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 heyhuynhgiabuu/pi-task --skill taskgit clone --depth 1 https://github.com/heyhuynhgiabuu/pi-taskWrote 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/heyhuynhgiabuu/pi-task/task)<a href="https://agentmods.dev/skills/heyhuynhgiabuu/pi-task/task"><img src="https://agentmods.dev/badge/skills/heyhuynhgiabuu/pi-task/task/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/heyhuynhgiabuu/pi-task/task"><img src="https://agentmods.dev/badge/skills/heyhuynhgiabuu/pi-task/task.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 | $0.00024 | $0.00989 |
| Opus 5.5 | $0.00010 | $0.00396 |
| Sonnet 5.5 | $0.00005 | $0.00198 |
| Haiku 4.5 | $0.00002 | $0.00099 |
Grade A, and why
task 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 3d 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 — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Task delegation
Use this guide only in a parent session with the delegation tool available (default task, renamed via PI_TASK_TOOL_NAME). The examples use the default name; call the actual tool name exposed in your session. This guide does not enable nested delegation in children. The tool's current agent list is authoritative; installed and project agents can override the bundled agents.
Delegate bounded, multistep work that benefits from independent judgment or fresh context. Do small lookups and straightforward edits directly. Choose the agent whose advertised scope fits; do not invent an agent name. Honor the parent session's approval, write, and side-effect constraints.
Complete the handoff
A child starts with fresh context. Referenced files are evidence, not a replacement for the parent's reasoning. Put the following in prompt: goal, scope, non-goals, write policy, acceptance criteria, verification recipe. Name relevant files, the review base or diff, and stop conditions explicitly.
Use parent_context for facts, decisions, and constraints the parent learned outside those files. Use proposed_changes for one item per intended change: its semantics and what acceptance requires. Do not ask the child to reconstruct an earlier conversation from file paths.
Reviewer tasks require explicit parent_context and non-empty proposed_changes. If there are no design changes, say so in an item rather than omitting the field.
{
"agent_type": "reviewer",
"description": "Audit retry settings",
"prompt": "Goal: audit the uncommitted retry-settings diff. Scope: src/subagent/durable.ts and its tests. Non-goals: unrelated cleanup. Write policy: read-only; no paid requests. Acceptance: ranked findings with file:line evidence and a verdict. Verification: read the diff, trace retry decisions, and run the focused tests.",
"parent_context": "The parent approved generation-level retry settings only; provider stream options are out of scope.",
"proposed_changes": [
"Durable children read the parent's retry policy live; cached harnesses must observe updated settings."
]
}
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.
- 3d ago First seen · 65 lines · 24 tokens per session scan A ab2745d4fd2f
task is a skill published in the GitHub repository heyhuynhgiabuu/pi-task (62 stars, last pushed 3d ago), licensed MIT. It adds 24 tokens to every session and 989 once invoked, about $0.0001 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 2026-10-13.
Other skills, from other repositories
skill-creator
Create or update Agent Skills (SKILL.md plus optional scripts, references, or assets). Use when someone asks to design a new Agent Skill, refine an existing one, or structure skills for Pi discovery, packaging, or other Agent Skills-compatible clients.
pi-ralph-wiggum
Long-running iterative development loops with pacing control and verifiable progress. Use when tasks require multiple iterations, many discrete steps, or periodic reflection with clear checkpoints; avoid for simple one-shot tasks or quick fixes.
pi
Use when needing a minimal, extensible terminal AI coding agent harness in TypeScript with plugin architecture. Pi: minimalist terminal AI coding agent emphasizing simplicity and composability.
ask-user
Use askuser as a decision, research, and requirements gate before ambiguous or high-stakes choices.
extending-pi
Guide for changing or extending Pi's behaviour. Use when someone wants to modify how Pi behaves, add capabilities, decide which Pi extension point or artifact to use, build or package an extension, create an Agent Skill, add prompt templates/themes/context/model providers, configure Pi resources, or asks whether a Pi…
typesafe-ai
Build AI-powered software with TypeSafe and Jev, or use pi-typesafe's typesafequestion tool for typed semantic judgments and probabilities. Use when a feature needs programmable common sense, when an LLM prompt-and-parse step could become a structured decision, or when designing, calling, evaluating, or…