skills.rest
dshills
dshillsCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

aperture

Select repository files and assess task feasibility before agent execution.

Updated Apr 18, 2026
Search Tags:#planning#feasibility#manifest#deterministic#gate#context-loading#token-budget
One-click install
npx skills add https://github.com/dshills/aperture --skill aperture
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aperture
Source: https://github.com/dshills/aperture/tree/main/examples/claude-code-skill
Command: npx skills add https://github.com/dshills/aperture --skill aperture

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Aperture gates coding tasks by deterministically planning and pruning repository context before an agent runs, producing a reproducible manifest and an explicit feasibility signal to decide if the task should proceed.

Core Features & Use Cases

  • Deterministic planning: selects the files and context the agent should load to fit token budgets, reducing noise and waste.
  • Feasibility scoring and gap detection: evaluates anchors, specs, tests, and config, surfacing blocking gaps with actionable remediation guidance.
  • Explainability and reproducibility: outputs a traceable decision tree and a manifest hash so runs are auditable.
  • Pre-agent gating across toolchains: integrates with various code assistants to guard the start of any coding task.

Quick Start

Describe a repo task and run aperture plan TASK.md to gate the coding task before execution.

Frequently Asked Questions about aperture

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I load repository context deterministically before an AI agent starts coding?▼

Deterministic context loading selects specific repository files to load based on token budgets, reducing noise and waste before agent execution. It identifies task anchors, specs, tests, and config to ensure safe gating.

What is pre-agent gating for coding tasks and when do I need it?▼

Pre-agent gating assesses task feasibility and prunes repository context before an agent runs, producing a reproducible manifest. It is needed for non-trivial repo task kickoffs, plans, or feasibility checks requiring budget headroom and essential inputs.

How do I check coding task feasibility and detect missing specs or tests?▼

Feasibility scoring evaluates anchors, specs, tests, and config to surface blocking gaps with actionable remediation guidance. It detects missing spec, test, or config files to ensure safe gating before agent execution.

Can I gate coding tasks across different code assistant toolchains?▼

Pre-agent gating integrates with various code assistants to guard the start of any coding task. It works across toolchains by deterministically planning and pruning repository context before execution.

How do I make AI coding runs reproducible and auditable?▼

Reproducible runs output a traceable decision tree and a manifest hash so planning decisions are auditable. This ensures deterministic file selection and context loading can be verified and repeated.

Why does my AI coding agent run out of token budget before finishing?▼

Token budget exhaustion happens when repository context is not pruned before execution. Deterministic planning selects files to fit token budgets, reducing noise and waste to prevent budget headroom issues.

Related Skills

Explore More Software Engineering Skills→
paradox123paradox123

context-bootstrap

Collect repository structure, runnable commands, and architecture anchors at startup.

Community
Intermediate
fhltafhlta

speckit-agent-context-update

Update coding agent context files to reference the latest Spec Kit plan.

Community
Intermediate
doodledooddoodledood
70

verify

Coordinate parallel automated verifiers to validate Manifest criteria.

Community
Advanced
jankneumannjankneumann
4

quick-task

Dispatch ad-hoc coding tasks to a configured vendor CLI.

Community
Intermediate