The model was enable to edit `internal/ui/model/ui.go` for me, because
it has ~ 108KB and the limit was 100KB.
(Without being able to read, the model also can't update it).
We're keeping the `fetch` tool limit to 100KB.
Andrey Nering
created
c7368b7
fix: fix thinking on/off toggle for certain openai-compat providers
Andrey Nering
created
59e9233
refactor(hyper): simplify by removing old code
Andrey Nering
created
54a57fd
chore(legal): @carlosgrillet has signed the CLA
Charm
created
eaf1cd7
chore(ui): change wording: rewrote input to rewrote output (#2742)
Christian Rocha
created
35ce257
chore(legal): @lloydzhou has signed the CLA
Charm
created
c9afdb1
chore(legal): @vorticalbox has signed the CLA
aeb529c
chore(legal): @georgeglarson has signed the CLA
Charm
created
86bb805
fix(tools/job_kill): use longer job_kill desc to improve reliability (#2747)
Christian Rocha
created
4d16e96
fix(app): replace single events channel with pubsub.Broker for fan-out (#2663)
Click to expand commit body
Fixes the SSE multi-consumer bug where multiple clients sharing the same
workspace received only a fraction of events (competing-consumer semantics
on a plain channel). Each caller to Events()/SubscribeEvents() now gets
its own dedicated channel via Broker.Subscribe(ctx).
💘 Generated with Crush
Assisted-by: AWS Claude Sonnet 4.6 via Crush <crush@charm.land>
Tai Groot
created
c84ca2d
refactor(hooks): move matcher compilation into the runner
Click to expand commit body
The runner now compiles its own matchers, so a config round-trip can't
silently strip them. Hooks with an invalid matcher are dropped with a
warning instead of quietly matching everything.
Christian Rocha
created
38a846d
fix(hooks): recompile matchers after config reload
Click to expand commit body
A hook with a matcher like "^bash$" was firing on every tool call
after any config write, because the compiled regex was dropped when
the config reloaded. Recompile on reload so filtering keeps working.
e5d95aa
fix(hyper): fix re-authorization flow not triggering on certain conditions (#2703)
Click to expand commit body
We attempt to refresh tokens in two scenarios:
* Pro-actively, when we expiration date passed before a request
* Reactively, when we try a request but a 401 was returned
The error / event handling that shows a dialog asking the user to
re-authenticate is dependent on code that runs for the second scenario.
This means that the dialog might not show on the first scenario, if
refreshing the token fails.
If refreshing a token fails, continue with the previous token. Let it
fail and the 401 will trigger a re-authorization dialog.
I caught this because I was switching between local vs. prod Hyper a lot.
Andrey Nering
created
e5b5a40
Merge pull request #2692 from charmbracelet/hypercrush
Click to expand commit body
chore(styles): overhaul style definitions
Christian Rocha
created
81daa26
fix(styles): use semantic names in styles + drop deadcode
Christian Rocha
created
4f3a308
chore(ui): formal hypercrush type treatment
Christian Rocha
created
511916a
fix(ui): logo and grad arguments from earlier refactor
15f20e6
chore(ui): add new letterforms: h, y, p, e, with alts
Christian Rocha
created
549f717
fix(agent): implement OnRetry logging with structured retry fields (#2700)
iceymoss
created
f669bd3
fix(lsp): replace sticky unavailable cache with retry backoff (#2498)
Click to expand commit body
Scope unavailable tracking to each Manager instance and retry missing server commands after a short backoff so temporary environment changes do not permanently block LSP startup.