Claude Skills by docxology

Harden an analytic workflow against being gamed, anchored, or fed planted evidence.

Detect adversary denial and deception by testing for what a deceiver would hide or plant.

Reason about attribution of an operation with calibrated confidence and alternatives.

Recognize social-engineering elicitation patterns in conversations and outreach.

Design canary tokens and decoys to detect probing and exfiltration (defensive).

Apply the MOM/POP/MOSES/EVE deception-detection checklists to a body of evidence.

Review behavioral and access indicators of insider risk within ethical/legal bounds.

Make every implicit assumption explicit and assess its load-bearing role.

Verify citations exist, say what they're claimed to say, and support the cited point.

Bind every claim to its supporting evidence; flag overclaims and unsupported assertions.

Review code for security defects, silent failures, and unsafe fallback behavior.

Assess dual-use, harm, and misuse potential of a project or capability.

Test an argument's internal consistency and the validity of its inferential steps.

Adversarial-then-constructive review of a project: claims, evidence, risks, gaps, and go/no-go.

Adversarially stress an artifact to find the failure mode its authors did not anticipate.

Assess whether a result can be regenerated from the stated data, code, and seeds.

Critique a study's design for validity threats, confounds, and inferential reach.

Check statistical methods, power, multiple comparisons, and inference against the claims.

Review a system's threat model for missing actors, surfaces, and assumptions.

Distinguish automated from human activity using behavioral and temporal signals.

Detect coordination signatures across accounts, timing, and content reuse.

Analyze how information propagates through a network to find amplifiers and bottlenecks.

Analyze how rival narratives compete for attention, belief, and durability.

Map the competing narratives, carriers, and audiences in an information space.

Assess how a platform's features enable or constrain manipulation.

Monitor an information space for emerging narratives and inflection points.

Document collection, handling, and hashing so evidence integrity is auditable.

Verify a public claim by tracing it to a primary source and corroborating independently.

Plan ethical, scoped collection tied to specific intelligence requirements.

Require independent corroboration before promoting a claim to a finding.

Audit a dataset's origin, licensing, sampling, and integrity before analytic use.

Confirm where imagery was taken using corroborating geographic features (defensive OSINT).

Triage media for reuse, editing, and context-collapse before treating it as evidence.

Use and sanity-check file/post metadata while accounting for stripping and spoofing.

Identify inauthentic personas from behavioral, temporal, and network signals.

Vet a source's identity, track record, motive, and access before relying on it.

Assign and justify confidence using source quality, corroboration, and assumption load.

Produce calibrated probability estimates with explicit reference classes and ranges.

Grade evidence by quality and relevance using an explicit, repeatable rubric.

Synthesize a body of sources into a structured, evidence-graded, gap-aware briefing.

Write findings bottom-line-up-front with traceable evidence and stated confidence.

Build multiple plausible futures around critical uncertainties to stress strategy.

Score evidence by diagnosticity across a full hypothesis set to find the least-disconfirmed explanation.

Cross-tabulate variables to organize evidence and reveal relationships.

Diagram claims, premises, and inferential links to expose logical structure and gaps.

Trace cause-effect chains and feedback loops driving a system's behavior.

Order events temporally to expose gaps, correlations, and causal sequencing.

Assess how each driver influences every other to find leverage and feedback.

Clarify Audience, Issue, Message, and Storyline so the product fits the decision it serves.

Build the strongest principled case against the consensus to test its robustness.