Skill
agent-browser
Browser automation for AI agents via inference.sh. Navigate web pages, interact with elements using @e refs, take screenshots, record video. Capabilities: web scraping, form filling, clicking, typing, drag-drop, file upload, JavaScript execution. Use for: web automation, data extraction, testing, agent browsing, research. Triggers: browser, web automation, scrape, navigate, click, fill form, screenshot, browse web, playwright, headless browser, web agent, surf internet, record video
○ declared by the publisher, as-isSafety
Safety · gateassessed 2mo ago
C4 severity-bearing · 16 informational
Graded from 4 severity-bearing findings on this release.
A gate; it passes or fails, never weighed against price or quality.
Reliability & Quality
n/a
Reliability & Qualityassessed 2mo ago
n/anot assessed
Not assessed; no producer has emitted into this category. An absence, not a zero.
Findings with severity, graded on the same letter scale as safety.
not assessed
Cost
n/a
Costassessed 2mo ago
n/anot assessed
Not assessed; no producer has emitted into this category. An absence, not a zero.
Declared external service dependencies. Never a monetary figure.
not assessed
Popularity
n/a
Popularity · not a figureassessed 2mo ago
n/anot assessed
Not assessed; no producer has emitted into this category. An absence, not a zero.
Context read off the linked repository. It enters no ranking and no grade.
not assessed
Performance tested for◌ not assessed
no controlled task evaluation on this release
Repo◌ not assessed
Vettd Note
No Vettd note yet for this asset.
Best for
Not for
Re-assessed 2mo ago · 1 of 7 categories assessed
Record lookup
◌ Vettd does not install assets; this record describes a copy you obtain yourself.
Custody chainSource ↗○ linked, not verified by Vettd
101-skills ◐101-skills/skills ○version = △ no commit pinnedsha256:becc…5afa ◐ signature △ not assessedlicense △ none declaredrecord c3361ee6-4360-4cdb-92ea-a73ffe0a6268 · live
Evidence · 1 of 7 categories · 20 signals · 1 soft spot
Where the record is weakest: 1 soft spot
Ranked worst first.
1Code executes bash but Bash tool not in allowed-toolsHighSafety · high · VTD-0087
/tmp/cisco-shim-hwy2ewwx/SKILL.md — Skill restricts tools to ['Bash(belt', '*)'] but code executes bash commands
The soft spots and their evidence are identical when you are signed in.sign in
Safety● gradedCmost recent 2mo ago20 signals · 1 soft spot
how it was established20 signals · 2 sources · scan
show
Code executes bash but Bash tool not in allowed-toolsflaggedsoft spot 1/tmp/cisco-shim-hwy2ewwx/SKILL.md — Skill restricts tools to ['Bash(belt', '*)'] but code executes bash commandscisco · scan · 2moHigh
References external URL — review for indirect prompt injection riskflaggedExternal URL(s) detected in SKILL.md — referenced content can change after auditvettd · scan · 2moMedium
Skill name collides with well-known skillflagged"agent-browser" matches a well-known skill name — may cause unintended invocationvettd · scan · 2moMedium
Code uses search/grep patterns but Grep tool not in allowed-toolsflagged/tmp/cisco-shim-hwy2ewwx/SKILL.md — Skill restricts tools to ['Bash(belt', '*)'] but code uses regex search patternscisco · scan · 2moLow
No scripts/ directoryflaggedConsider bundling reusable scripts for validation and automationvettd · scan · 2moInfo
Valid name fieldflaggedName "agent-browser" follows spec (lowercase, hyphens, ≤64 chars)vettd · scan · 2moInfo
No repository linkflaggedNo repository field found in SKILL.md frontmatter. Skills without a verifiable source repository cannot be externally audited.vettd · scan · 2moInfo
Validation loop referencedflaggedInstructions for the agent to validate its own work before proceedingvettd · scan · 2moInfo
Progressive disclosure usedflaggedSKILL.md body references files in references/, scripts/, or assets/ — agents can load additional context on demand instead of consuming everything upfrontvettd · scan · 2moInfo
Examples includedflaggedFound code blocks, input/output samples, or an examples section — concrete samples help agents pattern-match effectivelyvettd · scan · 2moInfo
references/ directory presentflaggedAdditional documentation files available for progressive disclosurevettd · scan · 2moInfo
SKILL.md body length is reasonableflagged282 lines (recommended: under 500)vettd · scan · 2moInfo
SKILL.md presentflaggedRequired skill definition file foundvettd · scan · 2moInfo
No evaluation test casesflaggedAdd an evals/ directory with test prompts and expected outputs to measure skill qualityvettd · scan · 2moInfo
Step-by-step workflow foundflaggedStructured procedures improve reliability for complex tasksvettd · scan · 2moInfo
Description includes usage contextflaggedGood: description explains when to activate the skillvettd · scan · 2moInfo
Description within character limitflagged487/1024 characters usedvettd · scan · 2moInfo
Skill does not specify a licenseflaggedSKILL.md — Skill manifest does not include a 'license' field. Specifying a license helps users understand usage terms.cisco · scan · 2moInfo
No secrets or unsafe code patterns detectedflaggedScanned all files for credentials, private keys, and code-level risks (eval, shell exec, destructive ops)vettd · scan · 2moInfo
No prompt injection or jailbreak signals detectedflaggedScanned text content for instruction override, jailbreak framing, credential solicitation, and embedded injection markersvettd · scan · 2moInfo
Reliability & Quality◌ not assessednot assessed0 signals
how it was establishednot assessed; no producer has emitted into this category
Not assessed; no producer has emitted into reliability & quality for this asset. That is different from a check that ran and found nothing, which the record would carry as a row of its own.
Performance◌ not assessednot assessed0 signals
how it was establishednot assessed; no producer has emitted into this category
Not assessed; no producer has emitted into performance for this asset. That is different from a check that ran and found nothing, which the record would carry as a row of its own.
Cost◌ not assessednot assessed0 signals
how it was establishednot assessed; no producer has emitted into this category
Not assessed; no producer has emitted into cost for this asset. That is different from a check that ran and found nothing, which the record would carry as a row of its own.
Compatibility◌ not assessednot assessed0 signals
how it was establishednot assessed; no producer has emitted into this category
Not assessed; no producer has emitted into compatibility for this asset. That is different from a check that ran and found nothing, which the record would carry as a row of its own.
Popularity◌ not assessednot assessed0 signals
how it was establishednot assessed; no producer has emitted into this category
Not assessed; no producer has emitted into popularity for this asset. That is different from a check that ran and found nothing, which the record would carry as a row of its own.
Characteristics◌ not assessednot assessed0 signals
how it was establishednot assessed; no producer has emitted into this category
Not assessed; no producer has emitted into characteristics for this asset. That is different from a check that ran and found nothing, which the record would carry as a row of its own.
Coverage1 of 7 categories assessed · shown
Safety
✓ assessed
assessed as findings and a letter grade, not as signals
Reliability & Quality
△ not assessed
no producer has emitted for this category
Performance
△ not assessed
no producer has emitted for this category
Cost
△ not assessed
no producer has emitted for this category
Compatibility
△ not assessed
no producer has emitted for this category
Popularity
△ not assessed
no producer has emitted for this category
Characteristics
△ not assessed
no producer has emitted for this category
Freshness & re-evaluation
assessed2mo ago · Cisco AI Defense · Vettd core● unreachable
source stateThe repository appears to have moved to inference-sh/skills.Unreachable
re-check policysource unreachable: re-probed on the scheduler's cadence until it answersnext check in 6d
re-run whena new release ships, the upstream source changes, or the scanner ruleset is revisedauto-scheduled
Limitations & open conflicts
limitationReliability & Quality, Performance, Cost, Compatibility, Popularity, Characteristics are not assessed: behaviour there is unmeasured, not clean△ gap
conflictno two sources report the same quantity on this record
Verify: does this record describe your copy?
What record c3361ee6-4360-4cdb-92ea-a73ffe0a6268 covers
scannerv9△ superseded by v11
validityassessed 2mo ago · Unreachable
this record was taken at digest sha256:becc25649700… compare it against the digest your own scan prints
Reproduce: re-run the evidence yourself
runs the same scanners against your own copy; the published record is never overwritten
How this record comparesagainst other published records · full comparison →
Signal kinds
▲ finding: something to act on, carries a severity● measurement: a reproducible quantityƒ estimate: inferred from a sample▪ fact: read directly off the artifact◆ classification: assigned from a taxonomy◌ nothing found: a producer looked and emitted no object◇ context only: described, never graded
Custody marks◐ computed by vettd○ declared by the publisher, as-is△ not assessed: nothing looked