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 deepklarity/harness-kit --skill hk-follow-breadcrumbgit clone --depth 1 https://github.com/deepklarity/harness-kitWrote 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/deepklarity/harness-kit/hk-follow-breadcrumb)<a href="https://agentmods.dev/skills/deepklarity/harness-kit/hk-follow-breadcrumb"><img src="https://agentmods.dev/badge/skills/deepklarity/harness-kit/hk-follow-breadcrumb/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/deepklarity/harness-kit/hk-follow-breadcrumb"><img src="https://agentmods.dev/badge/skills/deepklarity/harness-kit/hk-follow-breadcrumb.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00172 | $0.01211 |
| Opus 5 | $0.00086 | $0.00606 |
| Sonnet 5 | $0.00034 | $0.00242 |
| Haiku 4.5 | $0.00017 | $0.00121 |
Grade A, and why
hk-follow-breadcrumb 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 11d 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 — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/hk-follow-breadcrumb — Breadcrumb-First Exploration
This skill is about reading before searching. The docs/breadcrumb_analysis/ directory contains end-to-end workflow traces that have already been carefully researched and documented. Loading a relevant breadcrumb takes seconds and costs a fraction of what spawning exploration subagents costs. The breadcrumb docs are maintained alongside code changes, so they're the most reliable "map" of how flows work.
Context
<exploration_context> $ARGUMENTS </exploration_context>
If the context above is empty, look at what the user is asking about in the conversation and infer the exploration target.
Step 1: Load the index
Read docs/breadcrumb_analysis/_INDEX.md. This is the single source of truth for what's been documented.
The index has two sections that matter:
- Flows table — lists every breadcrumb folder and what it traces. Scan this to find which breadcrumb(s) cover the user's question.
- Quick navigation — maps common symptoms (e.g., "task stuck in IN_PROGRESS", "screenshots not showing") directly to the right DEBUG.md. If the user describes a symptom, check here first.
Step 2: Match the question to breadcrumbs
Compare what the user is asking against the index entries. There are three outcomes:
A. Direct hit — a breadcrumb covers exactly this flow
Load the relevant docs in order of increasing detail:
-
FLOW.md — Read this first. It's the high-level map (30-second read). Shows the path data takes, which files/functions are involved at each hop, and where branches occur. This alone often answers "where does X happen?" questions.
-
DETAILS.md — Read this if the user needs function-level specifics: what data goes in/out, what conditionals change behavior, what side effects exist. This answers "how does X work?" questions.
-
DEBUG.md — Read this if the user is debugging something broken, or wants to know how to test a flow. Contains log locations, grep patterns, quick commands, env vars, and common breakpoints. This answers "how do I debug/test X?" questions.
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.
- 11d ago First seen · 101 lines · 172 tokens per session scan A ca6553d2e8d4
hk-follow-breadcrumb is a skill published in the GitHub repository deepklarity/harness-kit (97 stars, last pushed 1mo ago), licensed MIT. It adds 172 tokens to every session and 1,211 once invoked, about $0.0009 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
neuron-debugger
Debug and monitor Neuron AI applications with Inspector APM, event observability, logging, and performance analysis. Use this skill whenever the user mentions debugging, monitoring, observability, performance analysis, tracing, Inspector, or needs to understand why an agent is behaving a certain way. Also trigger for…
diag-harness
Kernel-version skew check (ADR-027). Reports manifest surface + manifest kernel + installed kernel + verdict (match/patch-diff/minor-diff/major-diff). Exits 1 on minor/major skew with a copy-pasteable npm install @metaharness/[email protected] next step. Exits 2 if no .harness/manifest.json at path.
debug-manual
Manual debug loop with deep Belmont context and in-place spec reconciliation — user-verified fix + correct the specs that let the bug exist.
debug-auto
Auto debug loop — investigate, fix, verify using agent-dispatched pipeline.
debug
Debug router — choose between auto (agent-verified, narrow code fix) and manual (user-verified, deep spec context, in-place spec reconciliation).
neuron-debugger
Debug and monitor Neuron AI applications with Inspector APM, event observability, logging, and performance analysis. Use this skill whenever the user mentions debugging, monitoring, observability, performance analysis, tracing, Inspector, or needs to understand why an agent is behaving a certain way. Also trigger for…