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/nikolamin/claude-coordinator-kit/phase-loopnpx skills add nikolamin/claude-coordinator-kit --skill phase-loopgit clone --depth 1 https://github.com/nikolamin/claude-coordinator-kitWrote 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/nikolamin/claude-coordinator-kit/phase-loop)<a href="https://agentmods.dev/skills/nikolamin/claude-coordinator-kit/phase-loop"><img src="https://agentmods.dev/badge/skills/nikolamin/claude-coordinator-kit/phase-loop.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 | $0.00169 | $0.03865 |
| Opus 5 | $0.00084 | $0.01932 |
| Sonnet 5 | $0.00034 | $0.00773 |
| Haiku 4.5 | $0.00017 | $0.00386 |
Grade A, and why
phase-loop 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 today.
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 — 219 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Phase loop
This skill packages the coordinator's phase loop for delivery via a plugin. It is the same
content as the kit's PROCESS.md, which a file-copy install places at
docs/coordination/PROCESS.md. If this project used that file-copy install, docs/coordination/ STATE.md, docs/coordination/PROCESS.md, and docs/coordination/codex-setup.md are already
present under docs/coordination/ and this skill's instructions describe exactly what's already
on disk there — read STATE.md first when resuming, it's the fastest way to reconstruct where
things stand. See CLAUDE.md (if installed at the project root) for the standing behavioral rules
this loop runs under: model routing, verification standard, comms register, escalation.
The coordinator (Claude, main session) runs this loop. It never does concept/design/implementation
work itself — it spawns build agents, dispatches independent verifier agents to check output
against acceptance criteria, re-prompts unfinished agents, and escalates only non-trivial
questions to the user. The coordinator itself never performs the verification — that's always a
separate agent dispatch, per CLAUDE.md's Role section.
Phases
0. Bootstrap — Knowledge-base skeleton (below) created, docs/coordination/STATE.md
initialized, .coordinator-scratch/ created at the project root and gitignored — append the entry
to the project's .gitignore, creating that file if it doesn't exist yet (see CLAUDE.md's Agent
brief hygiene section) — committed. First thing any fresh coordinator session does if the skeleton
doesn't exist yet — this is the one named exception to "coordinator never self-executes" (see
CLAUDE.md Role section): fixed layout, no judgment, one-time. If the skeleton already exists,
don't recreate it — read STATE.md's Current section and resume from there instead, but first
check whether .coordinator-scratch/ exists and is gitignored; if either is missing (an
already-bootstrapped project, or one installed from an earlier kit version), create/gitignore it
now as the same one-time catch-up CLAUDE.md's Role section names, commit that .gitignore change
as its own small commit, then resume. Bootstrap also decides which path Phase 0.5/1 takes: if the
repo contains real code beyond the fresh doc skeleton (source files, build config, non-trivial git
history), it's an existing project — run Phase 0.5 before
interviewing. A genuinely empty/new repo is greenfield — skip straight to Concept. Either path,
ask the founder one early question before the first build agent is ever dispatched: does a push to
the project's main branch trigger a deploy? A greenfield repo has no Phase 0.5 to carry this, so
ask it here, during Bootstrap, before Concept starts; an existing project asks it as part of Phase
0.5 below. Record the answer as a durable decision in docs/coordination/STATE.md and state the
resulting branching/worktree convention plainly at that point.
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.
- today Changed · +1 lines 1f91ed10d376
- 4d ago First seen · 218 lines · 169 tokens per session scan A ee21e811e13d
phase-loop is a skill published in the GitHub repository nikolamin/claude-coordinator-kit (2 stars, last pushed yesterday), licensed MIT. It adds 169 tokens to every session and 3,865 once invoked, about $0.0008 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 skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
auto-perf-optimize
Run agent-driven VS Code performance or memory investigations. Use when asked to launch Code OSS, automate a VS Code scenario, run the Chat memory smoke runner, capture renderer heap snapshots, take workflow screenshots, compare run summaries, or drive a repeatable scenario before heap-snapshot analysis.
chat-perf
Run chat perf benchmarks and memory leak checks against the local dev build or any published VS Code version. Use when investigating chat rendering regressions, validating perf-sensitive changes to chat UI, or checking for memory leaks in the chat response pipeline.
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…