guard: the forge client send functions call the guard

hexajon commented

2026-09-29 16:40:31 +00:00

Owner

What to build

The forge client in internal/gitea receives the guard as an interface when it is built. Each function that sends a request calls the guard before a write (any method other than GET), with the effects of the endpoint. A client with no guard refuses every write. The per-handler out-of-scope checks in the MCP layer go, because the client now runs that rule on every surface.

Depends on

#653, for the guard package.
#654, because the client needs the declared effect of each endpoint.

Acceptance

madt_files create, update, delete, and batch, madt_git_objects apply patch, madt_repos sync_fork and merge_upstream, and madt_branches rename refuse a write to the default branch on the CLI and MCP.
A content write to a populated repo outside the session's scope is refused on the CLI too.
A delete that ADR 0029 omits is refused at the client.
Breaking for anyone who writes files to the default branch through madtea. The release notes say so.

## What to build The forge client in `internal/gitea` receives the guard as an interface when it is built. Each function that sends a request calls the guard before a write (any method other than GET), with the effects of the endpoint. A client with no guard refuses every write. The per-handler out-of-scope checks in the MCP layer go, because the client now runs that rule on every surface. ## Depends on - #653, for the guard package. - #654, because the client needs the declared effect of each endpoint. ## Acceptance - `madt_files` create, update, delete, and batch, `madt_git_objects` apply patch, `madt_repos` sync_fork and merge_upstream, and `madt_branches` rename refuse a write to the default branch on the CLI and MCP. - A content write to a populated repo outside the session's scope is refused on the CLI too. - A delete that ADR 0029 omits is refused at the client. - Breaking for anyone who writes files to the default branch through madtea. The release notes say so.

hexajon added a new dependency

2026-09-29 16:41:10 +00:00

#653 guard: add the guard package with its three effects

hexajon added a new dependency

2026-09-29 16:41:10 +00:00

#654 guard: declare the effects of every operation and every spec endpoint

hexajon referenced this issue

2026-09-29 19:05:28 +00:00

guard: the raw API passthrough keeps every limit #668

hexajon referenced this issue

2026-09-29 19:05:41 +00:00

guard: a parity test fails when a write skips the guard #669

hexajon added a new dependency

2026-09-29 19:05:52 +00:00

#668 guard: the raw API passthrough keeps every limit

hexajon added a new dependency

2026-09-29 19:05:53 +00:00

#669 guard: a parity test fails when a write skips the guard

hexajon referenced this issue

2026-09-29 19:06:46 +00:00

epic: one guard enforces the recorded limits on every surface #670

hexajon added a new dependency

2026-09-29 19:06:51 +00:00

#670 epic: one guard enforces the recorded limits on every surface

hexajon referenced this issue

2026-10-15 23:34:20 +00:00

guard: enable the forge-client guard - tag the recoverable deletes, add the completeness test, flip the install on #780

hexajon added a new dependency

2026-10-15 23:34:31 +00:00

#780 guard: enable the forge-client guard - tag the recoverable deletes, add the completeness test, flip the install on

hexajon referenced this issue

2026-10-15 23:42:58 +00:00

guard: enable the forge-client guard - tag the recoverable deletes, add the completeness test, flip the install on #780

hexajon removed a dependency

2026-10-15 23:43:21 +00:00

#780 guard: enable the forge-client guard - tag the recoverable deletes, add the completeness test, flip the install on

hexajon referenced this issue

2026-10-15 23:43:31 +00:00

guard: enable the forge-client guard - tag the recoverable deletes, add the completeness test, flip the install on #780

hexajon referenced this issue

2026-10-16 01:00:21 +00:00

api passthrough bypasses the forge write guard #784

hexajon referenced this issue from a pull request that will close it,

2026-10-16 02:41:35 +00:00

feat(guard): enforce the ADR 0046 write guard on every forge write surface #790

hexajon referenced this issue from a commit

2026-10-16 02:41:35 +00:00

feat(guard): #660 enforce the ADR 0046 forge write guard on the first-class CLI and MCP surfaces

hexajon commented

2026-10-17 16:49:11 +00:00

Author

Owner

The code for this landed on main, and the host tests pass: the forge client takes the write guard as an interface, the default-branch and out-of-scope writes are refused on both surfaces, and an omitted-class delete is refused at the client. This issue depends on #654, which stays open, so it is not closed yet. #654 has the effect-declaration mechanism in place, but its own dependency #538 (the vendored specs are stale against the live forge) is still open.

hexajon referenced this issue

2026-10-18 02:07:36 +00:00

test(parity): #669 fail the gate when a forge write skips the guard #801

hexajon referenced this issue from a commit

2026-10-18 02:07:36 +00:00

test(parity): #669 fail the gate when a forge write skips the guard

hexajon commented

2026-10-18 05:23:09 +00:00

Author

Owner

The v0.20.0 breaking set is complete and green on main. The code landed across the v0.20.0 branches, the full local gate passes, and the forge suite is green on both forges at both pinned versions on current main (31693766). Closing as done.

The v0.20.0 breaking set is complete and green on main. The code landed across the v0.20.0 branches, the full local gate passes, and the forge suite is green on both forges at both pinned versions on current main (31693766). Closing as done.

hexajon closed this issue

2026-10-18 05:25:19 +00:00

Rows
Columns

guard: the forge client send functions call the guard #660

What to build

Depends on

Acceptance