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 mp-juns/universal-research-mcp --skill research-workflowgit clone --depth 1 https://github.com/mp-juns/universal-research-mcpWrote 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/mp-juns/universal-research-mcp/research-workflow)<a href="https://agentmods.dev/skills/mp-juns/universal-research-mcp/research-workflow"><img src="https://agentmods.dev/badge/skills/mp-juns/universal-research-mcp/research-workflow/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/mp-juns/universal-research-mcp/research-workflow"><img src="https://agentmods.dev/badge/skills/mp-juns/universal-research-mcp/research-workflow.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 | $0.00032 | $0.01029 |
| Opus 5 | $0.00016 | $0.00515 |
| Sonnet 5 | $0.00006 | $0.00206 |
| Haiku 4.5 | $0.00003 | $0.00103 |
Grade A, and why
research-workflow 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 9d 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 — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Universal Research Workflow
Mandatory session scope confirmation
Before research tools or any task execution in EVERY NEW session, ask the user to confirm the current workspace and permission scope, then WAIT for their explicit reply. Apply session-scope.md: host shell; ordinary scoped file creation/editing allowed after confirmation; external network/download approval BEFORE EACH operation; ZERO agents. Ask even if the initial task sounds actionable. Defaults and a previous session's approval never authorize this session. Preserve an explicit same-session confirmation after resume/compaction; ask again if uncertain.
Use this skill for questions about previous research work, plans, protocols, decisions, results, failures, or artifacts managed by this project.
If research_profile_status reports a profile, use its declared retrieval mode
only after the corresponding local derived index is current. The profile can
declare code, build-definition, configuration, and documentation source scope,
but it does not register files or make them evidence-eligible. Never treat a
profile's provider route or registered Skill ID as permission to call an API,
read a secret, create a Skill, or start a subagent.
Evidence-first retrieval
- Call
memory_search_candidatesto obtain candidate records. - Treat every returned score, preview, and summary as candidate metadata.
- For any material claim, call
memory_fetch_evidencewith the returnedevent_id, path, line range, andsource_sha256asexpected_sha256. - Before reporting a material result, comparison, causal statement, release
decision, or other load-bearing fact, pass each evidence fetch response's
claim_gate_referenceobject unchanged tomemory_check_evidence_eligibility. Usemateriality="material"for a load-bearing factual statement. Do not substitute the fetch response's currentsha256forexpected_sha256. - Treat
eligibleonly as integrity/range/count eligibility. It does not establish semantic support, resolve conflicts, or prove the source true. - Do not state a claim when eligibility is
blocked. When it is eligible, review relevance and conflicts before stating a bounded conclusion. - State the verified path, line range, and uncertainty with the conclusion.
- If evidence conflicts, preserve the conflict rather than merging it.
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.
- 9d ago First seen · 96 lines · 32 tokens per session scan A 82c571574879
research-workflow is a skill published in the GitHub repository mp-juns/universal-research-mcp (0 stars, last pushed 11d ago), licensed MIT. It adds 32 tokens to every session and 1,029 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
neotoma-agent
Canonical turn protocol for agents writing to Neotoma. Bounded retrieval on the user message, store user message PARTOF conversation with REFERSTO edges to retrieved entities, store assistant reply the same way. Use this skill in any agent loop that should persist conversation state to Neotoma.
ensure-neotoma
Install Neotoma, configure MCP for your harness, and verify connectivity. Prerequisite for all memory skills.
claude-context-improver
Audit and improve the Claude Code context layer — CLAUDE.md guidance files, .claude/rules/ path-scoped rules, and companion codemap.md navigation maps — against Claude 5 context-engineering rules (judgement over rules, progressive disclosure, no cross-layer conflicts). Asks whether to optimize the current repository…
store-neotoma
Skill "store-neotoma" from markmhendrickson/neotoma, covering store-neotoma, purpose, scope, prerequisites and phase 0: preview (mandatory — execute first).
memory-organization
Audit and reorganize every persistent store — semanticmodels, metrics, referencesql, knowledge, memory, AGENTS.md, skills — verifying each item sits in the correct store per storage-classify, and surfacing duplicates, misclassifications, conflicts, and stale/erroneous entries. Produce a Remediation Plan, STOP for…
end
Session-end audit. Surfaces remaining work to complete or track, verifies all data intended for Neotoma storage from this session is actually stored, and offers to file what's missing. Invoke at the natural close of a working session, before context is lost.