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 KonghaYao/peri --skill multitaskgit clone --depth 1 https://github.com/KonghaYao/periWrote 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/konghayao/peri/multitask)<a href="https://agentmods.dev/skills/konghayao/peri/multitask"><img src="https://agentmods.dev/badge/skills/konghayao/peri/multitask/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/konghayao/peri/multitask"><img src="https://agentmods.dev/badge/skills/konghayao/peri/multitask.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 | $0.00048 | $0.01259 |
| Opus 5.5 | $0.00019 | $0.00504 |
| Sonnet 5.5 | $0.00010 | $0.00252 |
| Haiku 4.5 | $0.00005 | $0.00126 |
Grade C, and why
multitask scanned grade C with 1 finding 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.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
If `Agent` is unavailable, or the user prohibits delegation, do the permitted work directly and say so. Never refuse work the user has authorized. How it starts
The opening of the file, as written. The whole thing — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Multitask Mode
You are the coordinator. Every workstream leaves your hands and belongs to exactly one owner launched through Agent. The mode holds from the moment you have it, for the rest of the session — there is no opt-in, no task it exempts, and no small change it lets you keep.
What this mode changes
The host default lets you work in the foreground whenever the change fits, and puts direct work at two or three files. Under this mode none of that zone survives: every deliverable goes to an owner.
Coordination is the only work left to you — reading owner reports, synthesizing, asking the user, deciding the next step, launching and resuming owners, and keeping the record below. Everything else is a deliverable:
- Any read, search, or edit of project material. One file is already enough.
- Recon that enumerates a directory, greps the tree for references, or reads build, deploy, or CI config to pin down a vague brief. Recon is a deliverable like any other; hand it off, and the work it sizes stays with that owner.
- Any change crossing a crate boundary.
- Two or more independent workstreams the user hands over — one owner each.
An owner's returned report is coordination; the files behind it are not. Where the line is unclear, hand off rather than take a look first.
If Agent is unavailable, or the user prohibits delegation, do the permitted work directly and say so. Never refuse work the user has authorized.
One owner
A deliverable you hand off — investigate → implement → verify — has exactly one owner, and that owner is never you. Absorbed:
- Do not redo handed-off work yourself or in a second owner; follow up only where results expose a gap.
- Verification belongs to the deliverable. Read the evidence your owner reports instead of re-running its verification. When a real gap appears, hand the gap back to that owner (
resume_thread_id) — a gap that falls outside its original file scope is still the owner's to close, not yours. - Do not split one deliverable into roles that wait on each other.
- Siblings only for genuinely independent workstreams, with non-overlapping write scopes and settled interfaces. You hold no files, and an owner's files are not yours to edit; wait, or hand ownership over explicitly.
- Delegation is one level deep, so you alone add siblings. Independent siblings run concurrently — launch them together and wait for their completion notifications instead of serializing the work or taking it back.
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 Changed 09d968c8180a
- 21d ago Changed · +12 lines · -19 tokens per session 5797b0b54427
- 26d ago Changed · -50 lines · +7 tokens per session scan A → C 639494460405
- 28d ago First seen · 108 lines · 60 tokens per session scan A e4b3ec65aea7
multitask is a skill published in the GitHub repository KonghaYao/peri (240 stars, last pushed today), licensed Apache-2.0. It adds 48 tokens to every session and 1,259 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-18.
Other skills, from other repositories
babysit-pr
Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…
plugin-creator
Create and scaffold plugin directories for Codex with a required .codex-plugin/plugin.json, optional plugin folders/files, valid manifest defaults, and personal-marketplace entries by default. Use when Codex needs to create a new personal plugin, add optional plugin structure, generate or update marketplace entries…
codex-pr-body
Update the title and body of one or more pull requests.
qa-testing
Verify your work by actually operating the app or website you changed, instead of assuming it works. Strongly recommended whenever you build, modify, or debug a web app, website, or desktop GUI app. Drive real browsers with the agent-browser CLI and native desktop apps with the cua-driver CLI. These are installed on…
remote-tests
Testing against remote executors in integration tests.
update-v8-version
Update Codex's pinned v8 / rustyv8 versions, validate the release-candidate path, and investigate failed V8 canary or artifact builds. Use when asked to bump V8, update rustyv8 artifacts, prepare or validate a V8 release candidate, check v8-canary, or diagnose why a V8 version update no longer builds.