Cairn
NousResearch/hermes-agent

Prompt injection defense via untrusted tool names

Cluster 98587 · agent. Data as of

2 open items. Fix attempted, not merged.

The short answer

1open issue
1open PR
3items grouped
Known?
Yes. 3 related items grouped here, 0 issues already closed.
Fix event
No fix has been merged to main. 1 PR closed without merging; see the attempts below.
Tracker
The maintainers' tracker marks this unresolved. That is separate from the fix event.
Still open
1 issue and 1 PR.

These items address gaps in prompt-injection defenses by expanding _UNTRUSTED_TOOL_NAMES to cover terminal, execute_code, and skill_view, and by restricting how external content reaches the agent through web-cache reads, shell commands, and network code.

Open issues1

  • Issue#98587_UNTRUSTED_TOOL_NAMES doesn't cover terminal/execute_code/skill_view — indirect prompt injection can land through these tools too

Open PRs1

  • PR#101597fix(security): frame side-door external content — web-cache reads, fetching shell commands, network code

1 PR closed without merging

From the close comments: other 1

  • #98663 fix(security): add terminal/execute_code/skill_view to _UNTRUSTED_TOOL_NAMES for prompt-injection defense · other → #98587
How this cluster was grouped
  • Link evidence (structural links throughout): every member is connected by structural evidence: shared changed files, or a closing/superseded reference.
  • Pairwise link density 0.67: the share of item pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. 1.00 means every pair is linked.

Pick up a Lab task

Check whether a fix covers the open issues here, with your own agent. Bring back evidence for maintainers.

Open the Lab

At a glance

Subsystem
agent
Closed issues
0
Attempts
1 PR closed without merging

For your agent

The same facts, typed.

cluster/98587.json
cluster/98587.md