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 levi-qiao/longgraph-skill --skill loop-convergegit clone --depth 1 https://github.com/levi-qiao/longgraph-skillWrote 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/levi-qiao/longgraph-skill/loop-converge)<a href="https://agentmods.dev/skills/levi-qiao/longgraph-skill/loop-converge"><img src="https://agentmods.dev/badge/skills/levi-qiao/longgraph-skill/loop-converge/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/levi-qiao/longgraph-skill/loop-converge"><img src="https://agentmods.dev/badge/skills/levi-qiao/longgraph-skill/loop-converge.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 | $0.00046 | $0.00444 |
| Opus 5.5 | $0.00018 | $0.00178 |
| Sonnet 5.5 | $0.00009 | $0.00089 |
| Haiku 4.5 | $0.00005 | $0.00044 |
Grade A, and why
loop-converge 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 1mo 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
loop-converge — a loop-graph preset for slimming code
A thin authoring entry. It binds a code-convergence pack, starts the owner
interview, then follows loop-graph to compile a
normal two-node run (executor + supervisor). It is not a third runtime node
and it does not ship a second template set.
Fit check
- Use this when unused or duplicate code will take many verified rounds to remove, merge, or reuse, and an independent audit should keep the bar from sliding into a rewrite.
- If the request is a one-shot tidy that fits a normal host task, say so and send the task directly. Do not wrap it in a graph.
On invoke
- Inspect the workspace and the current host the same way loop-graph does. Never ask which client this is when context already identifies it.
- Read and bind
preset.md. That pack is the North Star, the supervisor requirement, the method guards, the knob overrides, the recommended shape, and the detector hints. Do not redesign them. - Start the owner interview immediately. With this pack bound, ask at most the three questions the pack names — scope, authority, launch mode — each as a recommended A/B (or A/B/C) choice. Do not ask for a North Star. Do not offer to omit the supervisor.
- Read and follow
../loop-graph/SKILL.mdfrom When called from a preset skill through generate and deliver. Compile only from loop-graph'stemplates/. This skill never executes the generated nodes.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago Changed · +1 lines · -64 tokens per session 7d2db1519019
- 1mo ago First seen · 36 lines · 110 tokens per session scan A 77cfc26a27de
loop-converge is a skill published in the GitHub repository levi-qiao/longgraph-skill (78 stars, last pushed 1mo ago), licensed MIT. It adds 46 tokens to every session and 444 once invoked, about $0.0002 per session on Opus 5.5. 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
fabric-spec
Starts a persistent Pi Fabric spec supervisor that audits the main session against a feature design spec and steers only when a requirement lacks verified evidence. Use for strict, unblocked spec compliance while the main agent keeps full freedom to orchestrate.
fabric-advisor
Starts a persistent Pi Fabric peer advisor that reviews the main agent at decision points and surfaces only concrete, material advice. Use for ambient correctness review without another extension.
review-team
A multi-reviewer code review process that checks a change from several specialist viewpoints and combines the results into one report. It can cover bugs, security, tests, dependencies, frontend behavior, and continuous-integration workflows.
dale-loop-pr
Monitor or repair one existing pull request using current review and CI evidence. Use for PR babysitting, scoped fixes, or carrying a PR to merge-ready state.
eval-code-quality
Specialized code quality evaluator for the Evaluate-Loop. Use this for evaluating code implementation tracks where the deliverable is functional code — features, API routes, state management, utilities. Checks build integrity, type safety, code patterns, error handling, dead code, imports, test coverage, and naming…
receiving-code-review
Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation.