Verified:GitHub

mission-spike

Run exploratory or implementation spike experiments to resolve mission uncertainty.

Updated May 22, 2026

Search Tags:#debugging #autonomous agents #coding #experimentation #evidence gathering #stateful workflow #implementation spike

Ready to run

npx skills add https://github.com/MathieuDoyon/mission --skill mission-spike

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: mission-spike
Source: https://github.com/MathieuDoyon/mission/tree/main/.claude/skills/mission-spike
Command: npx skills add https://github.com/MathieuDoyon/mission --skill mission-spike

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It resolves unknowns by running a focused experiment so the mission can plan and implement with evidence instead of speculation.

Core Features & Use Cases

Exploratory spikes for cheap learning (throwaway scripts and probes) that answer questions about APIs, data shapes, performance, or behavior.
Implementation spikes that draft rough real code in the source tree and tag it with a MISSION-SPIKE marker for later promotion.
Structured mission-report output capturing what was learned, artifacts, commit status (when applicable), and a promote checklist.

Quick Start

Ask the orchestrator to resolve one clear unknown by invoking mission-spike with the question, and direct it to run either an exploratory probe or an implementation spike depending on which is cheaper to confirm.

Frequently Asked Questions about mission-spike

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

FAQPage Schema

How do I resolve unknown API assumptions before writing production code?▼

Run an exploratory spike to resolve unknown API assumptions by probing data shapes and behavior with throwaway scripts. This generates concrete evidence captured in a mission-report, allowing you to proceed with implementation based on facts rather than speculation.

What is the best way to validate technical feasibility during a planning plateau?▼

Validating technical feasibility during a planning plateau involves running a focused spike experiment. By selecting exploratory or implementation mode, you test the smallest possible assumption, outputting a structured mission-report with artifacts and a promote checklist to guide routing decisions.

How do I run an implementation spike without polluting my main source tree?▼

Run an implementation spike to draft rough code directly in the source tree while tagging it with a MISSION-SPIKE marker. This isolates experimental code for later promotion and emits a mission-report detailing commit status and a checklist for integrating valid findings.

When should I use an exploratory probe versus an implementation spike?▼

Use an exploratory probe when you need cheap learning about APIs or performance using throwaway scripts. Choose an implementation spike when you need to draft rough code in the source tree to validate structural assumptions and generate a promotion checklist.

Why do my project plans repeatedly fail when facing unknown technical requirements?▼

Project plans repeatedly fail when facing unknown technical requirements because planning relies on speculation. Running an evidence-gathering spike experiment generates facts about data shapes and behavior, producing a mission-report that breaks the cycle of repeated plan failures.

What do I need to set up before running an autonomous agent spike experiment?▼

Before running an autonomous agent spike experiment, you must select either exploratory or implementation mode and set up the workspace. This preparation allows the agent to run the smallest experiment needed to generate evidence and emit a structured mission-report.