fix: CI failure + new Sentry finding — orphaned headers and agent test

1. Agent test: add researcher + trainer to "all disabled" test so it
   correctly expects "no primary visible agent" when ALL agents are off

2. Orphaned section headers: add pre-check that at least one entry fits
   before adding section header in both injectTrainingOnly and inject
   memory section (prevents header-only output inflating budget reports)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: anandgupta42

packages/opencode/.github/meta/commit.txt

@@ -1,22 +1,27 @@
-refactor: cut training_scan and training_validate, simplify docs
+fix: address Sentry review findings — 7 bugs fixed
 
-Research from 8 independent evaluations + SkillsBench (7,308 test runs)
-found that compact focused context beats comprehensive docs by 20pp.
-The training system's value is in correction capture (2-sec saves) and
-team propagation (git sync) — not in regex scanning or keyword grep.
+1. stripTrainingMeta/parseTrainingMeta regex: remove multiline `m` flag
+   that could match user content starting with `<!-- training` mid-string
+   (types.ts, store.ts)
 
-Removed:
-- training_scan (255 lines) — regex pattern counting, not discovery
-- training_validate (315 lines) — keyword grep, not validation
+2. training_save content limit: reduce from 2500 to 1800 chars to account
+   for ~200 char metadata overhead against MemoryStore's 2048 char limit
+   (training-save.ts)
 
-Simplified:
-- trainer.txt: removed scan/validate workflows, focused on guided
-  teaching and curation
-- agent-modes.md: updated trainer section with correction-focused example
-- training docs: complete rewrite with new pitch:
-  "Correct the agent once. It remembers forever. Your team inherits it."
-  Backed by SkillsBench research showing compact > comprehensive.
+3. injectTrainingOnly: change `break` to `continue` so budget-exceeding
+   section headers skip to next kind instead of stopping all injection
+   (memory/prompt.ts)
 
-Net: -753 lines. 152 tests pass.
+4. injectTrainingOnly: track itemCount and return empty string when no
+   items injected (was returning header-only string, inflating budget
+   reports) (memory/prompt.ts)
+
+5. projectDir cache: replace module-level singleton with Map keyed by
+   Instance.directory to prevent stale paths when AsyncLocalStorage
+   context changes across concurrent requests (memory/store.ts)
+
+6. budgetUsage side effect: already fixed — delegates to injectTrainingOnly
+   which is read-only (no applied count increment). Sentry comments were
+   against pre-refactor code.
 
 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

packages/opencode/src/memory/prompt.ts

@@ -199,7 +199,8 @@ export namespace MemoryPrompt {
     // Inject non-training memory blocks
     if (memoryBlocks.length > 0) {
       const memHeader = "\n### Memory\n"
-      if (used + memHeader.length < budget) {
+      const firstMemFormatted = formatBlock(memoryBlocks[0].block)
+      if (used + memHeader.length + firstMemFormatted.length + 2 < budget) {
         result += memHeader
         used += memHeader.length
 
@@ -270,16 +271,19 @@ export namespace MemoryPrompt {
 
       const section = KIND_HEADERS[kind]
       const sectionHeader = `\n### ${section.header}\n_${section.instruction}_\n`
-      if (used + sectionHeader.length > budget) continue
-      result += sectionHeader
-      used += sectionHeader.length
 
       const sorted = [...items].sort((a, b) => {
         const metaA = parseTrainingMeta(a.content)
         const metaB = parseTrainingMeta(b.content)
         return (metaB?.applied ?? 0) - (metaA?.applied ?? 0)
       })
 
+      // Check if header + at least one entry fits before adding header
+      const firstFormatted = formatTrainingEntry(sorted[0])
+      if (used + sectionHeader.length + firstFormatted.length + 2 > budget) continue
+      result += sectionHeader
+      used += sectionHeader.length
+
       for (const block of sorted) {
         const formatted = formatTrainingEntry(block)
         const needed = formatted.length + 2

packages/opencode/test/agent/agent.test.ts

@@ -683,6 +683,8 @@ test("defaultAgent throws when all primary agents are disabled", async () => {
         executive: { disable: true },
         validator: { disable: true },
         migrator: { disable: true },
+        researcher: { disable: true },
+        trainer: { disable: true },
         plan: { disable: true },
       },
     },