markdown · 10156 bytes Raw Blame History

Permissions

Every authorization decision in shithub flows through one function: policy.Can(ctx, deps, actor, action, repo) → Decision. Handlers, hooks, and the SSH/HTTP git transports all funnel through this single entrypoint. No surface reads ownership, visibility, or collaborator state inline; the lint guard at scripts/lint-policy-boundary.sh enforces the boundary in CI.

The shape

  • Actor — who's asking. Anonymous, logged-in user (with IsSuspended and IsSiteAdmin flags), or future org-team principal (S31).
  • Action — what they want to do, drawn from the constant registry in internal/auth/policy/actions.go. New actions go in their owning sprint's PR; the matrix test ensures every constant is covered for every actor archetype.
  • Resource — currently a RepoRef; issue/pull/org refs land in their owning sprints. The RepoRef carries OwnerOrgID shape today (zero) so S31 plugs in without retro-fitting the interface.
  • Decision{Allow bool; Reason string; Code DenyCode}. Allow drives control flow; Code lets handlers pick a friendly user-facing message without re-deriving from the resource fields. Reason is for logs and tests, never end-user surfaces.

Role hierarchy

Five collaborator tiers, mirroring GitHub:

Role Implies Granted by
read Clone/fetch a private repo, view issues/pulls invitation by maintainer
triage read + close/label/assign issues, no code write invitation by maintainer
write triage + push, branch create, PR create/comment invitation by maintainer
maintain write + most settings (general, branches) invitation by admin
admin maintain + delete/transfer/visibility/destructive invitation by admin/owner

The repo owner is implicit admin — there's no repo_collaborators row for them. repo_collaborators lives in migration 0019 and is keyed by (repo_id, user_id).

Action → minimum role table

The complete map (also enforced by the matrix test):

Action Min role on repo
repo:read read (private)
repo:write write
repo:admin admin
repo:settings:general maintain
repo:settings:collaborators admin
repo:settings:branches maintain
repo:archive admin
repo:delete admin
repo:transfer admin
repo:visibility admin
issue:read read (private)
issue:create write
issue:comment write
issue:close triage
issue:label triage
issue:assign triage
pull:read read (private)
pull:create write
pull:merge admin
pull:review write
pull:close write
star:create logged in
fork:create logged in
watch:set logged in

Read actions on public repos are short-circuited to allow before the role check — anyone (anonymous or otherwise) can read a public repo.

Decision precedence

Can() evaluates in a fixed order; the first matching rule produces the verdict. Ordered from most-decisive to least:

  1. Soft-deleted repo → deny (DenyRepoDeleted). Nothing else matters.
  2. Site-admin + read action → allow. (Write actions still go through the rest of the pipeline; broad admin overrides hide bugs and create insider-threat surface.)
  3. Suspended actor + write action → deny (DenyActorSuspended). Reads against public repos still allowed.
  4. Anonymous + private repo → deny (DenyVisibility). Caller maps to 404, not 403, to avoid existence leak.
  5. Public repo + read → allow.
  6. Compute effective role (owner ⇒ admin; collaborator ⇒ row.role; else RoleNone).
  7. Archived repo + write → deny (DenyArchived). Even owners can't push to archived repos.
  8. Min role for action vs effective role. Below threshold + private repo + no role → deny as visibility (404). Below threshold with any role → deny as DenyRoleTooLow (403).
  9. Login-required actions (star/fork) on anonymous → deny (DenyAnonymous).

Existence-leak guard

policy.Maybe404(decision, repo, actor) maps a denial to a status code that doesn't reveal whether a private repo exists. Convention:

  • Allow → 200.
  • Deny on a private repo, viewer is not the owner → 404.
  • Deny on a private repo, viewer is the owner (e.g. push to archived) → 403, since the owner already knows the repo exists.
  • Deny on a public repo → 403.

Handlers that care about user-facing message tone (e.g. the HTTP git handler's "repository is archived" stderr line) should switch on Decision.Code rather than parse Decision.Reason.

Per-request memoization

policy.WithCache(ctx) attaches a request-scoped memo. The web layer wires this in internal/web/middleware/policy.go::PolicyCache. Within one request, repeated Can() calls for the same (actor, repo) pair hit the cache. Across requests there's no cache — staleness is hard to get right and the per-request DB cost of fresh lookups is acceptable.

If a handler mutates collaborator state mid-request and re-checks policy in the same flight, call policy.InvalidateRepo(ctx, repoID) between the mutation and the re-check.

Suspended actors and the auth surfaces

The IsSuspended flag on Actor is the canonical input the policy package uses to deny writes by suspended accounts. Each entrypoint that constructs an actor must source it correctly:

  • Web (session)middleware.OptionalUser populates CurrentUser.IsSuspended from users.suspended_at. Handlers pass viewer.IsSuspended straight into policy.UserActor. The lookup is run on every request (no cookie-baked state), so an admin suspending an account takes effect on the user's next click.
  • Web (PAT)middleware.PATAuthMiddleware rejects requests whose owning user has suspended_at IS NOT NULL with a 401 before the handler runs. Code paths under PAT auth construct policy.UserActor(..., IsSuspended: false, ...) because the gate is upstream; the field is still passed for honesty and is correct by construction.
  • git over HTTPS (internal/web/handlers/githttp) — the basic- auth resolver (auth.go::resolveViaPAT/resolveViaPassword) rejects suspended owners with errBadCredentials before the policy check runs, so the policy.UserActor(..., false, ...) call in handler.go never sees a suspended actor. Suspension on the HTTPS git path is enforced at credential resolution, not at policy evaluation. If the credential resolver is ever reorganised to return a populated user even for suspended accounts, propagate the flag here.
  • git over SSH (internal/git/protocol/ssh_dispatch.go) — the dispatcher loads the user row before constructing the actor and passes user.SuspendedAt.Valid directly into policy.UserActor. The authorized_keys invocation also rejects up-front (see docs/internal/git-ssh.md), but the policy call is the defence-in-depth layer.
  • post-receive hook (cmd/shithubd/hook.go) — same shape as SSH dispatch: load user, pass SuspendedAt.Valid into the actor.

When adding a new auth entrypoint (e.g. an OAuth-bearing webhook ingest), the rule is: load the user record, source IsSuspended from users.suspended_at, and never hard-code false.

Site-admin scope

actor.IsSiteAdmin = true short-circuits to allow on read actions only. Write actions go through the normal role check, which means a site admin who is not a collaborator on a private repo cannot push or change settings without explicit impersonation (S34 ships the impersonation surface). Non-impersonated admin write attempts go through Can() like any other request and audit-log loudly via the S05 audit recorder.

Adding a new action

  1. Add the constant to internal/auth/policy/actions.go.
  2. Append it to AllActions in the same file.
  3. Add a case to minRoleFor(action) in policy.go. Unknown actions default to RoleAdmin (deny by default for strangers).
  4. The matrix test (policy_test.go) iterates AllActions and will automatically demand a verdict for every actor × resource × this action combination. Update mirrorMinRoleFor in the test file with the same minimum role.

If you add an action that involves a new resource type, add a *Ref struct in resources.go and an adapter in adapters.go, then overload Can with a new entrypoint (e.g. CanIssue, CanOrg).

Boundary lint

scripts/lint-policy-boundary.sh runs as part of make ci. It fails when the following patterns appear outside internal/auth/policy/, internal/repos/, and internal/web/handlers/repo/ (which constructs the policy actor at the lookup wrapper):

  • OwnerUserID == or == row.OwnerUserID — direct owner equality
  • Visibility == reposdb.Repo... — direct visibility branching
  • if X.IsArchived — archived as a control predicate

Test files everywhere are exempt — they legitimately seed state. If a new pattern surfaces (e.g. an issue handler reads issue.author_id), extend the script accordingly.

View source
1 # Permissions
2
3 Every authorization decision in shithub flows through one function:
4 `policy.Can(ctx, deps, actor, action, repo) → Decision`. Handlers,
5 hooks, and the SSH/HTTP git transports all funnel through this single
6 entrypoint. No surface reads ownership, visibility, or collaborator
7 state inline; the lint guard at `scripts/lint-policy-boundary.sh`
8 enforces the boundary in CI.
9
10 ## The shape
11
12 * **Actor** — who's asking. Anonymous, logged-in user (with `IsSuspended`
13 and `IsSiteAdmin` flags), or future org-team principal (S31).
14 * **Action** — what they want to do, drawn from the constant registry
15 in `internal/auth/policy/actions.go`. New actions go in their owning
16 sprint's PR; the matrix test ensures every constant is covered for
17 every actor archetype.
18 * **Resource** — currently a `RepoRef`; issue/pull/org refs land in
19 their owning sprints. The `RepoRef` carries `OwnerOrgID` shape today
20 (zero) so S31 plugs in without retro-fitting the interface.
21 * **Decision** — `{Allow bool; Reason string; Code DenyCode}`. Allow
22 drives control flow; Code lets handlers pick a friendly user-facing
23 message without re-deriving from the resource fields. Reason is for
24 logs and tests, never end-user surfaces.
25
26 ## Role hierarchy
27
28 Five collaborator tiers, mirroring GitHub:
29
30 | Role | Implies | Granted by |
31 | ---------- | ------------------------------------------------- | ------------------------- |
32 | `read` | Clone/fetch a private repo, view issues/pulls | invitation by maintainer |
33 | `triage` | read + close/label/assign issues, no code write | invitation by maintainer |
34 | `write` | triage + push, branch create, PR create/comment | invitation by maintainer |
35 | `maintain` | write + most settings (general, branches) | invitation by admin |
36 | `admin` | maintain + delete/transfer/visibility/destructive | invitation by admin/owner |
37
38 The repo **owner** is implicit `admin` — there's no
39 `repo_collaborators` row for them. `repo_collaborators` lives in
40 migration `0019` and is keyed by `(repo_id, user_id)`.
41
42 ## Action → minimum role table
43
44 The complete map (also enforced by the matrix test):
45
46 | Action | Min role on repo |
47 | ------------------------------------- | ---------------- |
48 | `repo:read` | `read` (private) |
49 | `repo:write` | `write` |
50 | `repo:admin` | `admin` |
51 | `repo:settings:general` | `maintain` |
52 | `repo:settings:collaborators` | `admin` |
53 | `repo:settings:branches` | `maintain` |
54 | `repo:archive` | `admin` |
55 | `repo:delete` | `admin` |
56 | `repo:transfer` | `admin` |
57 | `repo:visibility` | `admin` |
58 | `issue:read` | `read` (private) |
59 | `issue:create` | `write` |
60 | `issue:comment` | `write` |
61 | `issue:close` | `triage` |
62 | `issue:label` | `triage` |
63 | `issue:assign` | `triage` |
64 | `pull:read` | `read` (private) |
65 | `pull:create` | `write` |
66 | `pull:merge` | `admin` |
67 | `pull:review` | `write` |
68 | `pull:close` | `write` |
69 | `star:create` | logged in |
70 | `fork:create` | logged in |
71 | `watch:set` | logged in |
72
73 Read actions on **public** repos are short-circuited to allow before the
74 role check — anyone (anonymous or otherwise) can read a public repo.
75
76 ## Decision precedence
77
78 `Can()` evaluates in a fixed order; the first matching rule produces
79 the verdict. Ordered from most-decisive to least:
80
81 1. **Soft-deleted repo** → deny (`DenyRepoDeleted`). Nothing else
82 matters.
83 2. **Site-admin + read action** → allow. (Write actions still go
84 through the rest of the pipeline; broad admin overrides hide bugs
85 and create insider-threat surface.)
86 3. **Suspended actor + write action** → deny (`DenyActorSuspended`).
87 Reads against public repos still allowed.
88 4. **Anonymous + private repo** → deny (`DenyVisibility`). Caller maps
89 to 404, not 403, to avoid existence leak.
90 5. **Public repo + read** → allow.
91 6. Compute effective role (owner ⇒ admin; collaborator ⇒ row.role;
92 else `RoleNone`).
93 7. **Archived repo + write** → deny (`DenyArchived`). Even owners
94 can't push to archived repos.
95 8. **Min role for action** vs effective role. Below threshold + private
96 repo + no role → deny as visibility (404). Below threshold with any
97 role → deny as `DenyRoleTooLow` (403).
98 9. **Login-required actions** (star/fork) on anonymous → deny
99 (`DenyAnonymous`).
100
101 ## Existence-leak guard
102
103 `policy.Maybe404(decision, repo, actor)` maps a denial to a status
104 code that doesn't reveal whether a private repo exists. Convention:
105
106 * Allow → 200.
107 * Deny on a private repo, viewer is not the owner → **404**.
108 * Deny on a private repo, viewer **is** the owner (e.g. push to
109 archived) → **403**, since the owner already knows the repo exists.
110 * Deny on a public repo → **403**.
111
112 Handlers that care about user-facing message tone (e.g. the HTTP git
113 handler's "repository is archived" stderr line) should switch on
114 `Decision.Code` rather than parse `Decision.Reason`.
115
116 ## Per-request memoization
117
118 `policy.WithCache(ctx)` attaches a request-scoped memo. The web layer
119 wires this in `internal/web/middleware/policy.go::PolicyCache`. Within
120 one request, repeated `Can()` calls for the same `(actor, repo)` pair
121 hit the cache. Across requests there's no cache — staleness is hard to
122 get right and the per-request DB cost of fresh lookups is acceptable.
123
124 If a handler mutates collaborator state mid-request and re-checks
125 policy in the same flight, call `policy.InvalidateRepo(ctx, repoID)`
126 between the mutation and the re-check.
127
128 ## Suspended actors and the auth surfaces
129
130 The `IsSuspended` flag on `Actor` is the canonical input the policy
131 package uses to deny writes by suspended accounts. Each entrypoint
132 that constructs an actor must source it correctly:
133
134 * **Web (session)** — `middleware.OptionalUser` populates
135 `CurrentUser.IsSuspended` from `users.suspended_at`. Handlers pass
136 `viewer.IsSuspended` straight into `policy.UserActor`. The lookup
137 is run on every request (no cookie-baked state), so an admin
138 suspending an account takes effect on the user's next click.
139 * **Web (PAT)** — `middleware.PATAuthMiddleware` rejects requests
140 whose owning user has `suspended_at IS NOT NULL` with a 401 before
141 the handler runs. Code paths under PAT auth construct
142 `policy.UserActor(..., IsSuspended: false, ...)` because the gate
143 is upstream; the field is still passed for honesty and is correct
144 by construction.
145 * **git over HTTPS (`internal/web/handlers/githttp`)** — the basic-
146 auth resolver (`auth.go::resolveViaPAT`/`resolveViaPassword`)
147 rejects suspended owners with `errBadCredentials` *before* the
148 policy check runs, so the `policy.UserActor(..., false, ...)` call
149 in `handler.go` never sees a suspended actor. Suspension on the
150 HTTPS git path is enforced at credential resolution, not at policy
151 evaluation. If the credential resolver is ever reorganised to
152 return a populated user even for suspended accounts, propagate the
153 flag here.
154 * **git over SSH (`internal/git/protocol/ssh_dispatch.go`)** — the
155 dispatcher loads the user row before constructing the actor and
156 passes `user.SuspendedAt.Valid` directly into `policy.UserActor`.
157 The `authorized_keys` invocation also rejects up-front (see
158 `docs/internal/git-ssh.md`), but the policy call is the
159 defence-in-depth layer.
160 * **post-receive hook (`cmd/shithubd/hook.go`)** — same shape as
161 SSH dispatch: load user, pass `SuspendedAt.Valid` into the actor.
162
163 When adding a new auth entrypoint (e.g. an OAuth-bearing webhook
164 ingest), the rule is: load the user record, source `IsSuspended`
165 from `users.suspended_at`, and *never* hard-code `false`.
166
167 ## Site-admin scope
168
169 `actor.IsSiteAdmin = true` short-circuits to allow on read actions
170 only. Write actions go through the normal role check, which means a
171 site admin who is **not** a collaborator on a private repo cannot
172 push or change settings without explicit impersonation (S34 ships the
173 impersonation surface). Non-impersonated admin write attempts go
174 through Can() like any other request and audit-log loudly via the S05
175 audit recorder.
176
177 ## Adding a new action
178
179 1. Add the constant to `internal/auth/policy/actions.go`.
180 2. Append it to `AllActions` in the same file.
181 3. Add a case to `minRoleFor(action)` in `policy.go`. Unknown actions
182 default to `RoleAdmin` (deny by default for strangers).
183 4. The matrix test (`policy_test.go`) iterates `AllActions` and will
184 automatically demand a verdict for every actor × resource × this
185 action combination. Update `mirrorMinRoleFor` in the test file with
186 the same minimum role.
187
188 If you add an action that involves a new resource type, add a `*Ref`
189 struct in `resources.go` and an adapter in `adapters.go`, then
190 overload `Can` with a new entrypoint (e.g. `CanIssue`, `CanOrg`).
191
192 ## Boundary lint
193
194 `scripts/lint-policy-boundary.sh` runs as part of `make ci`. It fails
195 when the following patterns appear outside `internal/auth/policy/`,
196 `internal/repos/`, and `internal/web/handlers/repo/` (which constructs
197 the policy actor at the lookup wrapper):
198
199 * `OwnerUserID == ` or `== row.OwnerUserID` — direct owner equality
200 * `Visibility == reposdb.Repo...` — direct visibility branching
201 * `if X.IsArchived` — archived as a control predicate
202
203 Test files everywhere are exempt — they legitimately seed state. If a
204 new pattern surfaces (e.g. an issue handler reads `issue.author_id`),
205 extend the script accordingly.