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 Roshu18/bearstrike-ai --skill research-notesgit clone --depth 1 https://github.com/Roshu18/bearstrike-aiWrote 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/roshu18/bearstrike-ai/research-notes)<a href="https://agentmods.dev/skills/roshu18/bearstrike-ai/research-notes"><img src="https://agentmods.dev/badge/skills/roshu18/bearstrike-ai/research-notes/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/roshu18/bearstrike-ai/research-notes"><img src="https://agentmods.dev/badge/skills/roshu18/bearstrike-ai/research-notes.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 | $0.00020 | $0.00506 |
| Opus 5.5 | $0.00008 | $0.00202 |
| Sonnet 5.5 | $0.00004 | $0.00101 |
| Haiku 4.5 | $0.00002 | $0.00051 |
Grade A, and why
research-notes 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 1mo 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.
research-notes
Use this skill to guide planning and triage decisions before running tools.
Last-30-day high-signal patterns
- Access-control flaws remain top impact class (IDOR/BOLA, role boundary bypass).
- API auth/session confusion appears in GraphQL and invite/workspace workflows.
- SSRF remains high-impact where URL fetch parameters exist (swaggerUrl-style handlers).
- Misconfiguration leaks (Prometheus /metrics, debug/internal endpoints) provide chainable recon intel.
- Upload handling gaps (SVG script payloads) continue to produce stored XSS.
Frequent endpoint and parameter patterns
- Endpoint families: /api/v1/, /api/auth, /api/user, /internal, /graphql, /metrics, /verify-otp.
- Sensitive parameters: user_id, account_id, id, redirect_url, swaggerUrl.
- Trust-boundary headers: custom session blobs, X-Original-URL, X-Forwarded-*.
Operational guidance for MCP AI
- Start with low-noise recon and endpoint intelligence.
- Prioritize account-boundary tests before aggressive scanning.
- Keep one active hypothesis at a time until impact is confirmed.
- Use async jobs for expensive tools to avoid MCP timeout loops.
- Prefer context-aware payload sets (top 10 per class), not brute-force payload dumps.
Priority bug classes
- IDOR/BOLA and role boundary bypass
- auth/session handling flaws and OTP-flow bypass
- API overexposure and mass assignment patterns
- SSRF/open redirect request-routing issues
- high-confidence misconfiguration exposures
Recommended triage sequence
- Map live hosts and key endpoints.
- Score endpoints and test high-priority routes first.
- Execute access-control and auth boundary checks.
- Run constrained vulnerability templates and targeted payloads.
- Escalate to deeper scans only when signals are strong.
Public references
- OWASP Top 10: https://owasp.org/Top10/
- OWASP API Security: https://owasp.org/API-Security/
- PortSwigger Web Security Academy: https://portswigger.net/web-security
- HackerOne resources: https://www.hackerone.com/
- Bugcrowd resources: https://www.bugcrowd.com/resources/
- Intigriti research blog: https://www.intigriti.com/researchers/blog
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.
- 1mo ago First seen · 56 lines · 20 tokens per session scan A 5cd9fdf97eb8
research-notes is a skill published in the GitHub repository Roshu18/bearstrike-ai (8 stars, last pushed 6mo ago), licensed MIT. It adds 20 tokens to every session and 506 once invoked, about $0.0001 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 2026-08-31.
Other skills, from other repositories
gentech-blockrun
GenTech Labs' integration patterns for BlockRun MCP from Hermes Agent. Covers daily usage patterns, cost-optimized workflows, multi-tool pipelines, and reliable error handling for BlockRun's full toolset.
portainer-mcp-hygiene
How to drive the Portainer MCP server's tools correctly — both reading and mutating. Reading: project responses with select (JMESPath), where the heavy fields live (snapshots, status blocks, managed fields), how to handle non-JSON Docker/K8s proxy endpoints (container and pod logs, stats, exec), and how to interpret…
mcptoon
Compress MCP tool discovery with the mcptoon CLI. Trigger when a session has a large MCP tool catalog (many servers/tools), when the user mentions token cost, tool discovery, mcptoon, or asks to list/call MCP tools efficiently. Also route here when the user says the MCP tool list is too large, the agent context window…
kicad-library
Library management workflow for KiCAD — creating symbols, footprints, and managing libraries via MCP tools. Triggers on: "create a symbol", "make a footprint", "custom component", "register library", "find a part", "pin numbering", "new symbol", "new footprint", "add to library", "library path", "pad layout".
kicad-pcb
Workflow skill for KiCAD PCB layout and routing via MCP tools. Triggers on: "layout the board", "route traces", "PCB", "place footprints", "copper pour", "board outline", "differential pair", "board setup", "track width", "via", "zone", "design rules", "stackup", "silkscreen".
kicad-schematic
Workflow skill for KiCAD schematic design via MCP tools. Triggers on: "design a circuit", "add a component", "wire up", "connect pins", "build schematic", "place resistor", "place cap", "place IC", "schematic", "add symbol", "net label", "power rail".