Ant Stanley
Community@antstanley · Durban
Ant Stanley's skill registry delivers repository security scanning, spec-driven implementation planning, and semi-formal correctness verification for agentic coding environments.
Agent Skills by Ant Stanley
Showing 26 vetted skills indexed across 3 GitHub repositories.
antstanleyspec-builder
Executes a spec-planner plan by dispatching one sub-agent per task behind correctness and completeness gates.
antstanleyvalidate-done-certificate
Validates task implementations against done certificates to derive DONE, PARTIAL, or NOT_DONE verdicts.
antstanleysemi-formal-review
Reviews implemented code changes with a semi-formal certificate producing a derived correctness verdict.
antstanleyreasoning-semiformally
Applies structured certificate templates to verify patches, localize faults, and compare fixes.
antstanleythreat-model
Generates repository-scoped security threat models covering trust boundaries, attack surfaces, and severity calibration.
antstanleytrack-findings
Track validated security scan findings as Linear, Jira, or GitHub issues and advisories.
antstanleyvulnerability-writeup
Write source-backed vulnerability disclosure reports with PoCs from notes, findings, and scan output.
antstanleysecurity-diff-scan
Reviews Git diffs, commits, and pull requests for security regressions through phased scans.
antstanleyfinding-discovery
Identifies plausible security vulnerability candidates in code diffs and repositories.
antstanleytriage-finding
Triage existing security findings against a repository using static code evidence.
antstanleyattack-path-analysis
Traces validated security findings from source to sink and calibrates severity against a threat model.
antstanleyfix-finding
Implements and verifies minimal patches for validated security findings.
antstanleysecurity-scan
Audits every file in a repository or scoped path and produces a sealed scan contract with findings, coverage, report, and SARIF.
antstanleydefine-security-policy
Define, review, and update SECURITY.md policy guidance for repositories and components.
antstanleydeep-security-scan
Runs repeated multi-pass security discovery over a repository until candidate findings saturate.
antstanleypropose-security-hardening
Develop evidence-backed architectural security hardening proposals from vulnerability findings and source code.
antstanleyvalidation
Validates candidate security findings using PoCs, debugger traces, and static code analysis.
antstanleyspec-planner
Decompose a specification into a dependency-ordered graph of reviewable task packages with definitions of done.
antstanleydone-certificates
Authors task-specific semi-formal done certificates that a separate validating agent discharges.
antstanleydesign-guidelines
Create canonical design guidelines and acceptance criteria for a project's visual and interaction design.
antstanleydevelopment-guidelines
Generate a development-guidelines spec page defining coding standards, toolchain, and testing rules for a repository.
antstanleyspec-reviewer
Review canonical and change specs against code using semi-formal certificate templates.
antstanleyspec-creator
Create and maintain canonical design specifications as numbered markdown with JSON Schema sidecars.
antstanleyusing-jj-workspaces
Creates isolated jj workspaces as sibling directories for parallel sub-agent development.
Frequently Asked Questions About Ant Stanley
FAQPage SchemaWhat tasks can I accomplish with Ant Stanley's security skills?▼
You can run standard or deep multi-pass repository security scans, review pull request and commit diffs, discover and validate findings, trace attack paths from source to sink, triage imported scanner tickets, fix confirmed vulnerabilities, write disclosure reports, and propose architectural hardening with evidence-backed tradeoff analysis.
Who should use the spec-builder and spec-planner skills?▼
Engineers running agentic coding harnesses who want structured implementation. Spec-planner decomposes a specification into a dependency-ordered task graph with definitions of done; spec-builder dispatches one sub-agent per task in isolated jj or git workspaces, gating each through semi-formal review and done-certificate validation before merging.
What are the runtime requirements for spec-builder and jj workspaces?▼
Spec-builder needs a harness that dispatches sub-agents: Codex with agent dispatch, Claude Code or OpenCode core Task, or Pi with a subagents extension; otherwise it falls back to sequential single-agent execution. The using-jj-workspaces skill requires jj (jujutsu) installed and intercepts git-worktree requests in jj-managed repositories.
How does the semi-formal verification pipeline work?▼
Done-certificates authors a per-task verification protocol naming evidence and regression checks. After implementation, semi-formal-review issues a CORRECT to BUGGY verdict, then validate-done-certificate discharges each obligation to produce a DONE, PARTIAL, or NOT_DONE verdict. Reasoning-semiformally covers standalone patch verification and fault localization.
Can these skills file security findings into issue trackers?▼
Yes. The track-findings skill pushes validated findings from a sealed scan bundle into Linear, Jira, GitHub issues, or draft GitHub security advisories, handling up to 25 findings per batch with duplicate checks, exact previews, approval-gated writes, and readback confirmation. Ad-hoc tickets without a sealed scan are filed directly.