Works with the AI Agents you already use
structured analytic thinking
Weighs competing explanations for an uncertain judgement: hypotheses, key-assumptions check, evidence consistency matrix, diagnosticity, sensitivity, indicators to watch and calibrated confidence wording. Free sample.
Free
structured analytic thinking
Example session with this skill installed
Analyze why our API traffic spiked 400% yesterday. Evidence: E1 (Marketing ran a small ad), E2 (Logs show 80% 404s), E3 (Traffic came from 50 IPs in one region). Evaluate competing hypotheses.
- Read your context and instructions
- Compiled the structured analytic thinking
I have analyzed the API traffic spike using the structured analytic method. The analysis suggests the spike is likely due to a misconfiguration or broken integration rather than a successful marketing campaign or a malicious attack.
1. Question, scope and sources
Question: What caused the 400% API traffic spike on <date>?
Decision: Infrastructure scaling and security response.
Sources: E1 (Marketing report), E2 (System logs), E3 (Network telemetry).
2. Hypotheses
| ID | Hypothesis | Distinguishing claim | Origin |
|---|---|---|---|
| H1 | Marketing Campaign Success | High conversion/200 OK responses | User |
| H2 | Misconfiguration / Broken Client | High error rate from specific sources | Analyst |
| H3 | Distributed Denial of Service (DDoS) | High volume, malicious intent, deceptive headers | Standard |
3. Key assumptions check
| ID | Assumption | Basis | Solidity | If wrong |
|---|---|---|---|---|
| A1 | Log accuracy | System reliability | Solid | Analysis is invalid |
| A2 | Ad reach | E1 | Plausible | H1 becomes less likely |
4. Evidence and consistency matrix
| Item | H1 | H2 | H3 |
|---|---|---|---|
| V1: 80% 404 errors (E2) | II | C | C |
| V2: Traffic from 50 IPs (E3) | I | C | C |
| V3: Small ad spend (E1) | I | N | N |
| Inconsistency count | 4 | 0 | 0 |
5. Diagnosticity and sensitivity
The ranking is
robust. H2 and H3 explain the errors and IP distribution, whereas H1 is highly inconsistent with the 80% failure rate. H2 is ranked above H3 because the low number of IPs (50) is more consistent with a broken integration than a large-scale botnet.
6. Judgement and confidence
It is Probable (60-75%) that the spike was caused by a misconfigured client or broken integration.
Confidence: Moderate.
Reasons: Strong inconsistency in H1; limited diagnostic data to fully separate H2 from a small-scale H3. Any action taken on this judgement is a
Management decision required.
7. Indicators to watch
| ID | Indicator | Favours | Counts against | Where to look | Check by |
|---|---|---|---|---|---|
| IND1 | User agent string patterns | H2 | Randomized strings | Header logs | 24h |
| IND2 | Customer support tickets | H2 | No reports | Zendesk | 24h |
8. Gaps, bias notes and next steps
Information gap — organizational validation required: Need geographic correlation of the 50 IPs to known customer clusters.
This analysis supports, and does not replace, the analyst's professional judgement.
Next steps
- Inspect User-Agent strings for the 50 IPs to identify the client software.
- Cross-reference IP addresses with the marketing campaign target region.
- Check for recent API key issuances or client-side deployments.
Connects securely to your tools. The creator never sees your data.
About this skill
Cognitive bias and anchoring lead to flawed forecasts and missed threats. This skill enforces the Analysis of Competing Hypotheses (ACH) framework to help you weigh conflicting evidence and reach calibrated judgements without locking onto a single narrative.
What it does
- Hypothesis generation creates 3 to 6 mutually exclusive explanations for an event or trend, including a mandatory deception/manipulation hypothesis.
- Key-assumptions check identifies and rates the solidity of the underlying beliefs that support your conclusion.
- Consistency matrix tests every piece of evidence against every hypothesis to find contradictions rather than just confirmations.
- Diagnosticity analysis filters out evidence that supports all hypotheses equally, focusing only on data that actually separates them.
- Sensitivity testing determines if the lead hypothesis is robust or fragile by removing key evidence or assumptions to see if the ranking changes.
- Calibrated wording delivers findings using structured probability bands and confidence levels to avoid vague language.
How it works
- Intake & Scope starts by defining the core question, timeframe, and registering all evidence sources with unique identifiers.
- Matrix Construction builds the evidence-versus-hypothesis table, marking items as consistent, neutral, or inconsistent.
- Analytical Ranking ranks hypotheses by what they fail to explain, not by how much support they have.
- Validation produces a list of indicators to watch and identifies information gaps for future collection.
Frameworks & tools
This skill implements the Richards J. Heuer, Jr. methodology from the CIA's Psychology of Intelligence Analysis. It uses structured Markdown tables for evidence registers and consistency matrices.
Why this beats prompting it yourself
Standard prompts often suffer from confirmation bias, seeking evidence to prove the user's initial guess. This skill forces a "refutation-first" approach, ranking explanations by their inconsistency and explicitly testing for deception and sensitivity.
Use cases
- Threat assessment to determine if a security incident was a targeted attack or an accidental misconfiguration.
- Business forecasting to evaluate why a specific market metric moved and which future scenario is most likely.
- Root cause analysis to investigate complex system failures with incomplete logs.
- Strategic planning to identify "indicators to watch" that signal a shift in a competitor's direction.
Known limitations
Does not compute statistical probabilities from numeric data. It provides an analytical ranking of explanations based on qualitative evidence consistency.
Not for statistical forecasting, legal verdicts, or medical diagnosis. Requires initial evidence to function. Judgment wording follows fixed probability bands.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 5 days ago
- Free with an Agensi account
Permissions
No special permissions declared or detected