Issue Station
Sign In
Tools hereToolbox

openclaw - 💡(How to fix) Fix Integration: AIP Agent Guard -- skill signing, capability manifests, and audit for OpenClaw [2 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#71443•Fetched 2026-04-26 05:12:38
View on GitHub
Comments
2
Participants
2
Timeline
3
Reactions
0
Author
sunilpsunilp
Participants
steipetesteipetesunilpsunilp
Timeline (top)
commented ×2closed ×1

AIP Agent Guard is an open-source security plugin for OpenClaw that adds verifiable skill author identity, capability manifests with runtime enforcement, and an audit trail using the Agent Identity Protocol.

Available on ClawHub: clawhub install aip-agent-guard Available on npm: npm install -g aip-openclaw

Root Cause

AIP Agent Guard is an open-source security plugin for OpenClaw that adds verifiable skill author identity, capability manifests with runtime enforcement, and an audit trail using the Agent Identity Protocol.

Available on ClawHub: clawhub install aip-agent-guard Available on npm: npm install -g aip-openclaw

Code Example

aip-openclaw keygen
aip-openclaw sign ./my-skill/ --skill-name my-skill
aip-openclaw verify ./my-skill/

---

[manifest]
schema_version = 1
skill_name = "email-summarizer"

[capabilities.mcp_tools]
allow = ["read_email", "list_emails"]

[capabilities.filesystem]
read = ["/tmp/email-cache/*"]
write = "none"

[capabilities.shell]
enabled = false

---

{"ts":"2026-04-25T15:00:01Z","skill":"email-summarizer","tool":"read_email","action":"allow"}
{"ts":"2026-04-25T15:00:02Z","skill":"email-summarizer","tool":"write_file","action":"block","reason":"not in mcp_tools.allow"}
RAW_BUFFERClick to expand / collapse

Summary

AIP Agent Guard is an open-source security plugin for OpenClaw that adds verifiable skill author identity, capability manifests with runtime enforcement, and an audit trail using the Agent Identity Protocol.

Available on ClawHub: clawhub install aip-agent-guard Available on npm: npm install -g aip-openclaw

What problem it solves

OpenClaw skills run with unrestricted access by default. Community analyses have found malicious skills in registries, and several governments have restricted OpenClaw use over security concerns. The core gaps:

  1. No author verification -- anyone can publish a skill, no way to verify who wrote it
  2. No capability boundaries -- a "summarize email" skill can silently access the filesystem or network
  3. No audit trail -- no record of what tools a skill actually called

What it does

Skill Signing

Authors sign their skill directory with Ed25519. A merkle tree covers all files, so any tampering (modified code, added backdoor files) is detectable.

aip-openclaw keygen
aip-openclaw sign ./my-skill/ --skill-name my-skill
aip-openclaw verify ./my-skill/

Capability Manifests

Skills declare what they need in aip-manifest.toml:

[manifest]
schema_version = 1
skill_name = "email-summarizer"

[capabilities.mcp_tools]
allow = ["read_email", "list_emails"]

[capabilities.filesystem]
read = ["/tmp/email-cache/*"]
write = "none"

[capabilities.shell]
enabled = false

Auto-generate a draft manifest: aip-openclaw init ./my-skill/

Runtime Guard

The guard checks every tool call against the manifest and logs decisions:

{"ts":"2026-04-25T15:00:01Z","skill":"email-summarizer","tool":"read_email","action":"allow"}
{"ts":"2026-04-25T15:00:02Z","skill":"email-summarizer","tool":"write_file","action":"block","reason":"not in mcp_tools.allow"}

Progressive Security

LevelAuthor doesUser gets
0NothingSkill works with warnings
1aip-openclaw initCapability enforcement
2aip-openclaw signIdentity verification + enforcement

Built on

Links

Feedback welcome on the integration approach and how this could work with OpenClaw's Plugin SDK hooks.

extent analysis

TL;DR

To enhance security in OpenClaw, utilize the AIP Agent Guard plugin for verifiable skill author identity, capability manifests, and an audit trail.

Guidance

  • Review the AIP Agent Guard documentation to understand its features and installation process.
  • Use aip-openclaw init to auto-generate a draft capability manifest for your skill, ensuring you declare necessary capabilities.
  • Implement Ed25519 signing for your skills using aip-openclaw sign to enable identity verification.
  • Configure the aip-manifest.toml file to define the capabilities your skill requires, such as file system access or tool usage.

Example

[manifest]
schema_version = 1
skill_name = "my-skill"

[capabilities.mcp_tools]
allow = ["read_email", "list_emails"]

[capabilities.filesystem]
read = ["/tmp/email-cache/*"]
write = "none"

Notes

The effectiveness of the AIP Agent Guard depends on proper configuration and adherence to the security protocols outlined in the Agent Identity Protocol (AIP).

Recommendation

Apply the AIP Agent Guard plugin to your OpenClaw setup to leverage its security features, such as capability manifests and identity verification, to mitigate potential security risks associated with skill execution.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#request error#file not found#serialization error#model compatibility#GPU setup

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix Integration: AIP Agent Guard -- skill signing, capability manifests, and audit for OpenClaw [2 comments, 2 participants]