Skip to content
IAM Insights

Explainers · Glossary

Webhook

webhooks · webhook delivery · callback URL

A way for one service to notify another: the receiver gives the sender an HTTPS address, and the sender posts a signed message there when something happens, instead of the receiver polling.

In our briefings

What our briefings say about Webhook itself, with the most direct claims first.

  1. Chen and Su's draft lets an agent's access token end when its task does, not only when it expires. The client adds a lifecycle_binding to its Rich Authorization Requests (RAR) request, naming a task id and the states that end the task. The one binding type the draft defines is a task-status webhook: the task provider reports those states to the authorization server, which MUST link the task to the token and, when the task ends, "MUST immediately revoke the associated token and propagate this revocation information." Resource servers SHOULD NOT rely solely on the token's expiry and MUST check for revocation in real time, through a revocation feed or token introspection. The authors call the scheme "entirely dependent" on that notice arriving and keep expiry as the fallback; the authorization server has to trust the task provider's webhook, and they say only that mutual TLS and request signing SHOULD protect it. We read the cost as falling on every API: a token carrying the binding can no longer be trusted on its signature and expiry alone, and if the webhook never fires, nothing ends it early.

    A China Mobile draft would end an agent's token when its task ends, if a webhook says soBI-FIND-2026-1251

  2. Talos reads the development builds as a credentials-as-a-service product: the developer compiles one binary per customer with that customer's Discord webhook and AI provider API keys built in, and sells the self-directing layer as the difference. Stolen data reaches the operator's Discord encrypted with a key derived from the date, which Talos calls obfuscation rather than real separation from the developer. One development build carried what a Talos detection-rule comment calls the developer's own DeepSeek and Gemini keys. Talos ties the developer to carding forums going back to 2025 but names no forum or artifact.

    A Windows credential stealer lets four commercial AI models vote on its next move, choosing from four optionsBI-FIND-2026-0879

  3. Snyk's post carries two accounts of how the attacker got the right to publish, and does not reconcile them. Its body says "Investigators believe a maintainer’s npm token with publish rights was compromised, and malicious versions were then published directly to npm", and its opening calls a compromised token "the working theory". A "September 1 update" placed above both says the root cause "is now known" to be a flawed GitHub Actions CI workflow contributed in a pull request on 21 August, which "is estimated to have been generated by Claude Code", followed on 24 August by a malicious commit that changed CI so the npm publishing token would be sent to an attacker-controlled server through a webhook. The post names no source for the Claude Code estimate. The update says that commit "modified the CI workflow"; the commit it shows as figure 2 changes one file, a script named publish-resolve-data.js, adding lines that read the npm auth token from the environment and send it with curl to a webhook.site address. On either account, the malicious versions went out without provenance, which Snyk says recent valid versions had.

    Nx malware ran AI coding CLIs with guardrail-bypass flags to list sensitive file pathsBI-FIND-2025-0809

›Cited in 17 briefings

Every briefing that discusses this term, newest first.

  1. ChatGPT's MCP event subscriptions outlive the token that created them, and the draft sets no re-check schedule

    Article · · WorkOS Blog · Maria Paktiti · Oct 1, 2026

  2. A Block engineer says routing dangling DNS records to their owners cut takeover bug bounties to almost zero

    Podcast · · Cloud Security Podcast · How Block Solved DNS Vulnerabilities · Oct 13, 2026

  3. LiteLLM left a leaked admin key live; its first reported use, 136 days on, yielded an org owner's GitHub token

    Article · · LiteLLM blog · Oliver Jensen et al. · Oct 23, 2026

  4. AAuth carries the signals an agent's approval needs, Hardt says; the brain that decides does not exist yet

    Podcast · · Collateral Dialogue · Self-Driving Software: What OAuth Can’t Do for AI Agents | Dick Hardt, Hello · Sep 2, 2026

  5. A planted page note got browser agents to authorize access in all 73 attack runs; a server-side check stopped it

    Paper · · arXiv · Md Imran Hossen · Oct 10, 2026

  6. A Windows credential stealer lets four commercial AI models vote on its next move, choosing from four options

    Article · · Cisco Talos · Ryan Fetterman · Sep 22, 2026

  7. A2A 1.0 connects agents across companies and leaves out the person they act for

    Article · · A2A Protocol (Linux Foundation) · A2A Protocol project et al. · May 28, 2026

  8. Auth0's own Claude Code tutorial runs the agent on the developer's login and hands it fifteen admin scopes

    Article · · Auth0 Blog · Manish Hatwalne · Sep 29, 2026

  9. AAuth Events gives an agent an inbox, and the subscription tells the resource which agent is listening

    Draft · · IETF Datatracker · Dick Hardt · Sep 28, 2026

  10. By Sierra's own account, its 75,000 agent sessions measure activity, not outcome

    Article · · Sierra · Neil Rahilly · Jul 9, 2026

  11. Sierra's cloud agent runs 75,000 sessions a month and never holds a credential

    Article · · Sierra Engineering · Allen Chen et al. · Jul 15, 2026

  12. RFC 9967 lets SCIM report a deactivated account and leaves what that means to each sender and receiver

    Article · · RFC Editor · Phil Hunt et al. · May 14, 2026

  13. A China Mobile draft would end an agent's token when its task ends, if a webhook says so

    Draft · · IETF Datatracker · Meiling Chen et al. · Feb 4, 2026

  14. The Nx malware had local AI CLIs list wallet and key files with permission checks off

    Article · · StepSecurity · Ashish Kurmi · Aug 27, 2025

  15. Nx malware ran AI coding CLIs with guardrail-bypass flags to list sensitive file paths

    Article · · Snyk · Liran Tal · Aug 27, 2025

  16. Nx traces its stolen npm token to an injectable workflow still alive on old branches

    Article · · nrwl/nx security advisory (GitHub) · Nx maintainers · Aug 27, 2025

  17. An A2A threat model flags the Agent Card as an attack path, and its example flow never passes on the user's identity

    Paper · · arXiv · Idan Habler et al. · Apr 23, 2025