Guide · Agent skill
How to use the Iterative Depth Explorer agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
IterativeDepth is an agent skill from the LifeOS collection that makes an AI agent examine one problem several times, each pass through a different analytical lens, to surface requirements and edge cases a single look would miss. It suits developers and product owners who want a structured second look at a design before building it. By the end of this guide you will know what the skill does, how to install it, how to trigger it with natural requests, and what to check before trusting its output.
Iterative Depth Explorerdanielmiessler/LifeOS · MIT Download the skill Quick start
- 1Download the ZIP. Get the IterativeDepth ZIP from this page.
- 2Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
- 4Make a natural request. Ask the agent to explore a problem deeper, run a blind spot check, or tell you what you are missing.
- 5Review the criteria. Read the returned criteria and anti-criteria, and check which lenses produced them.
- 6Verify before acting. Confirm any factual claims independently and decide which criteria to adopt.
What is an agent skill, and what does IterativeDepth do?
An agent skill is a folder of instructions that a SKILL.md-compatible agent runtime, such as Claude Code, loads so the agent knows how to handle a certain kind of task. IterativeDepth is one such skill. Its SKILL.md describes an Explore intent that runs between two and eight sequential passes over the same problem.
Each pass uses a different lens. The skill's lens catalog includes angles such as the literal request, a stakeholder's view, how something could fail, and how it changes over time. Each pass is meant to produce new ISC criteria, which are binary-testable statements of what a good result looks like, plus anti-criteria. The results are merged and deduplicated across lenses, and the reasoning for each angle is shown.
- Runs 2 to 8 passes over the same problem, each through a different lens.
- Returns criteria and anti-criteria that a single pass would likely miss.
- Names the lenses used and any cross-angle finding that appeared only when two lenses collided.
- The method draws on 20 techniques from cognitive science, requirements engineering and design thinking, documented in ScientificFoundation.md.
Install the IterativeDepth skill
Download the IterativeDepth ZIP from this page. Unzip it into one of two locations: your personal skills folder at ~/.claude/skills/, or a project-specific .claude/skills/ folder inside the project you are working on. A personal install makes the skill available everywhere; a project install keeps it with that repository.
If you use the Claude apps instead of Claude Code, you can upload the skill there under Settings → Capabilities → Skills rather than unzipping it to a folder.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Check the skill's files after installing
The skill folder contains SKILL.md, plus Intents/Explore.md, TheLenses.md and ScientificFoundation.md. SKILL.md is the entry point that the agent reads first; the other files are read when the agent needs them, for example to choose lenses or to explain the method.
If you want to personalise the skill, the source describes an optional user customisation folder at <configRoot>/LIFEOS/USER/CUSTOMIZATIONS/SKILLS/IterativeDepth/, where configRoot is $CLAUDE_CONFIG_DIR when that variable is set, otherwise ~/.claude. Files there such as PREFERENCES.md, configurations and resources override the defaults. If the folder does not exist, the defaults apply.
- SKILL.md: entry point and trigger rules
- Intents/Explore.md: the Explore workflow and output requirements
- TheLenses.md: the catalog of eight exploration angles
- ScientificFoundation.md: the 20 techniques the method rests on
Trigger the skill with natural requests
You do not need a special command. The skill's description lists phrases that route to it: iterative depth, explore deeper, multi-angle analysis, surface hidden requirements, blind spot check, and what am I missing. Ask for one of these in plain language and the agent should pick the skill up.
Phrase your request around the problem and the goal. Naming the decision or design you want examined helps the agent choose lenses that fit it. The source notes that a security problem leans on failure and adversary angles, while a UX problem leans on the experiential one.
- "Explore this feature design deeper before I build it."
- "Do a blind spot check on this migration plan."
- "What am I missing in this checkout flow?"
- "Run a multi-angle analysis of the private versus public issue-routing decision."
Read the output and decide what to do with it
A good run returns criteria and anti-criteria that are each testable as a yes or no, deduplicated across lenses, with refinements to existing criteria reported along with what changed and why. Treat these as candidate requirements for your own review, not as settled facts.
The skill's own falsifier is that a reviewer finds the passes repeat one lens, miss a material edge case, or produce unsupported criteria. Check for those problems yourself before acting on the list.
- Look for criteria that are new compared with what you already had.
- Check that different lenses are actually different angles.
- Confirm each criterion is tied to the problem, not generic advice.
- Verify any factual claim independently, since lenses are analytical prompts and not sources of evidence.
What people use it for
- Reviewing a feature design before development starts to find edge cases the team has not discussed.
- Checking a routing decision, such as private versus public issue handling, across different outcomes.
- Auditing an existing set of requirements for missing failure cases and refined criteria.
- Looking at a decision from a stakeholder's view and from how it could fail, with the reasoning shown for each angle.
Frequently asked questions
How do I install the IterativeDepth skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
What does the IterativeDepth skill require?
It requires a SKILL.md-compatible agent runtime such as Claude Code. An optional LIFEOS user customisation folder under the Claude config directory can hold personal overrides. The source says it runs no scripts and uses no network.
Is IterativeDepth free?
The source lists the licence as MIT. It does not state pricing, so check the directory page and the repository for any terms beyond the licence.
How is IterativeDepth different from ApertureOscillation?
IterativeDepth looks for hidden requirements and edge cases by viewing one problem through different lenses. The skill description says scope and zoom analysis belongs to the separate ApertureOscillation skill, so use that one for tactical or strategic zoom.
Why does the skill not find the right answer every time?
The lenses are analytical prompts, not a source of verified evidence. Outputs can include unsupported criteria or miss a material edge case, so they need human review.
When should I not use IterativeDepth?
Skip it for simple, well-defined questions, where extra passes add cost without new insight. Use ApertureOscillation for scope or zoom questions.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use IterativeDepth if you are a developer or product owner who wants a structured blind-spot check on a design or decision before building, and you are willing to review its output. Pick something else for simple questions, for scope or zoom analysis (use ApertureOscillation), or for any task that needs verified facts.
Download the skill