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 pome-sh/digital-twins --skill pome-run-taskgit clone --depth 1 https://github.com/pome-sh/digital-twinsWrote 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/pome-sh/digital-twins/pome-run-task)<a href="https://agentmods.dev/skills/pome-sh/digital-twins/pome-run-task"><img src="https://agentmods.dev/badge/skills/pome-sh/digital-twins/pome-run-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/pome-sh/digital-twins/pome-run-task"><img src="https://agentmods.dev/badge/skills/pome-sh/digital-twins/pome-run-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 | $0.00123 | $0.02563 |
| Opus 5 | $0.00062 | $0.01282 |
| Sonnet 5 | $0.00025 | $0.00513 |
| Haiku 4.5 | $0.00012 | $0.00256 |
Grade A, and why
pome-run-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 6d 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 — 161 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Pome run task (Skill 4)
You are the coach: you talk to the builder and to the Pome control MCP
(mcp.pome.sh). The examinee is the sandbox clone Skill 0 (pome-intake)
registered, launched here for real against live twins. This skill runs one
already-verified task and scores it from the twin tape. Verify the seed
first (pome-verify-seed) — this skill does not re-check fairness.
Only one step here is examinee-runtime-specific: launching the examinee.
Minting, finalizing, scoring, reporting, and the fix loop are runtime-agnostic
(ADR-018). The launch policy — always_allow, closed-book web tools, memory
snapshot-clone, the network clamp — is owned by the examinee_launch spec, not
by this prose: you execute the spec, you do not restate it.
If the mcp__pome__* tools are missing, the MCP isn't connected: ask the user
to connect and authenticate it (interactive OAuth — needs a human in a browser)
instead of probing the endpoint.
SENSITIVE — the bearer. run_task returns agent_token, a
session-scoped JWT that is the bearer for every twin URL — a live credential.
It exists for one purpose: handing the examinee its twin authentication at
launch (§2). Pass it straight into the launcher's env or vault and then let go
of it — never write it to disk, into a task, or a log, never repeat it back to
the builder, and never keep it around "for later". Nothing downstream needs it:
finalize_run derives the bearer from session_id server-side. It dies at
expires_at regardless.
1. Mint the run
Mint a grp_-prefixed group_id now and reuse it for every trial of this
attempt — the baseline and any pre-fix flaky retries — so they aggregate as
one exam (aggregation keys on (group_id, task); never reuse a group_id
across different tasks). A post-fix rerun is the exception: it opens a new
group_id and links back to the baseline via baseline_group_id (see §5).
Then call run_task(task_id, agent_id, agent_version, group_id) (the
agent_id from intake). It seeds live twin sandboxes and returns session_id,
expires_at, agent_token, examinee_task (the prompt + twins the examinee
sees — no criteria), and examinee_launch (the full launch spec).
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.
- 6d ago Changed · +3 lines 0b3c47c7617b
- 10d ago First seen · 158 lines · 123 tokens per session scan A b70fba85ff24
pome-run-task is a skill published in the GitHub repository pome-sh/digital-twins (14 stars, last pushed 4d ago), licensed Apache-2.0. It adds 123 tokens to every session and 2,563 once invoked, about $0.0006 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 2026-08-30.
Other skills, from other repositories
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
cw-gates
Use before claiming any Codewhale change is done, green, or ready to land: the focused-to-broad verification ladder, the budget checks CI enforces, and the rules for what counts as a passing test.
codew-release-qa-sweep
Use before claiming Codewhale release work is done: run the full gate sweep and list the manual QA targets.
verify
Exercise the real app/API/CLI and collect observable evidence; tests alone do not count as end-to-end verification.
webapp-testing
Start/reuse a local app, wait for readiness, inspect rendered state/console/network, act from observed selectors, and verify with evidence.
mcp-app-verification
Comprehensive verification checklists for MCP Apps. Tests with basic-host reference, validates handler-before-connect, text fallback, resource URI linking, single-file bundling, host styling, CSP, and legacy pattern detection.