A manual quality-assurance harness for the senpi coding agent. It runs the real command-line tool in an isolated temporary environment and checks what users can observe.

Use it after changing senpi's AI, agent, coding-agent, or terminal interface packages. It tests four communication channels and includes self-checks for its helper scripts.

It catches problems that type checks and automated unit tests may miss. It also keeps tests away from real credentials and user data.

How code-yeongyu/senpi configures Claude Code and Codex

Reuse

Borrowing it

Nothing to install: this file belongs to code-yeongyu/senpi. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/code-yeongyu/senpi/main/.agents/skills/senpi-qa/SKILL.md

Clone the repo

git clone --depth 1 https://github.com/code-yeongyu/senpi

Made for: Claude Code, Codex.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/code-yeongyu/senpi/senpi-qa/github.svg)](https://agentmods.dev/skills/code-yeongyu/senpi/senpi-qa)

Your own site

<a href="https://agentmods.dev/skills/code-yeongyu/senpi/senpi-qa"><img src="https://agentmods.dev/badge/skills/code-yeongyu/senpi/senpi-qa/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/code-yeongyu/senpi/senpi-qa"><img src="https://agentmods.dev/badge/skills/code-yeongyu/senpi/senpi-qa.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 201 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 2,909 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits

NVIDIA SkillSpector warn 7 Sept 2026

SkillSpector: 3 findings, up to high

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

high Privilege Escalation · line 39
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Fix: Remove references to credential paths. Use environment variables or secrets managers. For docs, use placeholder paths (e.g., /path/to/config). Never load .env or token files in production code paths.
medium Rogue Agent · line 21
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
medium Agent Snooping · line 39
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Fix: Remove all code or instructions that list or read other skills' files or directories. Skills should operate independently; cross-skill access is a privilege escalation.

How audits are shown

Origin original No closer match found in the catalogue.

Token cost

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	$0.00201	$0.02909
Opus 5.5	$0.00080	$0.01164
Sonnet 5.5	$0.00040	$0.00582
Haiku 4.5	$0.00020	$0.00291

Measured 2mo ago against content hash e0ac7799dac4, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

senpi-qa 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 2mo ago.

The scan reads SKILL.md. This mod also ships 49 executable files (scenarios/fused-equivalence/cases/e01-parallel-reads.mjs, scenarios/fused-equivalence/cases/e02-global-sequential.mjs, scenarios/fused-equivalence/cases/e03-per-tool-waves.mjs, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

.agents/skills/senpi-qa/SKILL.md · 189 lines

How it starts

The opening of the file, as written. The whole thing — 189 lines — stays where its author put it; the contents beside it link to each section on GitHub.

senpi QA

QA the senpi coding agent (packages/{ai,agent,coding-agent,tui}) by driving the REAL CLI — not by reading code or trusting unit tests. Each channel runs the agent from source via tsx in an isolated sandbox and asserts observable behavior, so a passing run is evidence the user-facing surface actually works.

Every helper script ships a --self-test (or --self-check) that asserts its scenario against this machine. The scripts are therefore both the QA tools and their own regression checks.

Golden rules (read before running anything)

Isolation is mandatory. Everything spawns the CLI with SENPI_CODING_AGENT_DIR / SENPI_CODING_AGENT_SESSION_DIR pointed at a temp sandbox and PI_OFFLINE=1. QA must never write into the real ~/.senpi. scripts/lib/common.mjs does this for you — use it.
Never read or modify the real credentials. ~/.senpi/agent/auth.json is the user's real key store. Every script snapshots its sha256 and asserts it is unchanged at the end. If you script a run by hand, do the same (guardRealAuth() in common.mjs).
Deterministic loop = mock loop. To exercise the agent loop without real tokens, use Channel 3 (a local fake model server). Real-provider runs are for final smoke only and must use the user's existing auth, never a new key.
No src/ edits from this skill. It verifies; it does not fix. If QA finds a bug, report it with the captured evidence and let a follow-up change fix it.
The captured artifact IS the evidence. Write it under local-ignore/qa-evidence/<YYYYMMDD>-<slug>/. No artifact == the QA did not happen. local-ignore/ is gitignored — never commit evidence.

Setup (once)

node scripts/devenv-setup.mjs        # installs skill deps (node-pty), wires .env.local + .claude/skills
node .agents/skills/senpi-qa/scripts/lib/common.mjs --self-check   # confirm the harness

common.mjs --self-check confirms the repo resolves, a sandbox is created and auto-removed, a free port is allocatable, and the real auth file is untouched.

Read the full file on GitHub · 189 lines

Files

What ships with it

60 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.

Changes

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.

2mo ago First seen · 189 lines · 201 tokens per session scan A e0ac7799dac4

Subscribe to this mod's changes

senpi-qa is a skill published in the GitHub repository code-yeongyu/senpi (511 stars, last pushed yesterday), licensed MIT. It adds 201 tokens to every session and 2,909 once invoked, about $0.0008 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 2026-08-30.

E2E testing LLM Testing agent ai-agent cli coding-agent llm-tools

senpi-qa next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
senpi-qacode-yeongyu/senpi	Skill	511	yesterday	MIT	A	201	Claude Code, Codex
transformerssynthetic-sciences/openscience	Skill	4.3k	2d ago	Apache-2.0	A	63	Claude Code, Codex
knowledge-engineering-quality-and-delivery-real-api-qualification-and-e2eechoVic/blade-code	Skill	181	24d ago	MIT	A	182	Claude Code, Codex
knowledge-engineering-quality-and-delivery-unit-integration-and-shared-test-harnessesechoVic/blade-code	Skill	181	24d ago	MIT	A	184	Claude Code, Codex
segment-anything-modelsynthetic-sciences/openscience	Skill	4.3k	2d ago	Apache-2.0	A	45	Claude Code, Codex
hugging-face-clisynthetic-sciences/openscience	Skill	4.3k	2d ago	Apache-2.0	A	68	Claude Code, Codex
autogpt-agentssynthetic-sciences/openscience	Skill	4.3k	2d ago	Apache-2.0	A	39	Claude Code, Codex

senpi: Skill for Claude Code