CreepScope measures privilege lifetime after last legitimate use, not merely whether a call was initially authorized. Across 96 deterministic replay cells, step leases removed stale exposure by construction; in a small simulated mail agent episode, persistent grants enabled nine unauthorized sends while the leased configuration blocked three attempts. The…
Inspect the ClaimsCase · DiggingBeagle record
CreepScope tests agent permissions that survive their legitimate use
Controlled workflow replay distinguishes stale permission exposure from authority that a task never needed.
Scope
Research claim attributed to the cited report, with limitations from intake DB-RADAR-CP7-CP7A-003.
Assessment profile
Assessment method- Setting
- simulation
- Exploitation
- demonstrated
- Evidence
- Grade C
- Remediation
- mitigation available
- AI role
- AGAINST AI
Basis and provenance
Replay experiments deliberately specified last legitimate use and compared persistent versus step-limited grants. No field distribution of stale permissions was measured. Assessment uses the pinned Claims and source identities; public primary pages were not independently retrievable during this run. Impact remains unrated because field consequence or a comparable scoring rubric is not established.
Assessed Oct 15, 2026 using diggingbeagle.assessment/1.
Evidence basis: CLM-CP7A-CREEPSCOPE · CreepScope measures stale privileges surviving final legitimate task use — original research or disclosure
At a glance
Independent reproduction or full primary-document access in this research run was not achieved. Real-world incidence and broader model or deployment generalization are not established unless specifically stated.
Limits and uncertaintyFull account
CreepScope measures privilege lifetime after last legitimate use, not merely whether a call was initially authorized. Across 96 deterministic replay cells, step leases removed stale exposure by construction; in a small simulated mail agent episode, persistent grants enabled nine unauthorized sends while the leased configuration blocked three attempts. The tested oracle knows the final legitimate use, a difficult production assumption.
Lease expiry at the final authorized step eliminates the measured stale window by construction. Real services rarely know that final step in advance, so the production question is how revocation follows task state without breaking delayed authorized actions.
Claims & evidence
CLM-CP7A-CREEPSCOPEThe cited authors or maintainers report the following bounded finding: Controlled workflow replay distinguishes stale permission exposure from authority that a task never needed.supported
Basis: reported finding
- supportsCreepScope measures stale privileges surviving final legitimate task use — original research or disclosureprimary disclosure
Original primary document, reported finding and evaluation context; checkpoint DB-RADAR-CP7-CP7A-003
Unknowns and contradictions
- Independent reproduction or full primary-document access in this research run was not achieved.
- Real-world incidence and broader model or deployment generalization are not established unless specifically stated.
Sources and citation
Material revision history
- Oct 15, 2026 · Canonical change recorded · new in release · revision 150