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 nikzlabs/shipit --skill probegit clone --depth 1 https://github.com/nikzlabs/shipitWrote 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/nikzlabs/shipit/probe)<a href="https://agentmods.dev/skills/nikzlabs/shipit/probe"><img src="https://agentmods.dev/badge/skills/nikzlabs/shipit/probe/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/nikzlabs/shipit/probe"><img src="https://agentmods.dev/badge/skills/nikzlabs/shipit/probe.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 | $0.00028 | $0.01035 |
| Opus 5.5 | $0.00011 | $0.00414 |
| Sonnet 5.5 | $0.00006 | $0.00207 |
| Haiku 4.5 | $0.00003 | $0.00103 |
Grade A, and why
probe 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 19d 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.
What it actually says
The probe report
This skill ships with ShipIt's test plugin (docs/262). If you can read this
text through your skill listing, skills materialization (req 22) works: the
file was copied from the plugin checkout into your backend's discovery root
under a plugins--<alias>--probe namespace.
Run probe in the project workspace. It prints one JSON report and mutates
nothing. Flags: --bump increments the shared counter before reporting;
--host-check also tests HTTPS egress to the declared host (example.com),
which the manifest declares optional (reqs 23, 24) — the probe reports
whether it got through and exits 0 either way, so the Plugins card offers that
host rather than counting it as an unmet need. PROBE_TOKEN stays required, so
the fixture carries one of each. --stdin reads stdin to its end and reports
what arrived; without it the probe does not touch stdin, because a command that
reads stdin to its end waits until the caller's stdin ends. --hold <seconds>
makes the call run that long first, and records how long it ran.
| Field | Verifies |
|---|---|
cwd |
CLI runs with cwd = the project workspace (req 21) |
mode / env.SHIPIT_PLUGIN_COMMIT |
the self/consumer discriminator: a consumer generation carries its exact commit; repo: self runs the live tree, no commit (reqs 15, 27) |
credential.set |
PROBE_TOKEN injected for this command only (req 23) |
settings.greeting |
the validated settings file (req 26), and which side wrote it: the consuming project's overrides.settings is hello from the consuming project, the manifest's default is hello from the probe. Reading the default under this repo's own shipit.yaml, which sets the override, means the consumer half was ignored (nikzlabs/shipit#2298 finding 2) |
project.readable |
the workspace handle (req 21) |
state.counter |
shared state with the probe service — probe --bump, then reload its page: same number (reqs 17, 18) |
checkout.writable |
this surface's own mount — not the self/consumer signal (read mode), but it does check the rule directly: the plugin's tree is writable exactly when it is the project, so a consumer generation reports false and repo: self reports true (reqs 7, 15, 27). On the service surface it measures this fragment's own .:/app, which ShipIt forces read-only for a tracked generation and leaves as declared under repo: self |
stdin (only with --stdin) |
the command's stdin arrives whole, and while the caller still writes (req 17): bytes and sha256 equal the input's, and firstByteAfterMs / endAfterMs say when the probe read them. Give it valid UTF-8 (test-plugin/README.md has the fields) |
hold (after a --hold call) |
what became of the last held call: finished: false with a ranMs shorter than its seconds is a command that was killed before its end — the check that ShipIt stops a command whose caller went away (req 32; test-plugin/README.md has the fields) |
install.matchesActiveCommit |
install ran for the active generation; null under self-use, which has no generations (req 7) |
dependency.project / .plugin |
a real dependency loads from each mount of the tree, reported per root because /plugin and /project are separate mounts even when they are one tree. Under repo: self the CLI must show both resolved: true, used: true — the working tree's own agent.install prepares the tree the CLIs and services run out of (req 27); the service shows plugin: resolved false, since its fragment mounts only its own directory at /app. This is the one field that would have caught nikzlabs/shipit#2298 (test-plugin/README.md has the full table) |
The probe service renders the same report from the service surface at its
preview URL, with a counter button and an Agent Interface SDK section: it
awaits window.shipit.ready, checks embedded, and its "Send counter to
agent" button exercises the browser-to-agent path (req 3).
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.
- 19d ago Changed · +5 lines 356b35704544
- 1mo ago Changed · +4 lines ba43239eae06
- 2mo ago First seen · 33 lines · 28 tokens per session scan A 6199139f26a2
probe is a skill published in the GitHub repository nikzlabs/shipit (9 stars, last pushed today), licensed Apache-2.0. It adds 28 tokens to every session and 1,035 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 2026-08-31.
Other skills, from other repositories
ralph
Self-referential loop until task completion with configurable verification reviewer.
tdd
Test-first at pre-agreed seams — one test, one implementation, tracer-bullet red/green; independent expected values; substitutions only at the seam's boundary class; launch Phase 4 reference.
moai-ref-testing-pyramid
Test pyramid strategy, coverage targets, test patterns, and quality metrics reference. Agent-extending skill that amplifies manager-develop test-creation and quality-validation work with production-grade testing patterns. NOT for: production code implementation, architecture design, DevOps, security audits.
tdd
Build this test-first, one behavior at a time: $ARGUMENTS.
moai-workflow-loop
Ralph Engine - Automated feedback loop with LSP diagnostics and AST-grep integration for continuous code quality improvement. Use when implementing error-driven development, automated fixing, or continuous quality validation workflows.
moai-workflow-testing
Use when writing tests, measuring coverage, or running characterization, performance, or PR-review QA. Comprehensive specialist combining DDD testing, characterization tests, performance profiling, and TRUST 5 quality-assurance validation.