Claude Code Agent Development
anthropics/claude-plugins-official
Explains how to write agents for Claude Code plugins: the markdown file with YAML frontmatter, trigger descriptions, model and color settings, and system prompt design.
Protocol for iteratively refining sub-agent queries through follow-up questions to ensure sufficient context
Install Iterative Retrieval
$ npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install closedloop-ai/claude-plugins iterative-retrieval --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/closedloop-ai/claude-plugins.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/code/skills/iterative-retrieval .claude/skills/iterative-retrieval && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "iterative-retrieval" agent skill from https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrieval into .claude/skills/iterative-retrieval/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "iterative-retrieval", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrievalType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install closedloop-ai/claude-plugins iterative-retrieval --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/closedloop-ai/claude-plugins.git skills-src && mkdir -p .agents/skills && cp -r skills-src/plugins/code/skills/iterative-retrieval .agents/skills/iterative-retrieval && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "iterative-retrieval" agent skill from https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrieval into .agents/skills/iterative-retrieval/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "iterative-retrieval", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install closedloop-ai/claude-plugins iterative-retrieval --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/closedloop-ai/claude-plugins.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/plugins/code/skills/iterative-retrieval .cursor/skills/iterative-retrieval && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "iterative-retrieval" agent skill from https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrieval into .cursor/skills/iterative-retrieval/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "iterative-retrieval", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/closedloop-ai/claude-plugins.git --path plugins/code/skills/iterative-retrieval--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install closedloop-ai/claude-plugins iterative-retrieval --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/closedloop-ai/claude-plugins.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/plugins/code/skills/iterative-retrieval .gemini/skills/iterative-retrieval && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "iterative-retrieval" agent skill from https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrieval into .gemini/skills/iterative-retrieval/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "iterative-retrieval", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install closedloop-ai/claude-plugins iterative-retrievalInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/closedloop-ai/claude-plugins.git skills-src && mkdir -p .github/skills && cp -r skills-src/plugins/code/skills/iterative-retrieval .github/skills/iterative-retrieval && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "iterative-retrieval" agent skill from https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrieval into .github/skills/iterative-retrieval/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "iterative-retrieval", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install closedloop-ai/claude-plugins iterative-retrieval --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/closedloop-ai/claude-plugins.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/plugins/code/skills/iterative-retrieval .opencode/skills/iterative-retrieval && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "iterative-retrieval" agent skill from https://github.com/closedloop-ai/claude-plugins/tree/main/plugins/code/skills/iterative-retrieval into .opencode/skills/iterative-retrieval/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "iterative-retrieval", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
iterative-retrieval- GitHub stars
- 122
- Token cost
- ~1.6k tokens
- SKILL.md length
- 801 words
- Files
- 2 (incl. references)
- Skills in repo
- 41
- Repo updated
- First seen
- Licence
- Apache-2.0
At a glance
Protocol for iteratively refining sub-agent queries through follow-up questions to ensure sufficient context
- Works in 4 steps: Initial Dispatch → Sufficiency Evaluation → Refinement Request → …
- Tasks that involve Subagents
- SKILL.md covers Target Audience, Usage Model, The 4-Phase Protocol and Refinement Cycle Guidance, plus 3 more sections
- Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
What it does
Iterative Retrieval is an agent skill from closedloop-ai/claude-plugins. Protocol for iteratively refining sub-agent queries through follow-up questions to ensure sufficient context
Its SKILL.md is about 1.6k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files, including reference files (for example `references/examples.md`).
It sits in Agent Workflows, covering Subagents. The repository describes itself as: Open-source Claude Code plugins for multi-agent software delivery. Plan-first SDLC workflow, code review, LLM quality judges, and self-learning — grounded in your codebase… The licence is Apache-2.0.
When your agent uses it
- Tasks that involve Subagents
Example prompts
- “/iterative-retrieval”
Workflow steps
4 steps, taken from the step headings in SKILL.md.
What it can do on your machine
Read from SKILL.md and the folder at commit 476b54c. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
No scripts in the folder and no shell commands in SKILL.md.
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Iterative Retrieval loads about 1.6k tokens when it runs, and up to ~4.7k if it reads all its reference files. Until then it costs about 32 tokens; SKILL.md has 801 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from closedloop-ai/claude-plugins at commit 476b54c, republished under its Apache-2.0 licence (© closedloop-ai). 801 words, ~1,576 tokens.
Download SKILL.mdSave it as
.claude/skills/iterative-retrieval/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.- name
- iterative-retrieval
- description
- Protocol for iteratively refining sub-agent queries through follow-up questions to ensure sufficient context
Iterative Retrieval Skill
This skill enables orchestrators to iteratively refine sub-agent queries through follow-up questions, ensuring sub-agents gather sufficient context before the orchestrator accepts their output. This addresses the problem where orchestrators have semantic context that sub-agents lack, leading to incomplete summaries.
Target Audience
This skill is designed for the /code orchestrator only (code.md slash command). Sub-agents do not use this skill - they simply respond to queries. The orchestrator is responsible for evaluating responses and deciding whether to resume with follow-ups.
Note: Sub-agents do NOT see this skill documentation in their context. Only the orchestrator can invoke and follow the protocol. Sub-agents are unaware they're part of an iterative retrieval loop - they just respond to queries and follow-ups as normal requests.
Usage Model
This skill is optional and opt-in. Not every sub-agent call benefits from iterative refinement - simple lookups or well-defined queries don't need it. Invoke this skill when you anticipate that a sub-agent may return incomplete context due to semantic gaps.
The 4-Phase Protocol
This protocol provides a structured approach to iterative context gathering. All 4 phases represent the recommended workflow, but phases 2-4 are optional if the initial response is sufficient. Exercise judgment - if Phase 2 evaluation shows context is sufficient on first pass, phases 3-4 aren't needed.
Phase 1: Initial Dispatch
Define and dispatch the initial query with full context:
- Define PRIMARY OBJECTIVE: Clearly state what you ultimately need to accomplish
- Formulate INITIAL QUERIES: Specific questions or search criteria for the sub-agent
- Dispatch with BOTH: Send both the queries AND the primary objective to provide semantic context
- Store AGENT ID: Keep the
agent_idreturned from theTask(...)call for potential continuation via SendMessage
Phase 2: Sufficiency Evaluation
Evaluate whether the sub-agent's response provides sufficient context using the checklist below. If context is sufficient, skip to output (phases 3-4 not needed). If gaps exist, proceed to Phase 3.
Sufficiency Evaluation Checklist
Ask yourself these 4 questions:
- Direct Answer: Does the summary directly answer the orchestrator's primary objective?
- Obvious Gaps: Are there obvious gaps the orchestrator could identify? (e.g., missing error handling, incomplete data flow, unaddressed edge cases)
- Adjacent Information: Did the sub-agent mention adjacent or related information that wasn't fully explored?
- Confidence Check: Would the orchestrator be confident proceeding with ONLY this information?
If you answer "no" to question 1 or 4, OR "yes" to questions 2 or 3, the context is likely insufficient - proceed to Phase 3.
Show full SKILL.md (393 more words)Show less
Phase 3: Refinement Request
Continue the sub-agent with targeted follow-up questions using SendMessage:
- Continue via SendMessage: Use
SendMessage(to=<stored_agent_id>, summary=<5-10 word summary>, message=<prompt>)to continue the subagent; completed subagents auto-resume from transcript in the background with full prior context intact - Acknowledge what was useful: Briefly confirm what information was valuable
- Specify EXACTLY what's needed: Be precise about what additional context is required
- Ask about related info: "Is there related information that might be relevant to [primary objective]?"
Async flow: SendMessage returns immediately with a queued acknowledgment. The subagent then runs in the background and you will receive a <task-notification> when it finishes. Do not proceed to the next step until that notification arrives.
Fallback: If no agent_id is in memory (cross-session resume after a previous Claude Code session ended) or SendMessage returns an actual error, fall back to a fresh Task(...) launch with a self-contained prompt that includes all prior context needed.
Phase 4: Loop
Repeat Phases 2-3 until one of these conditions is met:
- Context is sufficient: All 4 checklist criteria are satisfied
- Maximum cycles reached: Recommended maximum of 3 refinement cycles (see guidance below)
- Source exhausted: Sub-agent confirms no additional relevant information exists
Refinement Cycle Guidance
The recommended maximum is 3 refinement cycles (initial dispatch + 2 follow-ups). This balances thoroughness against cost and latency.
Important: This is a recommendation, not an enforced limit. As guidance documentation, this skill cannot enforce limits - the orchestrator has final judgment. However, exceeding 3 cycles often indicates:
- The query scope is too broad
- The sub-agent lacks access to needed information
- The primary objective should be broken into smaller queries
Output Format
When iterative retrieval completes, report:
- Total refinement cycles used: Number of iterations (including initial dispatch)
- What additional context was gathered: Summarize what the follow-ups uncovered
- Agent ID for future continuation via SendMessage: Store for potential later follow-ups
Example:
Iterative Retrieval Summary:
- Cycles used: 2 (initial + 1 follow-up)
- Additional context gathered: Error handling patterns, retry logic implementation, timeout configuration
- Agent ID: agent-abc123 (available for future continuation via SendMessage)When to Use This Skill
Use iterative retrieval when:
- The orchestrator has semantic context the sub-agent lacks
- The query involves complex or interconnected information
- Initial results might miss important adjacent context
- The orchestrator can identify potential gaps in advance
Don't use iterative retrieval when:
- The query is simple and well-defined (e.g., "read file X")
- The sub-agent has all needed context upfront
- Time/cost constraints are critical
- The information needed is atomic and isolated
Examples
For detailed usage examples, see references/examples.md (if available).
© closedloop-ai, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 1 other file (references) in plugins/code/skills/iterative-retrieval of closedloop-ai/claude-plugins.
- SKILL.md
- references/examples.md
Open the folder on GitHubat commit 476b54c
Compare with similar skills
Iterative Retrieval next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Iterative Retrieval this skillclosedloop-ai/claude-plugins | 122 | — | ~1.6k | Automated safety check: Pass | Apache-2.0 | |
| Claude Code Agent Developmentanthropics/claude-plugins-official | 38k | 8 repos | ~2.8k | Automated safety check: Pass | Apache-2.0 | |
| Subagent Driven DevelopmentAsvarox/allkaraoke | 261 | 38 repos | ~1.2k | Automated safety check: Pass | None | |
| Dispatching Parallel Agentsultralisp/ultralisp | 258 | 41 repos | ~1.5k | Automated safety check: Pass | None | |
| Paseo Advisor Second Opiniongetpaseo/paseo | 20k | 1 repos | ~756 | Automated safety check: Pass | Custom licence | |
| Task Observerrebelytics/one-skill-to-rule-them-all | 3.2k | 1 repos | ~12k | Automated safety check: Pass | CC-BY-4.0 |
Similar skills
- Official
Subagent Driven Development
Asvarox/allkaraoke
A skill your agent uses when executing implementation plans with independent tasks in the current session
Dispatching Parallel Agents
ultralisp/ultralisp
A skill your agent uses when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
Paseo Advisor Second Opinion
getpaseo/paseo
Launches one separate agent through Paseo to give a second opinion on the current task, with a self-contained briefing and no permission to edit files.
Task Observer
rebelytics/one-skill-to-rule-them-all
Monitors task execution for skill improvement opportunities.
O2 Review Loop
openobserve/openobserve
Splits a change into planner, coder and independent reviewer roles: you confirm a spec, a subagent implements it, and a separate reviewer checks each round's local WIP commit.
More from closedloop-ai/claude-plugins
All 41 skills in this repoCodex Review
closedloop-ai/claude-plugins
Run Codex to review a plan file and return structured feedback with a verdict.
Critic Cache
closedloop-ai/claude-plugins
Check if critic reviews are still valid before re-running Phase 2.5 critics.
Cross Repo Cache
closedloop-ai/claude-plugins
Check if cross-repo coordinator results can be reused, avoiding redundant Sonnet agent launches.
Eval Cache
closedloop-ai/claude-plugins
Check for a cached plan-evaluation.json result before launching the plan-evaluator agent.
Find Plugin File
closedloop-ai/claude-plugins
This skill should be used when needing to locate files within the Claude Code plugins cache directory (~/.claude/plugins/cache).
Gh Monitor PR
closedloop-ai/claude-plugins
Start a detached GitHub pull-request monitor that wakes the exact launching Codex Desktop or CLI root through the managed Codex App Server when review, CI, conflict, merge-queue, closure, readiness…
Related
Categories
Questions about Iterative Retrieval
What does Iterative Retrieval do?
Protocol for iteratively refining sub-agent queries through follow-up questions to ensure sufficient context. Iterative Retrieval is an agent skill from closedloop-ai/claude-plugins.
When should I use Iterative Retrieval?
Iterative Retrieval fits situations like: tasks that involve Subagents.
How do I install Iterative Retrieval in Claude Code?
Run `npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a claude-code`. Or copy the skill folder (plugins/code/skills/iterative-retrieval in closedloop-ai/claude-plugins) into .claude/skills/iterative-retrieval in your project. Claude Code loads it when a task matches its description.
How do I install Iterative Retrieval in Codex?
Run `npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a codex`. Or copy the skill folder (plugins/code/skills/iterative-retrieval in closedloop-ai/claude-plugins) into .agents/skills/iterative-retrieval in your project. Codex loads it when a task matches its description.
Can I use Iterative Retrieval in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add closedloop-ai/claude-plugins --skill iterative-retrieval -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/iterative-retrieval, .gemini/skills/iterative-retrieval, .github/skills/iterative-retrieval and .opencode/skills/iterative-retrieval in your project.
What does Iterative Retrieval need to run?
SKILL.md names no scripts, command-line tools or credentials: Iterative Retrieval is instructions for the agent only.
Does Iterative Retrieval access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Iterative Retrieval safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Iterative Retrieval use?
Iterative Retrieval is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Iterative Retrieval use?
About 1.6k tokens (SKILL.md is roughly 6.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 3.1k tokens, read only when the agent opens those files.
What are the alternatives to Iterative Retrieval?
Skills that share tags, products or a category with Iterative Retrieval: Claude Code Agent Development (anthropics/claude-plugins-official, 38k stars), Subagent Driven Development (Asvarox/allkaraoke, 261 stars), Dispatching Parallel Agents (ultralisp/ultralisp, 258 stars) and Paseo Advisor Second Opinion (getpaseo/paseo, 20k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Iterative Retrieval?
closedloop-ai (a GitHub organization) maintains it in closedloop-ai/claude-plugins, which has 122 GitHub stars. The repository holds 41 skills in this directory. The repository was last updated on October 7, 2026.
Source: closedloop-ai/claude-plugins on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.