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 agentlas-ai/agentlas-desktop --skill direct-studygit clone --depth 1 https://github.com/agentlas-ai/agentlas-desktopWrote 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/agentlas-ai/agentlas-desktop/direct-study)<a href="https://agentmods.dev/skills/agentlas-ai/agentlas-desktop/direct-study"><img src="https://agentmods.dev/badge/skills/agentlas-ai/agentlas-desktop/direct-study/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/agentlas-ai/agentlas-desktop/direct-study"><img src="https://agentmods.dev/badge/skills/agentlas-ai/agentlas-desktop/direct-study.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 | $0.00055 | $0.04068 |
| Opus 5.5 | $0.00022 | $0.01627 |
| Sonnet 5.5 | $0.00011 | $0.00814 |
| Haiku 4.5 | $0.00006 | $0.00407 |
Grade A, and why
direct-study 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 14d 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 — 228 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Direct One Study
Consult ../../agent/agent.md, ../../agent/soul.md, the durable state schema and
../write-manuscript/SKILL.md when useful. Reading or acknowledging a manual is never an execution
prerequisite. The steps below are references to tools and evidence bindings, not a required order.
Default operating mode
Run autonomously: question -> acquire available data -> exploratory test -> inspect the result ->
revise the hypothesis -> next experiment. Search literature in parallel, guided by the results.
Data collection, feasibility scripts and exploration need no completed literature review or approved
analysis plan. Keep exploration distinct from confirmatory inference; formal plans and journal
validation apply only where the actual operation or final claim needs them. Open each result as it is produced and continue. Do not stop between stages
for confirmation. Stop only for a host-required human receipt that is actually pending under the
project's approval policy, a journal manual attestation, a Research Contract that Main returned
as a draft/checkpoint, or a genuine fork: two or more materially different directions, an ambiguous
request, or a missing required input. Main's exact Research Contract status approved plus its
approval receipt, including a standing-policy receipt, satisfies contract approval and requires no
second confirmation. The same rule applies when freeze_analysis_plan returns a frozen plan with
an exact standing-policy approval: proceed with its returned version and hash, and never claim a
human personally reviewed it. Incomplete plans and unresolved design decisions still fail closed.
Ask with concrete options and a recommendation ("A or B? I recommend A
because ..."). A bounded request ("just the literature review") ends at that scope with a report and
a proposed next step; an unqualified study request remains active until its criteria are verified or
a recorded blocker, budget exhaustion, or deadline stops it.
Record completion_scope: "full-study" for the whole study, or "bounded-deliverable" when the
researcher explicitly limits the deliverable. For full-study work, keep the same loop and selected
model through manuscript revision and the verified ready_to_submit gate. Inspect Main's progress
and continuation receipts after each turn. Repeated proposals, unchanged lifecycle updates, or
rewritten chat summaries do not establish progress. When Main reports a no-progress retry, choose
a materially different evidence-bound action or persist the concrete blocker.
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.
- 14d ago Changed · +3 lines · -8 tokens per session b7c7644122ae
- 21d ago Changed · +2 lines a39d52ee1733
- 24d ago Changed · +2 lines ff45485d3a33
- 1mo ago Changed · +28 lines d89af86284ce
- 1mo ago First seen · 193 lines · 63 tokens per session scan A df700b074b76
direct-study is a skill published in the GitHub repository agentlas-ai/agentlas-desktop (6 stars, last pushed today), licensed Apache-2.0. It adds 55 tokens to every session and 4,068 once invoked, about $0.0002 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 2026-09-04.
Other skills, from other repositories
tbtools
Use when the user asks about TBtools, TBtools-II, TBtools RPC API, TBtools CLI, or bioinformatics operations available through TBtools such as sequence manipulation, BLAST, GFF/GTF/GXF processing, expression tables, heatmaps, trees, MCScanX, DIAMOND, HMMER, MUSCLE, and IQ-TREE.
electron-expert
Expert in Electron framework, desktop app development, IPC, and cross-platform packaging. Use when the user mentions desktop, Node.js, cross platform, Windows, macOS, or Linux, or when the task involves Electron Architecture, Process Types, IPC Communication, or App Lifecycle.
autoresearch
Run an auto research / autoresearch loop for measurable improvements. Use when optimizing a skill, agent, prompt, code path, benchmark, or other target with a mechanical metric: baseline, mutate one thing, verify, keep or discard, log, chart progress, and optionally save improved custom skills or agents through Open…
architecture-diagram
How to draw and maintain the architecture diagram Ensemblr keeps for a workspace — ensemblrgetarchitecturediagram to fetch the stored one (it answers with null when nobody has drawn this workspace yet), the IR shape, how to derive it from the codebase, and ensemblrupdatearchitecturediagram to store it. Read this when…
research-orchestrator
Becomes a principal research scientist who orchestrates deep, multi-source research by coordinating evidence gathering, grading source quality, and synthesizing findings with explicit confidence levels. Use when the user needs a comprehensive research report on a complex topic, evidence-based analysis with source…
ensemblr
How Ensemblr works from inside it — the workspace/git-worktree model, the ensemblr control tools and what their answers mean, the committed .ensemblr/settings.toml (setup and run scripts, [git], [prompts], [infisical], environment variables), /.config/ensemblr/config.json, the Changes panel and diff comments, the…