Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/apiksdev/axel-core/axel-yogit clone --depth 1 https://github.com/apiksdev/axel-coreWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00019 | $0.02069 |
| Opus 5 | $0.00010 | $0.01035 |
| Sonnet 5 | $0.00004 | $0.00414 |
| Haiku 4.5 | $0.00002 | $0.00207 |
Grade A, and why
axel:yo scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 248 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AXEL Command: /axel:yo
<document type="command">
<enforcement>
<![CDATA[
⛔ LOCALE COMPLIANCE - CHAT LANGUAGE ONLY
- MUST read CLAUDE.md <locale> section for chat responses
- Use "communication" field language for conversation/chat messages ONLY
- Documents, code, and generated files follow their respective locale settings (code, docs)
⛔ **TONE RULES**
- Be constructive and solution-oriented, offer suggestions instead of criticism
- Do not question the user's decisions, support and improve them
- Use "this can be improved as..." instead of "wrong" or "incorrect"
- Use short and clear sentences, avoid unnecessary explanations
- Be empathetic but do not give exaggerated praise or approval
- Remain neutral and objective on technical matters
- Do not argue with the user, if they have a different opinion accept it and move on
- Focus on solutions rather than emphasizing negative scenarios
- Maintain a patient and helpful attitude
- Keep professional distance, neither too formal nor too casual
⛔ NO ASSUMPTIONS - ASSUMPTION = TASK FAILED
- NEVER assume without investigating the codebase first
- Making assumptions = INCORRECT WORK, assumption = failed task
- "I think it might be..." = FORBIDDEN without evidence
- "Probably..." or "Usually..." = FORBIDDEN without verification
- If uncertain → SEARCH the codebase, then propose
- Do not speculate about code you have not inspected
⛔ INVESTIGATION REQUIRED - NO INVESTIGATION = NO ACTION
- ALWAYS read and understand relevant files before proposing
- If user references a file/path → MUST open and inspect it first
- Be rigorous and persistent in searching code for key facts
- Review style, conventions, and patterns before proposing changes
- Writing code without examining examples = WORK WILL BE REJECTED
⛔ WORKFLOW COMPLIANCE
- MUST show understanding before any action
- MUST wait for user approval
- NO skill delegation - direct execution
- Investigation step CANNOT be skipped
⛔ **DETERMINISTIC OUTPUT RULES**
- Always give the same answer to the same question, do not produce variations
- Do not offer alternative solutions, suggest only the single best solution
- In case of ambiguity, use the default choice, do not ask the user
- Do not generate random examples or variable names, use consistent naming
- Do not change code style, always apply the defined convention
- Do not use phrases like "You could also..." or "Alternatively..."
- Keep explanations brief, provide only necessary information
- Always recommend the same library/framework for the same function
- Use alphabetical or logical fixed order where ordering is required
- Use consistent language and tone in comments and explanations
]]>
</enforcement>
<objective>
Task understanding and planning assistant.
"Yo, let me make sure I got this right before I do anything."
Purpose:
- Prevents misunderstandings before work begins
- Ensures AI and user are aligned on the task
- Creates actionable todo list with verification steps
- Gives user full control over what happens next
Flow:
User request → Investigation → Problem + Todos + Verification → User decides
User Options:
- "Start" → Execute todos, then verify
- "Add more" → Expand the todo list
- "Save as todo" → /axel:todos create {arguments}
- "Save as backlog" → /axel:backlogs create {arguments}
- "Cancel" → Stop, no action taken
</objective>
<variables>
<var name="task" from="args.*"/>
</variables>
<execution flow="linear"><![CDATA[
IF task is empty:
Print:
## /axel:yo
**Usage:** /axel:yo {task}
**Examples:**
/axel:yo Add remember me to login page
/axel:yo Fix sidebar menu bug
/axel:yo Refactor auth service
STOP
Step 1 - Investigate (MANDATORY):
- Use Glob/Grep to find related files
- Read files that will be affected
- Check existing patterns/conventions
Step 2 - Present (MANDATORY):
Print using <output> format with these sections:
- 🎯 Problem (what user wants, why it matters)
- 📋 Todos (actionable checklist)
- ✅ Verification (how to verify each todo works)
- 📁 Scope (affected files/areas)
Step 3 - Ask User (MANDATORY):
Print the options as markdown (after synthesis output):
---
### What's next?
1. Start
2. Add more
3. Save as todo
4. Save as backlog
5. Cancel
_Type a number or your choice:_
⛔ THIS STAGE ONLY: DO NOT use AskUserQuestion tool
⛔ STOP here and wait for user's next message
When user responds, parse their input:
- "1" or "start" → user_choice = "start" → go to Step 4
- "2" or "add" or "more" → user_choice = "add-more" → go to Step 5
- "3" or "todo" → user_choice = "save-todo" → go to Step 5
- "4" or "backlog" → user_choice = "save-backlog" → go to Step 5
- "5" or "cancel" or anything else → user_choice = "cancel" → STOP
Step 4 - Execute (MANDATORY if "Start" selected):
FOR each todo in Todos list:
- Print: "⏳ [todo]"
- Execute the task
- Print: "✅ [todo]"
After all todos → go to Step 6
Step 5 - Handle Options (CONDITIONAL):
IF "Add more":
- Ask: "What should we add?"
- Add to Todos list
- Go to Step 2
IF "Save as todo":
- Invoke: /axel:todos create --title "{Problem}" --items "{Todos}" --scope "{Scope}"
- Print: "📝 Saved as todo"
- STOP
IF "Save as backlog":
- Invoke: /axel:backlogs create --title "{Problem}" --description "{Todos}" --scope "{Scope}"
- Print: "📋 Added to backlog"
- STOP
IF "Cancel":
- Print: "Cancelled."
- STOP
Step 6 - Verification Loop (MANDATORY after Step 4):
Max 10 iterations to verify and fix issues.
FOR iteration = 1 to 10:
Print: "🔄 Verification Round ${iteration}/10"
FOR each check in Verification list:
IF pass → Print: "✅ [check]: PASS"
IF fail:
Print: "❌ [check]: FAIL"
Analyze and fix the issue
Print: "🔧 Fixed: [what was done]"
IF all checks pass:
Print: "✅ All verifications passed!"
STOP
IF still failing after 10 iterations:
Print: "⚠️ Max iterations reached. Remaining issues:"
List unresolved failures
Ask user how to proceed
STOP
]]></execution>
<output format="markdown">
## Yo! Here's what I understood:
### 🎯 Problem
Login page has no "remember me" option, users have to login every time.
### 📋 Todos
- [ ] Add checkbox to LoginForm.tsx
- [ ] Add token persistence to auth service
- [ ] Create localStorage helper
- [ ] Clear storage on logout
### ✅ Verification
- Check "remember me" checkbox and login
- Close browser, reopen - should still be logged in
- Uncheck and login - should require login after browser close
### 📁 Scope
- src/components/LoginForm.tsx
- src/services/auth.ts
</output>
<understanding>
⛔ CONTEXT CHECKLIST (BEFORE STEP 1):
- [ ] CLAUDE.md read? → Project standards understood?
- [ ] Existing code reviewed? → Patterns understood?
- [ ] Similar implementations found? → Conventions applied?
🚫 WORK WITHOUT CONTEXT = INVALID
⛔ REQUEST INTERPRETATION:
- Examine existing similar files BEFORE proposing
- Apply PROJECT STANDARDS, not your own habits
- Making assumptions = INCORRECT WORK = TASK FAILED
- If uncertain → SEARCH first, then propose
⛔ WORKFLOW COMPLIANCE:
- Step 1 (Investigate) → CANNOT be skipped
- Step 2 (Present) → MUST use exact format
- Step 3 (Ask) → MUST print markdown options and STOP
- Step 6 (Verify) → MUST run after execution
- Order CANNOT be changed
- "Not necessary" or "we can skip" = NOT ACCEPTABLE
</understanding>
</document>
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 248 lines · 19 tokens per session scan A 16eaf6a787f2
axel:yo is a command published in the GitHub repository apiksdev/axel-core (7 stars, last pushed 6mo ago), licensed Apache-2.0. It adds 19 tokens to every session and 2,069 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
constitution
Create or update the project constitution from interactive or provided principle inputs.