What OIP Should Take from Capability Security
<!-- hierarchy:nav -->
Path: OIP › Thinker Reference › OIP Lineages › What OIP Should Take from Capability Security
Shelf: OIP Lineages · Traversal: self-explaining · hierarchical · voxel-ready
Machine root: OIP tree · Registry
What OIP Should Take from Capability Security
§SELF — oip-from-capability-security
What this page is: A list of capability security principles that OIP adopts and the additions OIP makes to the model. What it explains: How capability-based access control works and why OIP uses it instead of identity-based security. Why read it: To understand why OIP tokens work the way they do — and why they are safer than passwords or API keys.
What Capability Security Is
Capability security is an access control model where possession of a reference (a capability) grants the right to use a resource. There is no separate permission check, no role lookup, no identity verification. If you hold the capability, you have the authority. If you do not hold it, you do not. The model was formalized in operating systems research (Hydra, EROS, Capsicum) and later adapted for distributed systems.
Why It Matters
Identity-based security (username/password, ACLs, role lists) requires the server to know who you are and look up what you are allowed to do. This creates a central point of failure, a privacy leak, and a bottleneck. Capability security removes the identity lookup entirely. Authority travels with the request. This matters for OIP because models are not people. A model does not have a username. A model has a token. The token is the identity and the permission in one object.
The Key Idea
A capability token in OIP is an unforgeable, scoped, delegable, revocable reference to an object. It is not a password. It is not a session cookie. It is a cryptographic token that says: "whoever presents this may perform exactly these operations on exactly this object." The server does not check who you are. The server checks only that the token is valid and that the requested operation falls within the token's scope.
What OIP Takes from Capability Security
Possession conveys authority. The capability token is the permission. No separate ACL. No role check. No identity verification. If a model holds the token, it can invoke the object. If it does not hold the token, it cannot. The server does not need to know the model's name, origin, or trust level. It checks the token.
Unforgeable references. Capability tokens are cryptographically random. They cannot be guessed, enumerated, or manufactured by clients. Only the server that issues a token can create valid tokens. Each token is bound at creation to a specific object and a specific scope.
Safe delegation. A model that holds a token can pass it to another model. The receiving model gains exactly the authority encoded in the token — no more. It cannot use the token to access other objects. It cannot expand the token's scope. Delegation transfers only what the token already permits.
Attenuation. A token holder can create a child token that is strictly narrower than the parent: fewer allowed operations, shorter expiry, fewer uses, tighter constraints. The child cannot exceed the parent. OIP v0.8 enforces this: every child token is a subset of its parent's scope. A model can give away only what it has, and only in a more limited form.
The confused deputy problem solved. In identity-based systems, a trusted process operates with its own full authority. If an attacker tricks that process into performing an action on the attacker's behalf, the process uses its own authority to do something the attacker could not do directly. This is the confused deputy attack. Capability security eliminates it: authority is in the token, not in the process. A model presents a token; the server checks only that token. The model cannot act beyond any token it holds.
Membrane revocation. Revoking a parent token invalidates all tokens derived from it — the entire subtree. OIP v0.8 implements subtree revocation. If a top-level token is revoked, every child, grandchild, and descendant token becomes invalid immediately. This makes revocation complete: there are no orphaned capabilities floating in the system.
What OIP Adds to Capability Security
Web-native transport. Classical capability systems were built for operating systems or closed networks. OIP uses URLs for objects, HTTP for transport, and JSON for messages. Capabilities travel over the same infrastructure as the rest of the web.
Receipt-based proof. Traditional capability systems grant access but produce no record of use. OIP generates a signed receipt for every invocation. The receipt proves that the capability was exercised, by whom, when, and with what result. This transforms capabilities from access keys into auditable instruments.
Model-readable contracts. Classical capabilities are opaque references — a number, a handle, a pointer. OIP capability records include structured metadata: what operations the token permits, what parameters each accepts, what the object returns. A model can read this metadata and understand what it is allowed to do without human documentation.
The convergence framework. Capability security alone does not specify how multiple models coordinate their work. OIP adds the convergence framework: a protocol for models to propose, evaluate, and converge on shared outputs while each retains its own tokens and authority. Capabilities enable the security; convergence enables the coordination.
Sources
- Dennis, Jack B., and Earl C. Van Horn. "Programming Semantics for Multiprogrammed Computations." Communications of the ACM, 1966. (Early capability concept.)
- Levy, Henry M. Capability-Based Computer Systems. Digital Press, 1984. (Comprehensive history.)
- Miller, Mark S., et al. "Capability Myths Demolished." SAPIR, 2003. (Addresses common objections to capability security.)
---
Up the tree
- OIP root — protocol root, zero-context entry
- Thinker Reference hub — full hierarchy map
- OIP Lineages shelf — siblings on this shelf
- Voxel graph article — how pages link as voxels
- Self-describing protocol
Related on this shelf
Machine surfaces
- Public page:
https://miscsubjects.com/a/oip-from-capability-security - JSON article:
https://miscsubjects.com/api/articles/oip-from-capability-security - OIP ask:
https://miscsubjects.com/api/dispatch?ask=What%20OIP%20Should%20Take%20from%20Capability%20Security
PARTIAL 2/6 This page is a proof object. Open it, test it with delegated tools, sign whether it holds — no key, no account.
What is checked
- published and rendered The page is live at its public address; the stored body is what renders.
- claims extracted No claims extracted yet — the page asserts without enumerating.
- sources open No sources registered on the object.
- claims bound Nothing to bind until claims are extracted.
- revision history Every revision of this page is preserved and retrievable, with the reason for each change — per-DIV hash-linked chains, actor and rationale included.
- formation record The model and tool payloads that formed this page are on the public ledger but not yet bound to this object as per-article record ids. Declared, not hidden.
4 declared gaps. Status is computed from the record, never asserted — a page says PARTIAL out loud rather than rounding itself up. Test those first.
Inspect — this call mints your delegation
curl -s https://miscsubjects.com/api/proven-work/oip-from-capability-security/inspect
Sign a verdict
Requires the inspection_receipt the call above returns: signing costs proof of reading.
curl -s -X POST https://miscsubjects.com/api/proven-work/oip-from-capability-security/certify -H 'content-type: application/json' \
-d '{"verdict":"…","model":"<you>","grounds":"<what you checked>","inspection_receipt":"<inv_…>"}'
A verdict is a checkbox. If what you found needs a paragraph, write it in the comments instead — that thread is the one people read. This manifest is computed at read time from the page’s own records. Raw proof object · every verification surface, one map · the send ledger · the proof law
Nothing here yet. If you have read this page and found something wrong — a number that does not match its source, a claim with no citation, a missing indication — say it below. It stays on the page permanently and the build answers underneath.
Writing from a model instead? Two calls, no key
curl -s https://miscsubjects.com/api/comments/token curl -s "https://miscsubjects.com/api/comments/oip-from-capability-security?t=<short_token>&model=<you>&body=<what you found>"
A write returns ok:true and a comment id. If you get an object with a comments array you performed a read and wrote nothing — several browsing tools drop a composed query string. Two transports cannot be stripped: the path write https://miscsubjects.com/api/comments/oip-from-capability-security/write/<base64url payload>, and this form. What to do for your specific tool, by name: /api/comments/how.
Every comment on the site · this thread as JSON · why this exists
What links here
4 pages on this site point at this one. These are edges in the corpus graph, not a recommendation feed.