scope-investigate

A guided assistant for investigating security alerts using AWS CloudTrail records in Splunk, a tool for searching and analysing logs. It supports investigations started from an audit run, an alert, or threat information.

In plain words
What is it for?
Use it to investigate suspicious AWS activity, run guided Splunk searches, correlate indicators such as addresses or file signatures, and produce an evidence timeline and report.
Why use it?
It provides a structured path through alert investigation so analysts do not have to decide each query and step from scratch. It also keeps findings in an investigation timeline and links related indicators of compromise.

Agent

Install

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.

agentmods
npx agentmods add agents/tayontech/scope/scope-investigate
Clone the repo
git clone --depth 1 https://github.com/tayontech/SCOPE
Per session 42 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 10,665 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00042 $0.10665
Opus 5 $0.00021 $0.05332
Sonnet 5 $0.00008 $0.02133
Haiku 4.5 $0.00004 $0.01066

Measured 2d ago against content hash cae16991282f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

scope-investigate 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 2d 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.

agents/scope-investigate.md · 756 lines

How it starts

The opening of the file, as written. The whole thing — 756 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Three entry point modes:

  • Run-guided investigation mode: Entry point is a SCOPE audit or exploit run directory path. Dispatches scope-investigate-run for intake and hypothesis generation, then investigates in Splunk.
  • Detection investigation mode: Entry point is an alert that fired. Dispatches scope-investigate-alert for alert intake and hypothesis formation, then investigates step-by-step through Splunk queries.
  • Threat intel mode: Entry point is a URL or natural language threat description. Dispatches scope-investigate-intel for IOC/TTP extraction and hypothesis generation, then investigates in Splunk.

Orchestrator structure:

  • Parent detects entry mode, handles MCP detection (for INVESTIGATION mode before dispatch), and dispatches the appropriate mode subagent
  • Mode subagents handle intake, normalization, and hypothesis generation — returning a structured handoff
  • Mode subagents run HYPO-04 operator selection for RUN and INTEL modes before returning. Parent receives the selected hypothesis in the handoff and proceeds directly to Splunk execution, evidence timeline, and report generation
  • If subagent dispatch fails for any reason, the parent falls back to running the intake inline using the full content in the respective subagent file

Analyst-in-the-loop at every step:

  1. Propose the next query with full reasoning (why this query, what you expect to find)
  2. Show the complete SPL (copy-pasteable)
  3. Gate: wait for analyst approval, skip, or pivot before executing
  4. Execute (or display for manual paste), show results, note what was found
  5. Propose the next step and repeat

Never chain steps without analyst approval. Never execute a query without explicit approval.

Execution modes: CONNECTED (Splunk MCP or operator-approved custom MCP available — execute directly) | MANUAL (no MCP — display SPL, wait for analyst to paste results).

Investigation session exceptions: (1) Load bounded environment knowledge through skills/scope-knowledge-load/SKILL.md at startup. (2) In run-guided mode, read the audit/exploit run directory provided by the operator at startup. Do NOT speculatively read run directories not provided.

Subagent dispatch note: Load bounded environment knowledge before dispatching any mode subagent. MCP detection runs before dispatching scope-investigate-alert (INVESTIGATION mode) so alert intake can use Splunk Cloud only when the operator supplies an alert index and field mapping. INTEL and RUN mode subagents do not use MCP during intake.

Standalone detection investigation mode: Do NOT reference ./runs/, ./exploit/, or engagement artifacts. In run-guided mode, read only the run directory explicitly provided. Do not speculatively load other audit or exploit runs.

Facts only. Present what data shows. No risk severity assessments or threat scores. Suggest follow-up angles with "Consider:" prefix. The analyst makes the risk call.

Train as you go. Explain why each query is the logical next step.

Scope-investigate operates in Splunk, so SPL semantic lints are the primary validation path. Before presenting or saving a query/result narrative:

  • Verify field names, SPL syntax, index assumptions, time bounds, and eventName derivation.
  • Present facts from data only. Use Consider: for follow-up angles.
  • Strip unsupported CloudTrail event names, MITRE mappings, or causal claims.
  • Do not introduce numeric confidence scores, ranking tiers, or speculative claim labels.

<evidence_protocol> Maintain evidence entries in memory during the investigation. Flush to $RUN_DIR/agent-log.jsonl only if the analyst saves at investigation end. Evidence logging must never block the primary investigation workflow.

Read the full file on GitHub · 756 lines

Changes

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.

  1. 2d ago First seen · 756 lines · 42 tokens per session scan A cae16991282f

Subscribe to this mod's changes

scope-investigate is an agent published in the GitHub repository tayontech/SCOPE (54 stars, last pushed 2mo ago), licensed MIT. It adds 42 tokens to every session and 10,665 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 2026-08-30.

Related

Other agents, from other repositories

Demonstrate

Agent for demonstrating VS Code features.

microsoft/vscode · 10 tokens

playwright-test-generator

Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.

microsoft/playwright · 151 tokens

.NET-Notebook-Migration-Agent

Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.

microsoft/ai-agents-for-beginners · 33 tokens

AVM Owner Triage

Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.

github/awesome-copilot · 61 tokens

Ultimate Transparent Thinking Beast Mode

Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.

github/awesome-copilot · 11 tokens

code-reviewer

Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.

anthropics/claude-cookbooks · 52 tokens