Borrowing it
Nothing to install: this file belongs to l-nmch/hailo-10h-llm-compiler. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/l-nmch/hailo-10h-llm-compiler/main/.claude/skills/new-finding/SKILL.mdgit clone --depth 1 https://github.com/l-nmch/hailo-10h-llm-compilerWrote 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/l-nmch/hailo-10h-llm-compiler/new-finding)<a href="https://agentmods.dev/skills/l-nmch/hailo-10h-llm-compiler/new-finding"><img src="https://agentmods.dev/badge/skills/l-nmch/hailo-10h-llm-compiler/new-finding/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/l-nmch/hailo-10h-llm-compiler/new-finding"><img src="https://agentmods.dev/badge/skills/l-nmch/hailo-10h-llm-compiler/new-finding.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 | $0.00046 | $0.00627 |
| Opus 5 | $0.00023 | $0.00313 |
| Sonnet 5 | $0.00009 | $0.00125 |
| Haiku 4.5 | $0.00005 | $0.00063 |
Grade A, and why
new-finding 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 — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Writing a finding page
A finding page is the unit of knowledge transfer in this repo. It exists so the next investigation (yours in six months, or someone else's) starts from the conclusion instead of from zero.
Shape (fixed, don't deviate)
- Symptom — what was observed, in terms a reader can reproduce (error message, numeric mismatch, behavior).
- Investigation — what was tried, including what did not work and what it ruled out. Negative results are as valuable as the fix.
- Root cause — the actual mechanism, backed by a source pointer (public HailoRT/DFC source, header, or structural comparison) wherever possible.
- Fix — the concrete change, with a file:line pointer into
pipeline/orruntime/if one exists. - Verification — the reproducible evidence: cosine similarity,
argmax match, structural diff of
.hn/HEF layout. A finding without a number attached is a hypothesis, not a finding.
If the bug is still open, write the same shape minus Fix, and add a "Current best hypothesis" + "Reproduce it" section (see open-tbt-cache-read.md for the template to copy).
Rules while writing
- No dates or timelines — the repo is evergreen.
- No proprietary excerpts — describe behavior and quote short identifiers (function/key names); don't paste long blocks of proprietary source. Link the public repo instead.
- No official HEFs/
.alls/.hngraphs, no firmware, no DFC wheels — neither pasted inline nor referenced as attached files. - Claims need evidence of a kind listed in docs/findings/index.md's "How to read these" section — structural comparison, public source reading, or on-hardware numerics.
After writing
- Add a row to the summary table in docs/findings/index.md.
- Update docs/status.md if the finding changes what works or what doesn't at the stage-by-stage level.
- If the finding closes an item in Investigation-Chronicle or opens a new one, note it there too — the wiki narrates, the repo is canonical (docs/ wins on conflict).
- Run the
preflightskill before committing.
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 · 56 lines · 46 tokens per session scan A 3ee3c5bed808
new-finding is a skill published in the GitHub repository l-nmch/hailo-10h-llm-compiler (14 stars, last pushed 10d ago), licensed MIT. It adds 46 tokens to every session and 627 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
trulens-evaluation-setup
Configure feedback functions and selectors for TruLens evaluations.
trulens-instrumentation
Instrument LLM apps with TruLens OTEL-based tracing - from setup to debugging and optimization.
trulens-diagnosis
Diagnose low evaluation scores and generate actionable improvement recommendations.
trulens-evaluation-workflow
Systematically evaluate your LLM application with TruLens.
trulens-blocking-guardrails
Configure and use feedback functions as runtime blocking guardrails.
trulens-dataset-curation
Create and curate evaluation datasets with ground truth for TruLens.