fix(loop): resume historical active targets

Author: codeacme17

loops/issue-dev-loop/SKILL.md

@@ -7,7 +7,7 @@ Run exactly one bounded issue cycle. Treat [`LOOP.md`](./LOOP.md) as the constit
 ## Start safely
 
 1. Read `LOOP.md`, `state.md`, and `dependencies.md` completely.
-2. Require absolute `ECHO_UI_LOOP_CONTROL_PLANE` and `ECHO_UI_LOOP_TARGET_ROOT` values from the scheduler. Run activation through `"$ECHO_UI_LOOP_CONTROL_PLANE/scripts/with-github-identity" --loop-root "$ECHO_UI_LOOP_TARGET_ROOT" automation -- node "$ECHO_UI_LOOP_CONTROL_PLANE/scripts/loopctl.mjs" validate --activation --loop-root "$ECHO_UI_LOOP_TARGET_ROOT"`. The installed launcher verifies its hash manifest and probes both configured profiles before starting validation.
+2. Require absolute `ECHO_UI_LOOP_CONTROL_PLANE` and `ECHO_UI_LOOP_TARGET_ROOT` values from the scheduler. Run activation through `"$ECHO_UI_LOOP_CONTROL_PLANE/scripts/with-github-identity" --loop-root "$ECHO_UI_LOOP_TARGET_ROOT" automation -- node "$ECHO_UI_LOOP_CONTROL_PLANE/scripts/loopctl.mjs" validate --activation --loop-root "$ECHO_UI_LOOP_TARGET_ROOT"`. The installed launcher verifies its hash manifest and probes both configured profiles before starting validation. When the target contains a durable active run created by an older owner-merged control plane, the launcher alone selects historical-target compatibility validation: installed runtime files remain hash-verified in the trusted bundle, while the target is checked for its stable state files, owner channel, JSON history, and evidence workflow. Callers cannot request this reduced target-only mode directly, and `restore-checkpoint` still requires the durable exact branch, head, and clean worktree.
 3. Read [`references/github-operations.md`](./references/github-operations.md). Run every operational `loopctl`, executor GitHub command, remote Git command, trigger, and reviewer publication through the installed control plane with the explicit target root. Never use the credential-refusing repository launcher, invoke the `.mjs` router directly, install control code from an issue branch, or alter global `gh` or Git credential configuration.
 4. Run `loopctl.mjs reconcile` through the automation wrapper to rebuild verified terminal history, pending/completed evolve state, and active runs from the append-only GitHub state journal. It tombstones local terminal-cache rows with no durable counterpart before recomputing metrics. For returned `workType: resume`, fetch the recorded branch through the wrapper, create a clean isolated worktree at the returned exact head, and run `restore-checkpoint --run-id <id>` inside that worktree. The restore command rejects the wrong branch, a dirty checkout, or any head other than the durable head. Resume it before selecting a new issue.
 5. Run `loopctl.mjs evolve-status`. If `evolveDue` is true, start `echo_ui_loop_evolver` with fresh context; do not silently replace it with product work.

loops/issue-dev-loop/references/github-operations.md

@@ -10,6 +10,8 @@ Run every executor GitHub command through:
 
 `ECHO_UI_LOOP_CONTROL_PLANE` must name the versioned installation created from a clean owner-merged `dev`; `ECHO_UI_LOOP_TARGET_ROOT` names the active worktree's `loops/issue-dev-loop`. Operational `loopctl` and trigger commands must use the scripts inside the installed root and pass `--loop-root "$ECHO_UI_LOOP_TARGET_ROOT"`. The repository launcher intentionally refuses credentials.
 
+For a durable active run whose target predates newer trusted runtime files, only the installed activation router may select historical-target compatibility validation. It continues to hash-verify the complete installed control plane and both identities, then validates the target's stable state files, owner channel, JSON history, and evidence workflow. A caller-supplied compatibility flag is rejected; the subsequent restore still enforces the journal's exact branch and head on a clean worktree.
+
 Run every reviewer publication command through the installed wrapper with role `reviewer`. Before reading either profile, it verifies every installed file, pins absolute Node/Git/`gh` executables, compares the target's security-critical owner-channel values to its trusted copy, removes token environment overrides, runs `gh api user`, and refuses an unexpected or owner identity. For Git, it clears global credential helpers and injects `gh auth git-credential` for the entire trusted child tree. Descendant `git` and `gh` processes pass through a role gate; arbitrary `sh`, `env`, Node scripts, caller PATH shims, and issue-worktree router changes are not authenticated. Never use owner credentials for executor or reviewer actions, never run raw remote `gh`/`git push` commands, and never call `gh auth setup-git`.
 
 Publish reviewer output only as a non-approving comment review:

loops/issue-dev-loop/scripts/lib/github-identity.mjs

@@ -240,10 +240,7 @@ export function hardenedGitArguments(args, { expectedRepository = null } = {}) {
     const rollback = lease?.match(
       /^--force-with-lease=(refs\/heads\/codex\/issue-[1-9][0-9]*):([0-9a-f]{40})$/,
     )
-    if (
-      rollback &&
-      sameArguments(args, ['push', lease, 'origin', `:${rollback[1]}`])
-    ) {
+    if (rollback && sameArguments(args, ['push', lease, 'origin', `:${rollback[1]}`])) {
       return ['push', lease, repositoryUrl, deleteRef]
     }
     const branch = args.at(-1)
@@ -783,9 +780,7 @@ function pullRequestMutationAllowed(kind, args, commandIndex, authorization, exp
   }
   if (kind === 'ready') {
     return (
-      parsed.values.size <= 1 &&
-      parsed.booleans.size === 1 &&
-      parsed.booleans.get('undo') === true
+      parsed.values.size <= 1 && parsed.booleans.size === 1 && parsed.booleans.get('undo') === true
     )
   }
   if (kind === 'comment') {
@@ -812,9 +807,7 @@ function reservedAutomationComment(body) {
 
 function checkpointPublicationAllowed(body, authorization) {
   const markers = [
-    ...body.matchAll(
-      /<!-- issue-dev-loop:checkpoint:([^:]+):sha256:([0-9a-f]{64}) -->/g,
-    ),
+    ...body.matchAll(/<!-- issue-dev-loop:checkpoint:([^:]+):sha256:([0-9a-f]{64}) -->/g),
   ]
   if (markers.length !== 1 || markers[0][1] !== authorization?.issue?.runId) {
     return false
@@ -1082,6 +1075,9 @@ function assertRootCommandPolicy({ role, tool, args, loopRoot, trustedLoopRoot,
       path.resolve(trustedLoopRoot, 'scripts', 'loopctl.mjs'),
       path.resolve(trustedLoopRoot, 'triggers', 'detect-work.mjs'),
     ])
+    if (args.includes('--target-compatibility')) {
+      throw new Error('target compatibility validation is reserved to wrapped activation')
+    }
     if (
       script &&
       allowedScripts.has(script) &&
@@ -1094,8 +1090,7 @@ function assertRootCommandPolicy({ role, tool, args, loopRoot, trustedLoopRoot,
   if (
     role === 'reviewer' &&
     tool === 'node' &&
-    path.resolve(args[0] ?? '') ===
-      path.resolve(trustedLoopRoot, 'scripts', 'publish-review.mjs')
+    path.resolve(args[0] ?? '') === path.resolve(trustedLoopRoot, 'scripts', 'publish-review.mjs')
   ) {
     parseReviewPublisherArguments(args.slice(1), {
       authorization,
@@ -1301,16 +1296,11 @@ function argumentAfter(args, name) {
 function withRootCommandIntent(authorization, { tool, args, trustedLoopRoot }) {
   const script = args[0] ? path.resolve(args[0]) : null
   const isTrustedLoopctl =
-    tool === 'node' &&
-    script === path.resolve(trustedLoopRoot, 'scripts', 'loopctl.mjs')
+    tool === 'node' && script === path.resolve(trustedLoopRoot, 'scripts', 'loopctl.mjs')
   const withIntent = isTrustedLoopctl
     ? { ...authorization, rootIntent: args[1] ?? null }
     : authorization
-  if (
-    !isTrustedLoopctl ||
-    args[1] !== 'start' ||
-    authorization.issue !== null
-  ) {
+  if (!isTrustedLoopctl || args[1] !== 'start' || authorization.issue !== null) {
     return withIntent
   }
   const issueNumber = Number(argumentAfter(args, '--issue'))
@@ -1478,10 +1468,7 @@ async function preflightPullRequestWrite({
   }
 
   if (['review', 'inline-review'].includes(intent.kind)) {
-    if (
-      livePullRequest.draft !== true ||
-      !intent.publication
-    ) {
+    if (livePullRequest.draft !== true || !intent.publication) {
       throw new Error('independent review publication requires the recorded Draft PR')
     }
     const publishedReviews = await githubPaginatedApi(
@@ -1499,9 +1486,7 @@ async function preflightPullRequestWrite({
             review.body?.includes(intent.publication.marker),
         )
       ) {
-        throw new Error(
-          `adjudication for ${intent.publication.findingId} is already published`,
-        )
+        throw new Error(`adjudication for ${intent.publication.findingId} is already published`)
       }
       const findingMarker = `<!-- issue-dev-loop:${runId}:${intent.publication.findingId} -->`
       let findingPublished = false
@@ -1935,7 +1920,13 @@ export async function runWithGitHubRole({
       ? hardenedGitArguments(args, { expectedRepository: channel.repository })
       : [...args]
   if (activationValidation) {
-    executionArgs = [args[0], 'validate', '--loop-root', path.resolve(loopRoot)]
+    executionArgs = [
+      args[0],
+      'validate',
+      ...(authorization.issue ? ['--target-compatibility'] : []),
+      '--loop-root',
+      path.resolve(loopRoot),
+    ]
   }
   const child = spawnCommand(executable, executionArgs, {
     env: childEnvironment,