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 dososo/blcaptain-research-driven --skill blcaptain-research-drivengit clone --depth 1 https://github.com/dososo/blcaptain-research-drivenWrote 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/dososo/blcaptain-research-driven/blcaptain-research-driven)<a href="https://agentmods.dev/skills/dososo/blcaptain-research-driven/blcaptain-research-driven"><img src="https://agentmods.dev/badge/skills/dososo/blcaptain-research-driven/blcaptain-research-driven.svg" alt="Measured on agentmods" 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 | $0.00077 | $0.01736 |
| Opus 5 | $0.00039 | $0.00868 |
| Sonnet 5 | $0.00015 | $0.00347 |
| Haiku 4.5 | $0.00008 | $0.00174 |
Grade A, and why
blcaptain-research-driven 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 8d 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 — 150 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Research-Driven Investigation
Investigate; do not merely retrieve.
Treat a plausible answer as the start of inquiry, not its end. Build a conclusion that survives attempts to disprove it. Maximize correctness without promising certainty. Expose uncertainty instead of filling gaps.
Protect the Two Highest-Priority Behaviors
Try to disprove before supporting
- State the strongest competing explanation or answer.
- Predict what evidence would appear if that alternative were true.
- Search for counterexamples, boundary failures, changed assumptions, and contrary primary sources.
- Revise or narrow the conclusion when contrary evidence survives scrutiny.
Do not call a search complete after collecting only supporting material.
Stop only after convergence
Stop when every applicable condition holds:
- Cover every material sub-question or name it as an open gap.
- Trace every key fact to an appropriate primary source when one exists.
- Corroborate every key conclusion with at least two genuinely independent sources.
- Test the strongest credible alternative or counterexample.
- Reconcile meaningful conflicts or state why they remain unresolved.
- Mark each conclusion with a certainty grade.
- Observe diminishing returns: the latest search round changes no conclusion, certainty grade, or material caveat.
Continue when any applicable item remains false. Stop earlier for a trivial or low-impact task when one authoritative source directly answers the whole question and no meaningful conflict, ambiguity, or freshness risk exists. State that reduced scope.
Use Three Levers
| Lever | Make it operational | Avoid |
|---|---|---|
| Thorough | Decompose the question into mutually exclusive, collectively exhaustive sub-questions; track coverage; apply the explicit STOP conditions. | Collecting many links without proving coverage. |
| Deep | Prefer the closest primary source; inspect definitions, mechanisms, boundaries, version history, and counterexamples; ask why the fact holds. | Repeating summaries that merely agree. |
| Correct | Cross-verify material claims; test alternatives; distinguish source independence; grade certainty. | Treating consensus, rank, or confidence as proof. |
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.
- 8d ago First seen · 150 lines · 77 tokens per session scan A 7284776e9860
blcaptain-research-driven is a skill published in the GitHub repository dososo/blcaptain-research-driven (2 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 77 tokens to every session and 1,736 once invoked, about $0.0004 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
karpathy-llm-wiki
Use when building or maintaining a personal LLM-powered knowledge base. Triggers: ingesting sources into a wiki, querying wiki knowledge, linting wiki quality, 'add to wiki', 'what do I know about', or any mention of 'LLM wiki' or 'Karpathy wiki'.
shellgames
Play board games on ShellGames.ai — Chess, Poker, Ludo, Tycoon, Memory, and Spymaster. Use when the agent wants to play games against humans or other AI agents, join tournaments, chat with players, check leaderboards, or manage a ShellGames account. Triggers on "play chess/poker/ludo/memory", "shellgames", "join…
curl-search
Web search using curl + multiple search engines (Baidu, Google, Bing, DuckDuckGo). Activates when user asks to search, look up, or query something online. Includes security enhancements: input sanitization, command injection protection, and URL encoding.
design-review-cycle
Run evidence-backed design, specification, and implementation-plan reviews through the local llm-cli-gateway stdio MCP server. Use for single-provider or complete cross-LLM design review before implementation. Covers all seven CLI request surfaces and Mistral Vibe's current accept-edits default.
model-routing
Select and dispatch llm-cli-gateway providers by live capability, target access, and task requirements. Use when choosing among Claude, Codex, Gemini, Grok, Mistral, Devin, and Cursor, or when selecting an explicit model through the local stdio gateway.
multi-llm-orchestration
Orchestrate the complete llm-cli-gateway provider surface through its local stdio MCP server. Use for parallel implementation, review, session, async-job, cache-aware prompt, and cross-LLM workflows across Claude, Codex, Gemini, Grok, Mistral, Devin, and Cursor.