Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/AleksandarBisevac/claude-pluginsnpx agentmods add skills/aleksandarbisevac/claude-plugins/orchestrating-parallel-workWrote 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/aleksandarbisevac/claude-plugins/orchestrating-parallel-work)<a href="https://agentmods.dev/skills/aleksandarbisevac/claude-plugins/orchestrating-parallel-work"><img src="https://agentmods.dev/badge/skills/aleksandarbisevac/claude-plugins/orchestrating-parallel-work/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/aleksandarbisevac/claude-plugins/orchestrating-parallel-work"><img src="https://agentmods.dev/badge/skills/aleksandarbisevac/claude-plugins/orchestrating-parallel-work.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 | $0.00000 | $0.02051 |
| Opus 5.5 | $0.00000 | $0.00820 |
| Sonnet 5.5 | $0.00000 | $0.00410 |
| Haiku 4.5 | $0.00000 | $0.00205 |
Grade A, and why
orchestrating-parallel-work 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 26d 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 — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Orchestrating parallel work
Running several agents against one repository is mostly bookkeeping, and the bookkeeping fails in a small number of ways that all look like carelessness and are not. Each rule below was written after the failure happened here, in a run where the orchestrator knew the rule already. Knowing is not the problem. The moment of action is.
So every rule names the command that checks it. A rule you cannot check at the moment you act is a rule you will follow when you are calm and skip when you are three agents deep.
Before you say what an agent is doing
A claim about work in flight is checked before it is spoken. ListAgents says what is
running; git status in the worktree says what has been written. Both are cheap.
This exists because an orchestrator here reported a task as running when no agent had been spawned for it — the task and its worktree had been created, the launch was intended, and the intention got written down as a fact. The worktree was clean and the agent list was empty; either check would have caught it in seconds.
The transcript size is not the signal. It does not grow while an agent works. The worktree does.
Taking work out of a worktree
Use tools/harvest-worktree.py <worktree>, which is what enforces every rule in this
section — it derives the base, refuses when the base is not derivable, and prints the
report the rest of this section asks you to read:
python3 tools/harvest-worktree.py <worktree> --out <patch> # 0 harvested, 3 no work,
# 1 refused, 2 usage
Never a branch name, because the branch has moved since the agent started. An
orchestrator here took git diff main from a worktree whose main had advanced, and
applying that patch silently reverted a sibling task's test cases. Nothing failed. It was
caught because a case count dropped, and only because someone looked.
And not its own HEAD either, whenever the agent committed — this section used to say
HEAD and that is right only for a worktree that left everything uncommitted. An agent that
committed has moved HEAD past its own base, so git diff HEAD shows the dirty tree alone
and silently drops every committed change. The base is the commit the worktree started
at, which the tool reads out of that worktree's own HEAD reflog; hand-composing it is where
both mistakes live.
What ships with it
1 file 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.
- 26d ago First seen · 159 lines · 0 tokens per session scan A 39dfe6ae12c4
orchestrating-parallel-work is a skill published in the GitHub repository AleksandarBisevac/claude-plugins (4 stars, last pushed yesterday), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,051 tokens. 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 2026-09-18.
Other skills, from other repositories
implementer
Delegate a coding task to Google's Antigravity CLI (agy staffer, fast Gemini), which edits the working tree directly and can perform explicitly requested Git delivery. Use when the user says /agy:implementer, "have agy fix/build X", or wants to hand a well-scoped coding task to the agy staffer instead of doing it in…
touch-audit
Mobile audit — app size, startup time, crash reporting, store compliance, accessibility, offline behavior. Use when asked for "mobile review", "app store readiness", "mobile performance", or "crash analysis".
atlas-changelog
Maintain per-repo and cross-repo changelogs — append structured entries after agent work. Use when asked to "log this change", "update changelog", "what changed", "change history".
prism-audit
Frontend audit — bundle size, dependencies, accessibility, performance, component quality. Use when asked for "frontend review", "performance audit", "accessibility check", or "bundle size".
forge-audit
Audit existing infrastructure for security issues, waste, and misconfigurations. Use when asked to "audit my infra", "check cloud setup", "infra review", "are we wasting money", "security check on infra", or "review my terraform".
lens-audit
Review existing analytics — find all dashboards and reports, check who uses them, whether metrics are defined, and whether they drive decisions. Recommend what to keep, kill, or add. Use when asked "are our dashboards useful", "analytics review", or "metrics audit".