Main merged the sandbox-driver adoption (#849) in a later form than this
branch was stacked on: the driver's own exec types replace fabro-sandbox's,
shell quoting moved to fabro-util, the sandbox lifecycle collapsed, and the
driver's events are stored as run events. This branch had deleted
`fabro-agent` and put the coding agent, the environment adapter, and the
steering hub on pebble.
The resolution takes main's sandbox API and re-applies pebble on top: the
`RunSandbox` `Environment` adapter moves to `pebble_environment.rs` (main's
`environment.rs` is the sandbox spec) and runs commands through `ExecSpec`
and `ExecControls`, feeding pebble's output sink from the driver's; the
driver-era `sandbox.*` names leave the known-event list, as on main, so a
stored event with that name and no driver shape is `Unknown` rather than an
error; `program_exit_code` matches pebble's non-exhaustive termination; the
Docker and Daytona smokes use main's constructor and credentials; the
remaining `fabro_agent` paths point at fabro-sandbox.
Pebble's `mcp` feature pins sandbox-driver, and the preview-url trait
objects only cross when both sides name one revision, so pebble moved to
main's `a92c0db6` (lithoscomputer/pebble#10) and fabro pins that pebble
revision until it lands on pebble main.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Replace fabro's hand-written agent loop with pebble's `CodingAgent` and
delete the `fabro-agent` crate.
Workflow: `PebbleBackend` builds one agent per stage over `RunSandbox`,
binds the stage's hooks as tool middleware, the interviewer as the
human-input provider, and a durable `EventSink` that writes every agent
event through the run event log before the agent goes on. Full-fidelity
threads continue across stages through `export`/`resume_from_export`.
Model failover takes the session record after the failed prompt and
continues it on the next route with `ResumeMode::UseModel`, so no tool
effect repeats. The steering hub targets pebble's control handle, with
a steering lease holding completion open while a human is paired.
Events: `EventBody::Agent` carries pebble's `CodingAgentEvent` envelope;
the per-variant bodies, the transcript projection, and the fabro-only
context-window, tool-summary, and skill types are gone in favor of
pebble's. The OpenAPI schemas, generated Rust and TypeScript clients,
and web readers follow.
Ask Fabro: the session runs a `CodingAgent` under a read-only permission
policy and a system prompt transform. Its conversation lives in a new
`run_session_records` table and resumes on the recorded model with the
event cursor advanced past the run log.
`fabro exec` builds the same agent over a local sandbox with pebble's
permission middleware and an interactive approval service.
The catalog fills in `metadata.agent.profile` for operator providers
that declare none, so pebble's lookup is the one resolution path.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The agent launched a sandbox MCP server with its own setsid wrapper and
PID handling, polled ss for the port in a shell loop, and read a fixed log
file for failures; the server listed a sandbox's services with its own ss
and procfs scripts and parsers, and told the API which one it had used.
The driver's services facet now does both: the agent spawns the server as
a service, waits for the port, reads the service's logs on failure, and
stops it on cancellation; the server lists the driver's listening ports,
grouped by port with the process names the sandbox can give. The
discovery source leaves the API and the web panel's iproute2 tip goes
with it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Fabro projected the driver's status into its own state enum, resource,
network, and timestamp types for the run sandbox and inventory endpoints,
losing the provider's state string, the network policy, the sandbox kind,
and the driver's own vocabulary along the way. The API now carries the
driver's SandboxStatus itself: SandboxDetails is fabro's run record beside
the status, SandboxInfo is the provider beside the status, and the
OpenAPI schema describes the driver's types (state, resources in the
units the driver reports, the network policy, the sandbox kind, workspace
ownership) which fabro-api reuses through with_replacement with round
trip tests proving identity and JSON parity. The projection types and
their conversion go; the web sandbox page and summary panel read the
status directly, and the TypeScript client is regenerated.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Both sides rewrote the same crates. This branch replaced fabro's sandbox
layer with the sandbox driver: one RunSandbox, no Sandbox trait, driver
events consumed directly, MockSandbox over the driver's doubles. Main
replaced fabro's LLM layer with lithos-llm: fabro-model deleted, the
catalog and provider ids from lithos, credentials through the lithos
CredentialProvider, clients built with build_client.
Every conflict was one of those two renames meeting in an import list or
a signature, so the rule was mechanical: sandbox names resolve to this
branch, LLM names to main. Where main's newer code still used the old
sandbox API — new session tests over Arc::new(MockSandbox), the SDK
example's LocalSandbox, test fakes typed as Arc<dyn Sandbox> — it is
ported to RunSandbox and the mock helper. Where this branch still used
fabro-model or Client::from_source, main's replacement stands. One
combined future in the CLI runner crossed clippy's size budget and is
boxed at its call.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The OpenAPI spec adopts the lithos request, response, content part,
tool, usage, and cost schemas. The completions endpoint returns the
lithos `Response` JSON verbatim and SSE carries lithos `StreamEvent`s
verbatim. The models and providers endpoints serve the fabro-types
catalog views, and the install and model-test flows probe providers
through fabro-llm.
The CLI builds its catalog from the operator overlay, drives `fabro exec`
through the server gateway adapter, and parses reasoning effort with the
shared controls. The web app reads content parts as lithos-tagged
objects. The TypeScript client is regenerated.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
SandboxProviderKind is now a validated string newtype instead of a
closed enum. The bundled kinds (local, docker, daytona) keep their
constants and a BundledProvider enum for the code paths that still
dispatch on them; any other well-formed sandbox-driver kind name is
accepted and names a plugin executable. EnvironmentProvider is gone:
environment settings carry SandboxProviderKind directly, and
is_clone_based is replaced by a workspace policy where local runs in a
designated directory and every other provider clones.
Server sandbox policy is keyed by kind. [server.sandbox.providers.<kind>]
accepts the bundled kinds with `enabled` and any plugin kind with its
launch settings (path, sha256, dev, args, env, inherit_env); bundled
kinds reject the plugin keys and a kind with no entry is disabled. The
OpenAPI schema, generated Rust and TypeScript clients, web settings
pages, and docs follow. The environments table drops its provider CHECK
enumeration in favour of the kind name rules so a plugin environment
can be stored.
Bundled-only code paths (run start, preflight, reconnect, terminal,
details) now fail with an explicit message for a plugin kind until the
driver construction function lands in the next step.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Collapse the role-paired materializer error variants into
`Credentials`/`Checkout` tagged with a `CheckoutRole`, route both
checkouts through one resolve-then-prepare helper, and replace the
test-only clone-URL field on the production materializer with a
`GitRemote` resolver seam. A workflow source in the target's repository
now reuses the already-resolved credentials instead of minting a second
token.
Also inline the one-line workflow-source normalizer, drop the `as_str`
wrapper on the new kind enum, remove the unused migration constant, move
rather than clone scheduler fields, and deduplicate the web form's
ref-validity rule and per-kind copy into a single table.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Importing them from the environment form component pulled headlessui
Disclosure into the automation form, which the automations-new test mocks
without it.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Share the clone-based provider predicate and provider label between the
automation form and environment settings instead of duplicating them
- Hoist repeated environments query state in the new-automation route
- Normalize empty environment ids to None so validation needs one check
- Merge the scheduler's record/clear error helpers and skip the clearing
write when no error is stored
- Guard the environment backfill with a cheap existence query
- Drop an unneeded id clone and a no-op migrator comment
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Automations now store an environment_id that must reference an enabled
Docker or Daytona environment. Each trigger fire resolves the current
environment definition and snapshots its settings into the run, and
deleting an environment still referenced by an automation is rejected
with a conflict.
Existing automations are backfilled conservatively: a compatible
environment named default is selected when present, otherwise the sole
compatible environment. Anything ambiguous is left incomplete and cannot
run until an operator selects an environment in the web UI.
Scheduler failures are recorded on the automation as last_error and
cleared after the next successful scheduled run.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Materializer derives the manifest GitContext from RunTarget::validate()
instead of hand-building it and re-parsing the repository slug
- Drop parse_github_repository_slug and InvalidRepositorySlug, now unused
- Store reuses Automation::git_target() instead of a private duplicate
- Legacy TOML import returns the target directly rather than a tuple
- Automation target migration updates columns with a single UPDATE ... FROM
- Web: share gitTarget(), targetFromFormValues(), and one SHA validator
across the automation form, list, detail, new, and edit views
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The Repo and Workflow dropdowns on the runs page derived their options
from the board query, which is disabled in list view. With ?view=list,
the options were always empty even when runs were visible.
Derive the options from whichever data source the current view loads:
the board query in columns view, or the current page of the paginated
list query in list view. Extract the option-building into an exported
buildFilterOptions helper that also keeps the active selection in the
options when no loaded run matches it, so the filter button never
renders an undefined label while paginating.
A future change will replace page-derived options with a facets
endpoint plus server-side repo/workflow query params.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The playground's local manifest projection and builder now emit only
the resolved target path; workflow-name fallback, titles, and the
workflows-map key are unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Resolve the model catalog table conflict in docs/public/core-concepts/models.mdx
by keeping both changes: this branch's `kimi` -> `moonshot` provider rename for
the Kimi rows, and main's new DeepSeek V4 rows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Path safety now lives in one place. The NUL-byte and drive-letter rules
move from a server-only helper into the store's own filename validation,
so uploads reject those paths at write time instead of only the ZIP read
path catching them. The download still re-checks, because artifacts
stored before the rule existed can still carry an unsafe path, but it
now skips a bad path rather than failing the whole archive.
Promote is_boundary_stage to RunProjection and drop the three identical
private copies. The ZIP download used a node-name match instead, which
would have dropped artifacts from a working node that happened to be
named "start".
Compress the archive. Entries were Stored while the response was also
excluded from transfer compression, so text artifacts moved at full
size. async_zip gains the deflate feature; async-compression and flate2
were already in the lock file.
Log archive failures unconditionally. The send-succeeded guard meant a
client that had already disconnected left no record at all, which is the
case where the log is the only evidence.
Also: collapse the duplicate 500 arms, drop the dead stage-ID tiebreaker
and the cached order in the selection map, name the accessible label
after the visible one, share the run URL prefix between the two download
href builders, and document the mid-stream truncation behavior in the
OpenAPI description.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The parallel stage summary rendered a Duration tile directly below
StageMetaBar, which already shows the same stage's duration with a live
ticking clock and a started-at tooltip. The two disagreed while running:
the meta bar counted up, the tile showed the static word "running". The
cancelled-stage bug lived only in the duplicate.
Drop the tile. The meta bar owns duration for every stage renderer, and
it was already correct for cancelled, pending and skipped stages. That
removes the three-way duration branch, the "--" sentinel decode, and the
ACTIVE_STAGE_STATES and formatDurationMs imports.
With the tile gone, ParallelOverview.durationMs is dead, as were
successCount, failureCount and isComplete — the renderer counts the
branch rows it draws. ParallelOverview reduces to branch identity.
For run event write failures, log the first at error with run_id and
event name, the rest at debug, and summarize new losses at flush. A
broken sink fails for every event, so a bare error would emit one
"investigate me" line per event for the life of the run.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The steering dock on run pages now starts collapsed, staying out of
the way until the operator opens it. A run that is interrupted and
waiting for steering still forces the dock open.
While collapsed, the whole dock header is a click target that expands
it. Clicks on buttons in the bar (Interrupt, the chevron) keep their
own behavior, and the chevron remains the keyboard/assistive-tech
toggle. The interview dock shares the shell, so it gets the same
click-to-open behavior.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Simplification pass over the for_each branch. No behavior change.
Share what was duplicated:
- Node::prompt_or_label replaces the "prompt, else label" fallback that
agent, prompt, and the for_each item injector each wrote out.
- context::lookup_flat replaces the "exact key, then strip context."
lookup that condition.rs had twice and the for_each source had again.
- is_llm_handler_type replaces the inline agent/prompt match, so the
runtime and the for_each_contract rule agree by construction.
- find_join_node now takes branch ids, so a for_each fan-out passes its
template target instead of needing find_join_for_target.
- collect_events moves to test_support; parallel and integration tests
shared one copy already.
- One ScriptedHandler replaces four test handlers that differed only in
what they returned.
Straighten the branch retry loop:
- Reserve the branch scope once before the loop instead of guarding it
with an Option, which removes three expect() calls.
- acquire_branch_permit and backoff_or_cancel replace the cancel-aware
select! blocks the loop repeated verbatim.
- Keep the match arms in Executor::execute_with_retry order so the two
loops stay easy to compare.
Drop redundant state:
- BranchPlan::is_for_each derives from template_target_id.
- for_each_contract checks node type before source shape, so one
mistake reports one diagnostic.
Prove the new wire fields survive the OpenAPI boundary: the fabro-api
round-trip fixture now carries index and item_label.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Conflicts were between this branch's parallel-branch identity work and
main's stage billing, review targets, and live stage timing.
- Stage fixtures: main added `billing` to each per-file `makeStage`; this
branch had hoisted one builder into `lib/test-utils`. Kept the hoisted
builder and gave it `billing: makeBilledTokenCounts()`, so both intents
hold and the field list stays in one place. `stage-sidebar.test.ts` also
builds raw `RunStage` wire payloads, so it keeps importing
`makeBilledTokenCounts` directly.
- Import lists (`run_projection.rs`, `fabro-api/src/lib.rs`,
`run_state.rs`, `stage_projection_round_trip.rs`): unioned both sides —
`ParallelBranchId` alongside `timing`, `ReviewTarget`,
`ReviewTargetKind`, `AttrValue`, `Node`, and
`StageToolBatchProjection`.
- `fabro-server` tests: git interleaved two unrelated new tests into one
body. Split them back into
`list_run_stages_exposes_parallel_branch_identity` and
`run_billing_includes_live_stage_timing_in_rows_and_totals`.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Branch indexes are sparse. A branch queued behind max_parallel reserves
no stage identity until it acquires the semaphore, tokio task order is
not index order, and a branch cancelled while queued never reserves one
at all. Sizing the row list by `stagesByBranchIndex.size` treated an
entry count as a dense index range, so a running branch at index 2 with
nothing at 0 or 1 rendered as a single "pending" placeholder and the
running branch disappeared. Size from the highest index observed.
Also:
- Derive the Succeeded/Failed tiles from the rendered rows instead of the
completed-event rollup, so the tiles cannot contradict the list. This
drops the isComplete fork and both live counters.
- Show the fallback branch count in the Branches tile, which previously
read "-" above N rows in exactly the case the fallback exists for.
- BranchRow carries `label` and `stageId`; ChildRow owns the route it
links to. `id` had become a display label on one path and a raw node id
on the other, and a view model should not hold a URL.
- Drop `branchIndex`, which only ever served as the React key and always
equalled the array index.
- Cover the sparse-index and pending-placeholder paths, and derive the
rollup counts in `completedEvent` instead of passing contradictory ones.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Branches bypass the engine's stage.started/stage.completed lifecycle, so
no SWR key invalidated the stages list while a fork ran. The new live
branch rows stayed frozen at their first observed state until an
incidental refetch. Map parallel.* events to the stages list, run events,
and graph keys.
Also:
- Label branch rows with formatStageLabel so a re-entered branch renders
as `review_glm@2`, matching the sidebar and waterfall.
- Build branch rows in one pass and count live outcomes in one loop.
- Name ParallelBranchId in the OpenAPI spec and reuse fabro_types::
ParallelBranchId, replacing two copies of an inline string format.
- Hoist makeStage and textContent into lib/test-utils so widening Stage
cannot leave per-file fixtures stale (tests are excluded from
typecheck, so the two component-test copies had already gone stale).
- Query stat tiles by data-stat instead of an exact Tailwind class.
- Reuse append_scoped_stage_event's body via append_event_with_scope and
add test_branch_event instead of poking envelope fields.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Conflict in apps/fabro-web/app/components/interview-dock.tsx. Main moved
the dock onto the shared collapsible `RunDockShell` and replaced the
local button constants with shared ones.
Kept main's structure whole and re-applied the review target rendering
onto it: the question paragraph in the shell's `body` becomes the linked
`ReviewTargetQuestion` when the target passes `safeReviewTarget`, and
plain text otherwise. Both now share main's paragraph classes through
`QUESTION_TEXT`, so the two renderings stay visually identical.
`peek` keeps using `question.text`, which is the correct plain-text
collapsed summary for a review target question.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The review question sentence was written in four places and the URL
safety rules in three. Collapse each to one definition.
- Add `ReviewTarget::question_text_with_link` as the single definition of
the question wording. `question_text()` and the Slack header both use
it, so a wording change is now one edit.
- Delete `ReviewTargetKind::noun()`. The enum already derives
`strum::Display` with the same snake_case output.
- Share one `review_target_line` helper between the console interviewer
and the CLI attach client, which held a byte-identical copy. Print only
the URL: `question.text` already carries the label and the noun.
- Trim the web-side check to the URL scheme, host, and credentials, which
are what a raw `href` can act on. Label length and control characters
cannot affect the DOM and stay server-side.
- Split validation from presentation in the web UI. `safeReviewTarget`
returns the target or null, and each caller picks its own fallback, so
an unsafe target now falls back to the same Markdown rendering as a
question with no target.
- Derive the resource noun from `kind` in the web UI instead of
hardcoding "document".
- Use `ReviewTargetKind.DOCUMENT` and the shared `isRecord` guard when
parsing events, instead of a raw string and a hand-rolled object check
that accepted arrays.
- Drop `deny_unknown_fields` from the wire struct. The OpenAPI schema
leaves `additionalProperties` permissive, so an added field would
otherwise make persisted events unreadable.
- Import `ReviewTarget` by name, and stop naming Slack in a fabro-types
error message.
- Document that `review_target=true` replaces the gate's `label`.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
