.01608 |
| Sonnet 5 | $0.00002 | $0.00643 |
| Haiku 4.5 | $0.00001 | $0.00322 |
Grade A, and why
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 2d 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 — 394 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/improve — Continuous Improvement Loop
Scans the codebase for improvement opportunities, fixes them one at a time, and loops until no improvements remain or the token budget is exhausted.
This is Boris Cherny's "agent prompting itself" pattern applied to codebase health: the agent scans → prioritizes → fixes → checkpoints → scans again. No human input between iterations.
Protocol engine:
agents/_ralph-loop-protocol.md+agents/_self-prompt-protocol.md
Usage
/improve Scan and fix all improvement categories
/improve arch Architectural improvements only
/improve duplication Duplicate code / abstraction consolidation
/improve quality Code quality (GuardLoop patterns GL-01 through GL-10)
/improve security Security posture improvements
/improve --dry-run Show what would be fixed, do not apply any changes
/improve --resume Resume a previously interrupted improvement loop
/improve --budget [N] Set max iterations (default: 10)
/improve --pr Open a PR with all fixes when the loop completes
What This Command Does
This is a self-prompting loop — the agent finds the work, does the work, finds more work, and stops when there is nothing left.
LOOP {
1. SCAN — Identify all improvement opportunities in scope
2. PRIORITIZE — Rank by impact, select the single highest-priority item
3. FIX — Apply the fix (surgical — one item per iteration, no scope creep)
4. VERIFY — Confirm the fix did not regress anything
5. CHECKPOINT — Record: what was fixed, what remains, what was learned
6. JUDGE — Any improvements remaining? Budget sufficient to continue?
7. SELF-PROMPT — Formulate: "I fixed X. Still need to address Y at [location]."
8. → RE-ENTER at step 2 with that self-prompt as current_task
}
EXIT when:
- Backlog is empty (no improvements remain in scope)
- Token budget < 20% remaining
- Max iterations reached
- Same item appears in remaining work twice with no change (oscillation)
- An item cannot be fixed without user input (blocked)
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.
- 2d ago First seen · 394 lines · 9 tokens per session scan A d9b52a97c1c3
improve is a skill published in the GitHub repository samibs/skillfoundry (12 stars, last pushed 1mo ago), licensed MIT. It adds 9 tokens to every session and 3,216 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 2026-08-30.
Other skills, from other repositories
memorix-troubleshooting
Use when Memorix MCP, setup, project binding, HTTP control plane, hooks, skills, or agent integration is missing, stale, or failing.
memorix-sessions
Use when resuming work, preparing handoff context, binding an HTTP control-plane project, or deciding whether sessionstart is useful.
ring:applying-composition-patterns
React composition patterns that scale. Avoid boolean prop proliferation by using compound components, lifting state, and composing internals. Use when refactoring components with boolean prop proliferation, building flexible component libraries, or during architecture review. Skip for simple components with 1-2 props…
ring:searching-code
Forensic code search and analysis with optional Chain of Draft (CoD) ultra-concise mode. Five-phase methodology (clarification, planning, execution, analysis, synthesis) with severity assessment. Use for targeted investigation of specific patterns, bugs, or vulnerabilities. Skip for broad architecture mapping (use…
ring:exploring-codebases
Exploring a codebase across phases: scopes the target, detects architecture, components, and layers, deep-dives each discovered perspective, then synthesizes findings into actionable guidance with file:line evidence. Use to understand how a feature or system works before planning changes, or to orient on an unfamiliar…
ring:auditing-dependency-security
Auditing a dependency for supply-chain risk before install (pip/npm/go/cargo): checks typosquatting, maintainer/age risk, vulnerability DBs (OSV, GHSA, Socket), and lockfile hash pinning, then emits a risk score and approve/conditional/escalate/block decision. Use when adding or updating a dependency, reviewing a…