Files
archie/internal/app/agentworker/task_execution.go
T
sam 4750ac097b feat(workflow): post line-anchored review comments with suggested fixes (#891)
* feat(workflow): post line-anchored review comments with suggested fixes

Bead archie-core-q9au (docs/prds/inline-review.md). A review's findings reached
the operator only as a flat list in the PR body, so the author had to map every
finding back to its line by hand and apply mechanical fixes by hand. Now the
line-anchored findings are also posted as inline review comments, carrying a
one-click suggestion when the reviewer could state the fix.

Domain (internal/domain/workflow):
- ReviewFinding gains Suggestion; Validate rejects one on anything but a
  confirmed finding, so the agent's record_finding call is refused with
  something it can act on. Validate splits into validateAnchor /
  validateClassification / validateDisposition, which also brings it back under
  the cyclop limit the gate enforces.
- StagePostReviewComments runs after StageOpenPR in the implement workflow. It
  posts only line-anchored findings (Line > 0); a whole-file finding has no line
  to attach to and stays in the body list, which remains the complete record.
  It is best-effort: the PR is open and the body already lists every finding, so
  a failed comment is logged, never a park -- the same reasoning as OpenPR's
  existing best-effort LinkBranch call.
- The suggestion fence is gated on Verdict == confirmed where the fence is
  written, because it is a one-click apply: a wrong suggestion breaks the code
  and the author has to notice.

Forge (internal/forge): ReviewCommentWriter + GitHub/Gitea implementations,
type-asserted like PullRequestReader so a forge that cannot is skipped rather
than faked (NoopForge is pinned as not satisfying it). GitHub loops one
CreateComment per comment and joins the failures, so one refused anchor cannot
drop the rest; Gitea carries them on a single COMMENT-state review.

Two deliberate departures from the PRD's sketch, both recorded in it:
- The interface is batch (CreateReviewComments), not one call per comment:
  Gitea's only inline shape is one submitted review, and the set then crosses
  NATS once instead of once per finding.
- The head SHA is resolved inside each forge implementation rather than by the
  caller. The stage runs in archie-agent, which holds no forge credentials and
  has no PullRequestReader; the daemon-side forgerpc handler is the only place
  that does.

Transport (internal/forgerpc): workflow.Forger gains the method, so the agent
reaches the daemon over the new create_review_comments subject; the server
type-asserts the writer and answers an incapable forge with an error the stage
logs. This is the one proxied method whose absence on an image skew does not
park a task.

The reviewer prompt and record_finding schema ask for a suggestion on a
confirmed, mechanically fixable defect only.

Regenerating the wfextract symbol table -- the documented way a new exported
workflow symbol reaches interpreted stage plugins -- also picked up symbols that
were missing from the committed table (the Status* and Source* constants,
ReviewCheck, Store, Task, and the _Store wrapper). Nil-guards were re-applied to
every wrapper afterwards, as that file's header requires.

Evidence: task check exit 0 (incl. 229 UI tests). Three mutations of the new
stage logic each fail a test that names the behaviour -- dropping the
confirmed-only gate, dropping the Line > 0 filter, and letting a posting failure
escape -- so the tests discriminate rather than passing vacuously.

* fix(build): drop the Dockerfile pin assertion

The guard fetched golangci-lint's go.mod at image build time to fail the build
when GOLANGCI_LINT_VERSION and GOFUMPT_VERSION disagreed. It only ever tripped
on self-inflicted pin drift, but it added a network fetch and a new failure
mode to the release build, which is not where that belongs.

The pins stay correct (golangci-lint 2.13.2, whose vendored gofumpt v0.11.0
equals GOFUMPT_VERSION) and the comment above them still states the invariant,
so the reasoning survives without the build-time machinery. The pin change
itself reaches the agent image through the normal release path: CI rebuilds
and pushes archie-agent on an archie/vX.Y.Z release, and a gateway-only
release deliberately does not.

* feat(workflow): add the remediate workflow for PR review remediation

Implements step 5 of docs/prds/pr-review-remediation.md: a remediate
workflow that resumes the task's existing worktree onto its PR branch
(worktree.Manager.Resume, not Prepare's refresh), runs a builder agent
against the review's actionable comments, commits and pushes to the
same branch, and replies via the forge's ReplyToReview. Task.RetryCount
doubles as the remediation round counter, capped by the repo's
max_retries; the round-cap stage parks the task and posts one comment
when it's hit.

Task gains a ReviewPayload field (JSON-encoded review unit) with full
store/proto/staterpc plumbing. workflow.Forger gains Comment and
ReplyToReview, proxied to archie-agent via a new forgerpc RPC. Resume
and its Dir lookup are added as an unexported optional Trees capability
rather than widening the exported interface, which is already at
golangci-lint's interfacebloat cap and is projected into wfextract's
Yaegi symbol table for repository-authored stage plugins.

The daemon-side reaction consumer that dispatches into this workflow
(subscribing ARCHIE_REACTIONS, grouping comments by review, enforcing
per-task serialization) is tracked separately in archie-core-8li9.4.

* feat(messaging): extract standalone messaging service binary and process composition

* feat(messaging): adapt channel frontends to consume Gateway ChatContract

* feat(messaging): remove in-process channels and Router from daemon bootstrap

Telegram, email and webhook now run only in the Messaging Service, which
reaches this process through the Gateway's ChatContract. The daemon builds
no channel and no gateway.Router, so the wiring that existed only to serve
them goes with it: setupGateways, setupEmailGateway, setupWebhookGateway,
setupTelegramGateway, buildTelegramRouter, makeTelegramReload and the
release-announcement path helper.

The seams those paths owned move to where they can still be honoured:

- Channel ValidateConfig runs in archiemessaging.compose, which now returns
  an error, so a configured-but-invalid front-end fails the service rather
  than being dropped silently.
- The inbound rate limiter moves to the Gateway process, which owns the only
  Router and therefore the only place a per-sender budget applies to every
  channel's turns.
- The daemon's gateway readiness probe becomes a ContractProbe against the
  Gateway. It cannot observe a channel lifecycle it no longer runs, and a
  channel-count probe would have reported every daemon degraded.
- telegramSetup is now chatSetup (in chat_setup.go), trimmed to the fields a
  chat turn runner actually reads.

Telegram operator features that were wired only in the daemon (/update,
release announcements, version reporting, SIGHUP reload) are not yet wired
in the Messaging Service: archie-core-8cda.6.6. Channel lifecycle reporting:
archie-core-8cda.6.8. The Gateway process builds no worktree manager, so
review_pr is unadvertised on every channel: archie-core-8cda.6.7.

* feat(messaging): re-wire the Telegram operator surface in the Messaging Service

setupTelegramGateway satisfied eight telegram.Gateway seams from the daemon's
own process. Each is resolved by where the fact it needs is owned, not by
where it used to be built; the decision table is in
docs/architecture/migration-decisions.md.

Wired here: /update (built from this service's own [chat.telegram] check and
install commands), the update-report relay (same sha256(bot_user)[:8] path the
daemon wrote, so a pending report is found rather than written afresh),
show_tool_calls, and /restart's config reload, which re-resolves the token and
allowlist from this service's own file and refuses a reload that resolves to
an empty token.

/version is sourced from the Gateway: ChatSnapshot gains a Version field
carrying the Gateway's rendered build block, read per call and bounded by the
dependency timeout. cmd/archie-messaging deliberately takes no version
ldflags, so it cannot report a partial upgrade as a matched one.

RunningVersions and ReleaseAnnouncements stay nil. Both need a component's
real running version, and the only versions this process can obtain are a
different binary's stamps -- reporting them would manufacture exactly the
false success the update check exists to catch, and the announcer without
parseable versions is a no-op that reads as configured. archie-core-8cda.6.9
carries the contract that unblocks them. Dangerous stays nil: no composition
has ever supplied a command authority.

Enrich is dropped from the update service: componentInstallTypeEnricher read
[nats], which this service must not decode, so the check command's own
reported install type stands.

* fix(workflow): guard inline review comments against head drift

Two findings from the review of q9au.

A comment's line numbers describe exactly one revision of the branch, but the
forge resolved the pull request's head at posting time -- after the review had
already read the worktree. A push in that window (a collaborator, a retry) meant
the findings were anchored to whatever now occupied those lines, mislabelling
instead of failing. The revision the review measured now travels with the
comment set: OpenPR records it from the same worktree read it already performs
(TaskContext.ReviewedHeadSHA), StagePostReviewComments threads it through
Forger to forgerpc, and the forge refuses the whole set once the head it reads
has moved off it -- which the best-effort stage logs, the PR body remaining the
record. An unmeasured revision posts unverified rather than dropping every
finding, so a worktree that cannot report one degrades instead of regressing.

Suggestion was documented single-line only, because GitHub applies a suggestion
fence up to its first line and a multi-line replacement would silently corrupt
the file. Only the verdict was validated, so a multi-line suggestion reached the
renderer and posted as that truncated fence. Validate now rejects an embedded
newline, which makes record_finding refuse it with feedback the model can act on
instead of the reviewer discovering the corruption later.
2026-09-19 23:37:23 +10:00

320 lines
12 KiB
Go

package agentworker
import (
"context"
"errors"
"fmt"
"io/fs"
"log/slog"
"os"
"path/filepath"
"strconv"
"github.com/samcharles93/archie-core/internal/agentexec"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/workflow"
"github.com/samcharles93/archie-core/internal/domain/workflow/skillbuild"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
"github.com/samcharles93/archie-core/internal/domain/workflow/wfeval"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/installtype"
"github.com/samcharles93/archie-core/internal/storage"
"github.com/samcharles93/archie-core/internal/taskrun"
"github.com/samcharles93/archie-core/internal/worktree"
)
// hybridTrees implements workflow.Trees by splitting operations: Prepare
// and Push (network ops needing the daemon's forge credential) proxy over
// worktreerpc; CommitAll/Diff/ChangedFiles/ChangedLines are local git
// operations run directly against the container's bind-mounted worktree.
//
// Prepare returns the bind-mounted worktree that archied prepared before the
// container started. The sandbox has no remote prepare capability.
type remoteTrees interface {
Push(ctx context.Context) error
}
type hybridTrees struct {
push remoteTrees
local *worktree.Manager
localDir string
branch string
// worktreeUID and worktreeGID are the daemon's own host UID/GID
// (WORKTREE_UID/WORKTREE_GID, set by containerEnv), or -1 when unset.
// The agent runs as root inside the container, so a commit it writes
// to the bind-mounted worktree lands owned by UID 0 on the host; Push
// reconciles ownership back to the daemon's UID before handing off,
// since the daemon -- running as its own non-root host user -- is the
// one that reads those objects to actually push (archie-core#520).
worktreeUID, worktreeGID int
}
func (h *hybridTrees) Prepare(ctx context.Context, owner, repo, base string, issue int, title, body, labels string) (dir, branch string, err error) {
return h.localDir, h.branch, nil
}
func (h *hybridTrees) CommitAll(ctx context.Context, dir, message string) (bool, error) {
changed, commitErr := h.local.CommitAll(ctx, dir, message)
if ownershipErr := h.reconcileOwnership(dir); ownershipErr != nil {
ownershipErr = fmt.Errorf("reconcile worktree ownership after commit: %w", ownershipErr)
return changed, errors.Join(commitErr, ownershipErr)
}
return changed, commitErr
}
func (h *hybridTrees) Push(ctx context.Context, dir, branch string) error {
if err := h.reconcileOwnership(dir); err != nil {
return fmt.Errorf("reconcile worktree ownership before push: %w", err)
}
return h.push.Push(ctx)
}
func (h *hybridTrees) reconcileOwnership(dir string) error {
if h.worktreeUID < 0 || h.worktreeGID < 0 {
return nil
}
return chownTree(dir, h.worktreeUID, h.worktreeGID)
}
// restoreWorktreeOwnership returns the deferred half of the ownership
// contract: the agent runs as root inside the container, so a run that ends
// without ever reaching CommitAll or Push (a gate failure that parks the task,
// a stage error, an early return) has to hand the bind-mounted worktree back to
// the daemon's own UID on its way out. The daemon is the process that later
// cleans and resets that directory as a non-root host user, and it cannot even
// unlink inside a root-owned directory -- so a parked attempt otherwise left a
// worktree the next attempt could not start from, no matter what the refresh
// did.
//
// Returns the func rather than deferring internally so the caller's own defer
// runs it at the caller's return, not at this function's.
func restoreWorktreeOwnership(trees *hybridTrees, dir string, log *slog.Logger) func() {
return func() {
if err := trees.reconcileOwnership(dir); err != nil {
log.Warn("worktree ownership restore failed", "dir", dir, "err", err)
}
}
}
func (h *hybridTrees) Diff(ctx context.Context, dir, base string) (string, error) {
return h.local.Diff(ctx, dir, base)
}
func (h *hybridTrees) ChangedFiles(ctx context.Context, dir, base string) ([]string, error) {
return h.local.ChangedFiles(ctx, dir, base)
}
func (h *hybridTrees) Snapshot(ctx context.Context, dir, destDir string) error {
return h.local.Snapshot(ctx, dir, destDir)
}
// Resume is a no-op here for the same reason Prepare is: resuming a PR
// branch needs the daemon's forge credential to fetch, so archied must run
// it against the bind-mounted worktree before the container starts, the
// same way it runs Prepare today. The container's worktree is already
// resumed by the time a remediate run reaches this stage.
func (h *hybridTrees) Resume(ctx context.Context, dir, branch string) error {
return nil
}
// Dir returns the bind-mounted worktree path archied already prepared,
// ignoring its arguments -- the sandbox has no way to independently derive
// or verify the path, same as Prepare above.
func (h *hybridTrees) Dir(owner, repo string, issue int) string {
return h.localDir
}
func (h *hybridTrees) ChangedLines(ctx context.Context, dir, base string) (int, error) {
return h.local.ChangedLines(ctx, dir, base)
}
// ChangedFileStats is the capability the workflow captures a change through.
// It is not part of workflow.Trees: it is asserted for as the unexported
// optional interface package workflow declares, so a Trees implementation
// without it degrades to no capture. Without this forwarder the in-container
// path -- which is every production run, archie-agent executing the whole
// workflow -- would be exactly that silent no-op.
func (h *hybridTrees) ChangedFileStats(ctx context.Context, dir, base string) (task.ChangeStats, error) {
return h.local.ChangedFileStats(ctx, dir, base)
}
var _ workflow.Trees = (*hybridTrees)(nil)
// chownTree recursively chowns dir to uid:gid so the daemon -- running as
// its own host user -- can read loose objects the agent committed as root
// inside the container.
//
// Lchown, not Chown: a repo can legitimately track a symlink whose target
// doesn't exist inside the container (a relative link outside the
// checkout, or to a tool the image doesn't ship). Chown follows the link
// and fails on a dangling target, which would abort a push that would
// otherwise have succeeded -- worse than the permission error this is
// fixing. Lchown changes the link itself and never touches the target.
func chownTree(dir string, uid, gid int) error {
return filepath.WalkDir(dir, func(path string, _ fs.DirEntry, err error) error {
if err != nil {
return err
}
return os.Lchown(path, uid, gid)
})
}
// worktreeOwnerID parses a WORKTREE_UID/WORKTREE_GID environment value,
// returning -1 (unset/unconfigured) when it is empty or not a valid
// non-negative integer.
func worktreeOwnerID(env string) int {
if env == "" {
return -1
}
id, err := strconv.Atoi(env)
if err != nil || id < 0 {
return -1
}
return id
}
type taskDependencies struct {
forge workflow.Forger
store workflow.Store
trees remoteTrees
events agentexec.EventPublisher
}
type runnerFactory func(map[string]agentexec.Provider, *slog.Logger) agentexec.Runner
func newTaskRunner(providers map[string]agentexec.Provider, log *slog.Logger) agentexec.Runner {
return agentexec.NewLoopRunner(agentexec.NewRuntime(providers), log)
}
// applyToolLimits wires the task's carried tool policy into a *LoopRunner's
// result cap/spill, so a worker-executed stage enforces the same limits the
// daemon's own chat path applies (config.Config.Tools.Policy, carried
// non-secret via TaskConfig.ToolPolicy). A runner that isn't a *LoopRunner
// (e.g. a test fake) is left untouched.
func applyToolLimits(agent agentexec.Runner, policy config.ToolPolicy) {
runner, ok := agent.(*agentexec.LoopRunner)
if !ok {
return
}
runner.Limits = agentexec.ToolLimits{
MaxResultChars: policy.MaxResultChars,
SpillDir: policy.SpillDir,
}
}
// routeTask applies the request-carried routing bindings and selects the
// workflow for a task. The daemon's package-level SetKindWorkflows state
// never crosses the process boundary, so runTask installs the resolved
// bindings here -- this worker process is where workflow.Route actually runs.
func routeTask(req taskrun.Request, registry workflow.Registry) workflow.Workflow {
workflow.SetKindWorkflows(req.KindWorkflows)
workflow.SetLabelWorkflows(req.LabelWorkflows)
return workflow.Route(req.Task, registry)
}
// runTask builds a workflow.Registry from the container's mounted worktree,
// routes and runs the entire workflow, and reports its terminal outcome.
// Store remains archied's authority; Response.Task is a logging snapshot.
func runTask(ctx context.Context, req taskrun.Request, dependencies taskDependencies, newRunner runnerFactory, workDir string, log *slog.Logger) (*taskrun.Response, error) {
registry, err := skillbuild.BuildRegistry(workDir)
if err != nil {
return nil, fmt.Errorf("build registry: %w", err)
}
wf := routeTask(req, registry)
trees := &hybridTrees{
push: dependencies.trees,
local: &worktree.Manager{
WorkDir: workDir,
BotUser: req.Cfg.BotUser,
BotEmail: req.Cfg.BotEmail,
},
localDir: workDir,
branch: req.Task.Branch,
worktreeUID: worktreeOwnerID(os.Getenv("WORKTREE_UID")),
worktreeGID: worktreeOwnerID(os.Getenv("WORKTREE_GID")),
}
// The agent process runs as root inside the container, so everything this
// run writes into the bind-mounted worktree is owned by UID 0 on the host;
// CommitAll and Push hand it back on the success path (archie-core#520).
// Doing the same here covers every OTHER way a run can end -- a gate
// failure that parks the task, a stage error, an early return -- because
// the daemon is the one that later cleans and resets this directory, as its
// own non-root host user, and it cannot even unlink a root-owned directory.
// Without this, a parked attempt left a worktree the next attempt could not
// start from.
defer restoreWorktreeOwnership(trees, workDir, log)()
// Start MCP providers and build a local tool registry.
mcpSet, mcpErr := startMCPProviders(ctx, req.MCPServers, log)
if mcpSet != nil {
defer mcpSet.cleanup(ctx, log)
}
if mcpErr != nil {
log.Warn("mcp providers had errors, continuing with available tools", "err", mcpErr)
}
// Build a runner with MCP-discovered tools when available.
var agent agentexec.Runner
if mcpSet != nil && mcpSet.registry != nil {
agent = agentexec.NewLoopRunner(agentexec.NewRuntime(req.Providers), log, mcpSet.registry)
} else {
agent = newRunner(req.Providers, log)
}
if agent == nil {
return nil, fmt.Errorf("no agent runner configured for task %d", req.Task.ID)
}
applyToolLimits(agent, req.Cfg.ToolPolicy)
agent = persistentRunner{Runner: agent, enabled: req.Repo.PersistentStorage}
// A workflow run in this process publishes to an in-process *events.Bus
// the daemon cannot see -- archied and archie-agent are separate
// processes connected only by NATS. Without this bridge, tc.Emit is a
// silent no-op for every stage/outcome/park event this run produces,
// which is why the dashboard timeline showed nothing for any task
// executed through the container/NATS path (archie-core-518). Nil
// dependencies.events (a caller with no NATS connection, e.g. tests
// that construct taskDependencies directly) leaves bus nil, and
// TaskContext.Emit is already nil-safe.
var bus *events.Bus
if dependencies.events != nil {
bus = events.NewBus()
sub := bus.Subscribe(64)
defer sub.Close()
go agentexec.ForwardTaskEvents(sub, dependencies.events, req.Task.ID, log)
}
tc := &workflow.TaskContext{
Task: req.Task,
Repo: req.Repo,
Cfg: req.Cfg.ToConfig(),
Forge: dependencies.forge,
Store: dependencies.store,
Trees: trees,
Agent: agent,
Reviewer: newReviewerFor(req),
Bus: bus,
Log: log,
CustomStages: wfeval.Discover,
}
if req.Repo.PersistentStorage {
memory, readErr := readProjectMemory(storage.MemoryPath)
if readErr != nil {
return nil, fmt.Errorf("read project memory: %w", readErr)
}
if memory != "" {
tc.SystemPrompt = func() string { return memory }
}
}
workflow.Run(ctx, wf, tc)
return &taskrun.Response{
Task: tc.Task,
Status: tc.Outcome.Status,
AgentVersion: Version(),
AgentInstallType: installtype.Type(),
}, nil
}