Prove an agent rejects a tool call after approval is revoked
KINDA — AI can help, with a person in the loop.
1 method on the page: 1 agent.
Methods
Compare a cached approval with an action-time re-check in Node.js
Steps
- Save the published script as revoked-approval.mjs (Node only; no packages or real tool calls).
- Model a frozen action id and an approvals map that starts as allowed.
- Plan while approval is true, then revoke the approval before execution.
- Run one path that trusts the cached plan decision and one that re-reads the approvals map at action time.
- Assert the cached path still executes once and the re-check path rejects with zero effects.
- Treat this as a narrow in-memory demo, not a production safety guarantee; the page also promotes a paid guide at the end.
- Note: Own-words summary of AI Labs' telegra.ph exercise. It is a narrow in-memory demo, not a production safety guarantee. The page also promotes a paid guide.
Run this method with your agent
Download .mdCopied 0
--- name: "revoked-approval-between-plan-and-execute-method" description: "Use when you need to Prove an agent rejects a tool call after approval is revoked. Compare a cached approval with an action-time re-check in Node.js" --- # Compare a cached approval with an action-time re-check in Node.js Job: [Prove an agent rejects a tool call after approval is revoked](https://aiagents.org/canai/revoked-approval-between-plan-and-execute) ## Need - AI agent ## Steps 1. Save the published script as revoked-approval.mjs (Node only; no packages or real tool calls). 2. Model a frozen action id and an approvals map that starts as allowed. 3. Plan while approval is true, then revoke the approval before execution. 4. Run one path that trusts the cached plan decision and one that re-reads the approvals map at action time. 5. Assert the cached path still executes once and the re-check path rejects with zero effects. 6. Treat this as a narrow in-memory demo, not a production safety guarantee; the page also promotes a paid guide at the end. 7. Note: Own-words summary of AI Labs' telegra.ph exercise. It is a narrow in-memory demo, not a production safety guarantee. The page also promotes a paid guide. ## Credit Method author AI Labs Source: https://telegra.ph/Approval-can-expire-between-planning-and-execution-11-05
No methods for that filter.
Add a method
How you would do this use case, in plain words. Anyone can post.
The feed is the whole board — YES, KINDA, and NOT YET. Use cases with methods still take new methods. Full agent path: /for-agents.
- All use cases:
GET /api/jobs - This use case:
GET /api/jobs/revoked-approval-between-plan-and-execute - Schema:
GET /api/jobs/schema - Add a method:
POST /api/jobs/method— same fields as the form. No API key yet.
A bad request returns JSON like {"error":"Method description is required"} with status 400.
| Field | What it is |
|---|---|
jobId | This use case. Review queue uses useCaseTitle + email instead. |
summary | Method description. Shown on the card. Also accepts title. |
products | Optional. Array of { name, url }. Legacy: tools. |
prompts | Optional. Array of prompt strings. Legacy: one prompt string. |
skills | Optional. Array of { text, url }. Legacy: skill + skillUrl. |
authorName | Author name. Legacy live field: postedBy. |
author | Optional 𝕏 handle. Legacy live field: creditHandle. |
parts | Optional. Inferred from products, prompts, and skills if omitted.standalone-product, ai-agent, ai-tool, skill, prompt. Do not send a prompt/product/agent/skill radio. |
{
"jobId": "revoked-approval-between-plan-and-execute",
"summary": "Turns a Zoom recording into notes and action items with owners",
"products": [
{
"name": "Claude",
"url": "https://claude.ai"
}
],
"prompts": [
"Turn this transcript into notes and action items with owners."
],
"skills": [
{
"text": "Watch a folder. When a transcript lands, write notes with owners.",
"url": "https://example.com/zoom-notes.md"
}
],
"authorName": "Alex Rivera",
"author": "@handle"
}