pantheon-opencode 1.6.0-beta.4 → 1.6.0-beta.6

package/scripts/run-node-tests.mjs

@@ -0,0 +1,189 @@
+#!/usr/bin/env node
+/**
+ * Run `node --test` over an explicit file list, and FAIL when zero tests are collected.
+ *
+ * Why this exists: `node --test <glob>` exits 0 when the glob matches nothing (Node >= 22).
+ * A test script whose file set is absent from the checkout -- or absent from the published
+ * package, because the `files` allow-list omits it -- therefore reports success while
+ * collecting nothing. That is the worst possible failure mode for a release gate: it
+ * manufactures confidence.
+ *
+ * This runner closes that hole from two independent sides:
+ *   1. It expands the globs ITSELF and refuses to start the runner with zero files, so the
+ *      runner is never handed an unexpanded pattern that silently matches nothing.
+ *   2. After a run that exited 0, it re-reads the collected count from the machine-readable
+ *      TAP summary and fails closed when that count is 0, unparseable, or missing.
+ *
+ * Side 2 is deliberately independent of side 1: a future Node change to `--test` glob
+ * handling, a glob that expands to a directory of empty files, or an emptied test file all
+ * still get caught.
+ *
+ * Human-readable output is unchanged: the default `spec` reporter still goes to stdout, and
+ * the TAP reporter is diverted to a scratch file that is read once and deleted.
+ *
+ * Usage:
+ *   node scripts/run-node-tests.mjs [--label <name>] [-- <node flags...>] <glob>...
+ *
+ * Examples:
+ *   node scripts/run-node-tests.mjs 'tests/*.mjs' 'tests/pantheon/*.mjs'
+ *   node scripts/run-node-tests.mjs -- --experimental-test-coverage 'tests/*.mjs'
+ */
+import { spawnSync } from 'node:child_process'
+import { globSync, mkdtempSync, readFileSync, rmSync } from 'node:fs'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
+import { fileURLToPath } from 'node:url'
+
+const ROOT = join(fileURLToPath(new URL('.', import.meta.url)), '..')
+
+function usage(message) {
+  console.error(`zero-test-guard: ${message}`)
+  console.error(
+    'usage: node scripts/run-node-tests.mjs [--label <name>] [-- <node flags...>] <glob>...',
+  )
+  process.exit(1)
+}
+
+/** Parse `[--label <name>] [-- <node flags...>] <glob>...`. */
+function parseArgs(argv) {
+  const separator = argv.indexOf('--')
+  const own = separator === -1 ? argv : argv.slice(0, separator)
+  const nodeFlags = separator === -1 ? [] : argv.slice(separator + 1)
+
+  let label = 'node tests'
+  const globs = []
+  for (let i = 0; i < own.length; i++) {
+    if (own[i] === '--label') {
+      const value = own[i + 1]
+      if (value === undefined) usage('--label requires a value')
+      label = value
+      i += 1
+      continue
+    }
+    if (own[i].startsWith('-')) usage(`unknown option ${own[i]}`)
+    globs.push(own[i])
+  }
+  if (globs.length === 0) usage('at least one glob is required')
+  return { label, globs, nodeFlags }
+}
+
+/**
+ * Expand globs to a deduped, ordered file list.
+ * Order follows the positional order of the globs so that the suite runs in the same order
+ * as the shell used to; within a glob, matches are sorted for determinism.
+ */
+function expand(globs) {
+  const files = []
+  const seen = new Set()
+  for (const pattern of globs) {
+    const matches = globSync(pattern, { cwd: ROOT })
+      .filter((match) => !match.includes('node_modules') && !match.includes('__pycache__'))
+      .sort()
+    for (const match of matches) {
+      if (seen.has(match)) continue
+      seen.add(match)
+      files.push(match)
+    }
+  }
+  return files
+}
+
+function reportMissingFiles(label, globs) {
+  console.error(`zero-test-guard: ${label} matched 0 test files — refusing to report success.`)
+  console.error(`  globs: ${globs.join(' ')}`)
+  console.error('  This is the "test suite silently passes when it collects nothing" failure.')
+  console.error('  Either the test files are missing from this checkout, or they are missing')
+  console.error('  from the published package (check the "files" allow-list in package.json).')
+}
+
+function reportZeroCollected(label, files) {
+  console.error(`zero-test-guard: ${label} collected 0 tests — refusing to report success.`)
+  console.error(`  ${files.length} file(s) matched, but not one test case was collected:`)
+  for (const file of files) console.error(`    ${file}`)
+}
+
+/**
+ * Count the real test cases collected in a TAP summary.
+ *
+ * Node counts a test FILE that declares no test cases as one passing "test"
+ * (`ok 1 - tests/foo.test.mjs`), so the summary's `# tests N` is >= the file count and
+ * can never reach 0 once any file matched. Relying on it alone would let a suite whose
+ * every test file was emptied out report success.
+ *
+ * The reliable discriminator is the subtest NAME: a file-level placeholder is named with
+ * the file path that was passed to the runner, a real test case is named with its test
+ * name. So a top-level subtest whose name is one of our matched paths is a placeholder
+ * and is not counted as a collected case.
+ *
+ * Returns null when the summary cannot be read or parsed -- the caller fails closed.
+ */
+function countCollectedTests(tapPath, files) {
+  let tap = ''
+  try {
+    tap = readFileSync(tapPath, 'utf8')
+  } catch {
+    return null
+  }
+  // The summary count is still asserted to exist: without it the stream is not TAP.
+  if (!/^#\s*tests\s+\d+\s*$/m.test(tap)) return null
+
+  const placeholders = new Set(files)
+  let collected = 0
+  for (const match of tap.matchAll(/^# Subtest: (.+)$/gm)) {
+    if (placeholders.has(match[1].trim())) continue
+    collected += 1
+  }
+  return collected
+}
+
+const { label, globs, nodeFlags } = parseArgs(process.argv.slice(2))
+const files = expand(globs)
+
+if (files.length === 0) {
+  reportMissingFiles(label, globs)
+  process.exit(1)
+}
+
+const scratch = mkdtempSync(join(tmpdir(), 'pantheon-node-tests-'))
+const tapPath = join(scratch, 'summary.tap')
+
+try {
+  const run = spawnSync(
+    process.execPath,
+    [
+      '--test',
+      '--test-reporter=spec',
+      '--test-reporter-destination=stdout',
+      '--test-reporter=tap',
+      `--test-reporter-destination=${tapPath}`,
+      ...nodeFlags,
+      ...files,
+    ],
+    { cwd: ROOT, stdio: 'inherit' },
+  )
+
+  if (run.error) {
+    console.error(`zero-test-guard: could not start the test runner: ${run.error.message}`)
+    process.exit(1)
+  }
+  if (run.signal) {
+    console.error(`zero-test-guard: the test runner was killed by ${run.signal}`)
+    process.exit(1)
+  }
+  // Real failures keep the runner's own exit code; the guard only speaks up when the
+  // runner claimed success it cannot justify.
+  if (run.status !== 0) process.exit(run.status)
+
+  const collected = countCollectedTests(tapPath, files)
+  if (collected === null) {
+    console.error('zero-test-guard: could not read the collected test count from the TAP summary.')
+    console.error('  Failing closed: an unreadable summary cannot certify that any test ran.')
+    process.exit(1)
+  }
+  if (collected === 0) {
+    reportZeroCollected(label, files)
+    process.exit(1)
+  }
+} finally {
+  rmSync(scratch, { recursive: true, force: true })
+}

package/scripts/sync-artifacts.mjs

@@ -31,13 +31,13 @@ import {
 
 /** Canonical MCP script sources — mirrors installOpenCode phase 2.10. */
 const MCP_SCRIPTS = {
-  'mcp_resources_server.py': () => join(ROOT, 'src', 'mcp', 'mcp_resources_server.py'),
-  'code_mode_server.py': () => join(ROOT, 'src', 'mcp', 'code_mode_server.py'),
-  'memory_mcp_server.py': () => join(ROOT, 'src', 'mcp', 'memory_mcp_server.py'),
-  'mcp_persistence_server.py': () => join(ROOT, 'src', 'mcp', 'mcp_persistence_server.py'),
+  'mcp_resources.py': () => join(ROOT, 'src', 'mcp', 'mcp_resources.py'),
+  'code_mode.py': () => join(ROOT, 'src', 'mcp', 'code_mode.py'),
+  'memory_mcp.py': () => join(ROOT, 'src', 'mcp', 'memory_mcp.py'),
+  'mcp_persistence.py': () => join(ROOT, 'src', 'mcp', 'mcp_persistence.py'),
   '_pantheon_paths.py': () => join(ROOT, 'src', 'mcp', '_pantheon_paths.py'),
   'mcp_codemap_module.py': () => join(ROOT, 'src', 'mcp', 'mcp_codemap_module.py'),
-  'pantheon_vision_server.py': () => join(ROOT, 'src', 'mcp', 'pantheon_vision_server.py'),
+  'pantheon_vision.py': () => join(ROOT, 'src', 'mcp', 'pantheon_vision.py'),
   'eval_store.py': () => join(ROOT, 'src', 'mcp', 'eval_store.py'),
   'scrub-secrets.py': () => join(ROOT, 'scripts', 'scrub-secrets.py'),
 }

package/scripts/test-opencode-v2-sandbox.sh

@@ -241,11 +241,11 @@ try:
 except OSError:
     raise SystemExit(1)
 expected = {
-    "code_mode_server.py",
-    "memory_mcp_server.py",
-    "mcp_persistence_server.py",
-    "mcp_resources_server.py",
-    "pantheon_vision_server.py",
+    "code_mode.py",
+    "memory_mcp.py",
+    "mcp_persistence.py",
+    "mcp_resources.py",
+    "pantheon_vision.py",
 }
 pattern = re.compile(
     r"spawning process \{\s*"
@@ -515,11 +515,11 @@ try:
 except OSError:
     raise SystemExit(1)
 expected = {
-    "code_mode_server.py",
-    "memory_mcp_server.py",
-    "mcp_persistence_server.py",
-    "mcp_resources_server.py",
-    "pantheon_vision_server.py",
+    "code_mode.py",
+    "memory_mcp.py",
+    "mcp_persistence.py",
+    "mcp_resources.py",
+    "pantheon_vision.py",
 }
 pattern = re.compile(
     r"spawning process \{\s*"
@@ -642,11 +642,11 @@ config_path = Path(os.environ["CONFIG_PATH"])
 runtime_dir = os.environ["RUNTIME_DIR"]
 python_bin = os.environ["MCP_PYTHON"]
 managed = {
-    "pantheon-code-mode": "code_mode_server.py",
-    "pantheon-memory": "memory_mcp_server.py",
-    "pantheon-persistence": "mcp_persistence_server.py",
-    "pantheon-resources": "mcp_resources_server.py",
-    "pantheon-vision": "pantheon_vision_server.py",
+    "pantheon-code-mode": "code_mode.py",
+    "pantheon-memory": "memory_mcp.py",
+    "pantheon-persistence": "mcp_persistence.py",
+    "pantheon-resources": "mcp_resources.py",
+    "pantheon-vision": "pantheon_vision.py",
 }
 try:
     config = json.loads(config_path.read_text(encoding="utf-8"))
@@ -829,7 +829,7 @@ else
 fi
 
 echo "--- Gate (b): pantheon://agents CONTENT check ---"
-RESOURCES_SRV="$PANTHEON_GLOBAL/scripts/mcp_resources_server.py"
+RESOURCES_SRV="$PANTHEON_GLOBAL/scripts/mcp_resources.py"
 if [ ! -f "$RESOURCES_SRV" ]; then
   record_status "pantheon://agents resource" "FAIL" "resources server not found at $RESOURCES_SRV"
 else
@@ -1013,11 +1013,11 @@ config_path = Path(os.environ["CONFIG_PATH"])
 runtime_dir = os.environ["RUNTIME_DIR"]
 python_bin = os.environ["MCP_PYTHON"]
 managed = {
-    "pantheon-code-mode": "code_mode_server.py",
-    "pantheon-memory": "memory_mcp_server.py",
-    "pantheon-persistence": "mcp_persistence_server.py",
-    "pantheon-resources": "mcp_resources_server.py",
-    "pantheon-vision": "pantheon_vision_server.py",
+    "pantheon-code-mode": "code_mode.py",
+    "pantheon-memory": "memory_mcp.py",
+    "pantheon-persistence": "mcp_persistence.py",
+    "pantheon-resources": "mcp_resources.py",
+    "pantheon-vision": "pantheon_vision.py",
 }
 
 try:

package/skills-lock.json

@@ -1,11 +1,11 @@
 {
   "src/skills/agent-coordination/SKILL.md": "cfd040ef29e3dd0fd5e0e67773874facf7c06e3284898afe1fddba4456c3f1aa",
   "src/skills/artifact-management/SKILL.md": "c60e493666bedf46e9113d0dd20106489d5c5a73905e1370fcfc1c0d71c59185",
-  "src/skills/auto-continue/SKILL.md": "4fe256d44ee9828d622a0d8b63cfc6bb39a317189a9ccf19494a6efd4527a7a9",
+  "src/skills/auto-continue/SKILL.md": "a7a4ec7ab35970bec36bcc286b95881b43a75e3bcdb626a17ebb9e397dd1a708",
   "src/skills/clonedeps/SKILL.md": "3e0b43cf96da98c83699dc48db3fd58fbfda84b9022bb70d3747875c81c3e124",
   "src/skills/code-review-checklist/SKILL.md": "0e59367c65ef4608db80a0674279f32cfc3f67dca7236f233b61a47fe506579f",
   "src/skills/codemap/SKILL.md": "87af5271e1576e154cadff313eab16264b726e43d094520736b7a5b03b409a4a",
-  "src/skills/context-compression/SKILL.md": "ebc721c715c1fa20e4ebf5177007fde94e001540b16cef4316393a665633c16f",
+  "src/skills/context-compression/SKILL.md": "9c65c132beb0dc7bd20335cfd7f41c589f5168a72dcf5d51740bc66f2865e027",
   "src/skills/git-workflow-and-versioning/SKILL.md": "685a94041f1b4ce2a430301f517b89102055696267c369a7e8fb191606659276",
   "src/skills/incremental-implementation/SKILL.md": "7065ddb45da3c6b6751de07f290e92e6cb072e949a2e3aaae13ee7762f0a70fc",
   "src/skills/loop-engineering/SKILL.md": "f804e545821b9cca8fe0ec517307ed902446f28c2ca8ff9194c0f987cf7229e3",

package/src/agents/aphrodite.md

@@ -122,7 +122,7 @@ Frontend: `tdd-with-agents`, `visual-review-pipeline`, `incremental-implementati
 
 
 ## Session Context Retrieval
-When dispatched by Zeus, call context_get(slug, "latest", session_id=SESSION_ID)
+When dispatched by Zeus, call context_get(slug=slug, key="latest", session_id=SESSION_ID)
 if session_slug is provided in dispatch metadata. Apply any "remaining_tasks",
 "current_phase", or "gotchas" from the retrieved context.
 If context_get returns None, proceed fresh (first phase or expired session).

package/src/agents/apollo.md

@@ -38,7 +38,10 @@ mcp_tools:
 - Generate structured summaries (not raw dumps)
 
 ### 2. External Research
-- Web search via bifrost MCP tools for documentation, blog posts, GitHub repos
+- Web search through whatever web-search MCP the USER has configured — Pantheon
+  bundles none and does not assume one. If no search tool is available, skip
+  external search and say so; never present an unverified recollection as a
+  citation.
 - Context7 for library documentation
 - Read URLs with webfetch for known resource URLs
 
@@ -50,7 +53,9 @@ mcp_tools:
 ##  TOOLS NOT AVAILABLE
 - bash - forbidden (cannot run commands)
 - edit - forbidden (read-only agent)
-- bifrost - use bifrost MCP tools for search
+- Web search by default - no search MCP is bundled or pre-configured. Use one
+  ONLY if the user's own config exposes it; otherwise state that external search
+  was unavailable instead of improvising.
 
 ## MCP Security
 - Never embed credentials in URLs (grep for token=, key=, secret=)

package/src/agents/athena.md

@@ -15,6 +15,7 @@ skills:
   - memory-bank
 permission:
   edit: deny
+  hashline_edit: deny
   bash: deny
   read: allow
   grep: allow

package/src/agents/demeter.md

@@ -111,7 +111,7 @@ Compress working context with the `context-compression` skill (L1, Pantheon-nati
 
 
 ## Session Context Retrieval
-When dispatched by Zeus, call context_get(slug, "latest", session_id=SESSION_ID)
+When dispatched by Zeus, call context_get(slug=slug, key="latest", session_id=SESSION_ID)
 if session_slug is provided in dispatch metadata. Apply any "remaining_tasks",
 "current_phase", or "gotchas" from the retrieved context.
 If context_get returns None, proceed fresh (first phase or expired session).

package/src/agents/hephaestus.md

@@ -64,7 +64,7 @@ Compress working context with the `context-compression` skill (L1, Pantheon-nati
 
 
 ## Session Context Retrieval
-When dispatched by Zeus, call context_get(slug, "latest", session_id=SESSION_ID)
+When dispatched by Zeus, call context_get(slug=slug, key="latest", session_id=SESSION_ID)
 if session_slug is provided in dispatch metadata. Apply any "remaining_tasks",
 "current_phase", or "gotchas" from the retrieved context.
 If context_get returns None, proceed fresh (first phase or expired session).

package/src/agents/hermes.md

@@ -270,7 +270,7 @@ Implementação: `tdd-with-agents`, `incremental-implementation`, `code-review-c
 
 
 ## Session Context Retrieval
-When dispatched by Zeus, call context_get(slug, "latest", session_id=SESSION_ID)
+When dispatched by Zeus, call context_get(slug=slug, key="latest", session_id=SESSION_ID)
 if session_slug is provided in dispatch metadata. Apply any "remaining_tasks",
 "current_phase", or "gotchas" from the retrieved context.
 If context_get returns None, proceed fresh (first phase or expired session).

package/src/agents/iris.md

@@ -11,6 +11,7 @@ permission:
     gh *: allow
   webfetch: allow
   edit: deny
+  hashline_edit: deny
   task:
     "*": deny
 temperature: 0.2

package/src/agents/nyx.md

@@ -9,6 +9,7 @@ permission:
   bash: allow
   webfetch: allow
   edit: deny
+  hashline_edit: deny
   task:
     "*": deny
 temperature: 0.1

package/src/agents/themis.md

@@ -8,6 +8,7 @@ permission:
   grep: allow
   bash: allow
   edit: deny
+  hashline_edit: deny
   task:
     "*": deny
 temperature: 0.1

package/src/agents/zeus.md

@@ -5,6 +5,7 @@ reasoning_effort: medium
 permission:
   read: allow
   edit: deny
+  hashline_edit: deny
   bash: deny
   task:
     "*": allow
@@ -32,7 +33,7 @@ skills:
 ## Memory Protocol
 
 **Auto-Store:** Ao receber subtask_summary, chame `memory_store()` com summary/files_changed/tests/status. Sempre.
-**Pre-work:** `memory_search("<feature>", top_k=3)` antes de planejar qualquer coisa.
+**Pre-work:** `memory_search(query="<feature>", top_k=3)` antes de planejar qualquer coisa.
 
 ## Golden Rule
 
@@ -79,7 +80,7 @@ O fluxo é SEMPRE: **Planejar → Especificar → Delegar → Revisar**. Zeus nu
 Antes de usar a arvore de roteamento, consulte o memory:
 
 ```
-memory_search(task_prompt, top_k=2)
+memory_search(query=task_prompt, top_k=2)
   → score > 0.85?
     SIM → usa resultado cacheado (agent, background, pattern)
     NAO → aplica arvore de roteamento + memory_store() pra proxima vez
@@ -90,8 +91,8 @@ memory_search(task_prompt, top_k=2)
 Para padroes de delegacao recorrentes, grave no KV:
 
 ```
-kv_store("delegation:<pattern>", "{agent: ..., background: true/false}")
-kv_get("delegation:<pattern>") → reusa decisao sem memory_search
+kv_store(namespace="deleg", key="delegation:<pattern>", value="{agent: ..., background: true/false}")
+kv_get(namespace="deleg", key="delegation:<pattern>") → reusa decisao sem memory_search
 ```
 
 ### Telemetria de delegacao (Nyx P1-3)
@@ -186,12 +187,12 @@ Wave announcement obrigatorio.
 Limite maximo de 2 niveis de nesting: Zeus -> subagente -> sub-subagente.
 
 ```
-depth = kv_get("deleg:depth") ?? 0
+depth = kv_get(namespace="deleg", key="deleg:depth") ?? 0
 if depth >= 2 → NAO delegar, ESCALAR para o usuario
-else → kv_store("deleg:depth", depth + 1)
+else → kv_store(namespace="deleg", key="deleg:depth", value=String(depth + 1))
 
 Quando subagente retornar:
-  kv_store("deleg:depth", max(0, depth - 1))
+  kv_store(namespace="deleg", key="deleg:depth", value=String(max(0, depth - 1)))
 ```
 
 Zeus (nivel 0) -> Apollo/Hermes (nivel 1) -> sub-subagente (nivel 2 max).
@@ -220,7 +221,7 @@ Quando um subagente RECUSA ou retorna "escopo fora do meu dominio" (ex: edit: de
 
 - **Causa provavel:** agente errado selecionado no roteamento (match por palavra-chave, nao por capacidade).
 - **ACAO:** RE-ROTEAR imediatamente para o agente correto — NAO retry com prompt reformulado.
-- Registre o caso no cache de delegacao (`kv_store("deleg:<pattern>", ...)` para aprendizado futuro).
+- Registre o caso no cache de delegacao (`kv_store(namespace="deleg", key="deleg:<pattern>", value=...)` para aprendizado futuro).
 - Recusa legitima do agente = comportamento correto do guard; o defeito esta na selecao (zeus), nao no agente.
 
 ## TODO Enforcer (Auto-Retry)
@@ -237,7 +238,7 @@ Apos task_status(wait=true), verifique:
     "reviewer-only", "fora do meu dominio", ou similar negativa de dominio)?
       SIM → NAO retry com prompt rephrased.
             RE-ROTEAR para o agente correto (ver "Taxonomia de Recusa (P0-3)").
-            Registre o caso: kv_store("deleg:<pattern>", "{agente_correto, ...}").
+            Registre o caso: kv_store(namespace="deleg", key="deleg:<pattern>", value="{agente_correto, ...}").
 
     FALHA REAL (timeout, crash, resposta vazia, context exceeded)?
       SIM → retry 1x com prompt rephrased (diferente, mais especifico)
@@ -284,7 +285,35 @@ in `task-result-guard.ts` (plugin `tool.execute.after` chain): an
 empty/whitespace-only result is converted into an explicit error so a silent
 child failure is never mistaken for success. No manual retry helper is required.
 
-### /cost — pantheon_cost tool (WIRED no plugin)
+### /cost — pantheon_cost tool (disponivel com o version gate corrigido)
+`pantheon_cost` volta a existir assim que o version gate corrigido estiver
+instalado **e após o usuário re-rodar `init`**: em host 2.x, `auto` resolve
+para `v2` e registra o entry de diretorio `src/plugin-v2`, que expoe
+exatamente 3 tools — `hashline_edit`, `pantheon_cost` e `pantheon_model`
+(V2 nao registra as 3 goal tools; ver `docs/INSTALLATION.md` ->
+"OpenCode V1/V2 — contrato de plugin").
+**Caveat**: a tool so reaparece para quem re-rodar `init` depois do fix. O
+`postinstall` nao re-executa `init`, entao ate la o config existente continua
+resolvendo para V1 e a tool segue ausente — nesse estado, tratar `/cost` como
+indisponivel.
+
+Causa raiz como era na tag `1.6.0-beta.4` (registro para quem ainda tem um
+install quebrado): o host `opencode v2.0.22` rejeita caminhos de arquivo `.ts` nos
+arrays de plugin do config (recusa `configured plugin path must be a directory`,
+medida em runtime nesse host; o repo so registra a parafrase `must be a
+directory`), e a selecao de geracao caia sempre em V1 — a geracao V1 escreve
+entradas `.ts` sob a chave singular `plugin` (`config.plugin`, o bloco de merge
+em `scripts/install/opencode.mjs`) e o host descarta todas, logo nenhum
+modulo Pantheon entra no processo Node. **O que mudou**: `resolveOpenCodeVersion`
+agora tem default `'auto'` nos dois call sites e no proprio signature, e `auto`
+consulta o host. Precedencia `auto`: flag explicito > `OPENCODE_VERSION` >
+basename `opencode2` > probe `--version` do host (major >= 2 => v2) > soft-fail
+para V1 com warning visivel. Precedencia e criterios do probe estao em
+`docs/INSTALLATION.md` -> "OpenCode V1/V2 — contrato de plugin".
+
+Comportamento da tool — **NAO VERIFICADO em runtime**: em 1.6.0-beta.4 o modulo
+(`src/pantheon/cost-command.ts`) nunca era carregado no processo, e o fix nao foi
+exercitado end-to-end, entao este descritivo permanece nao verificado.
 `pantheon_cost({ days?: number })` le o `opencode.db` READ-ONLY usando o backend
 unico `node:sqlite`; se indisponivel, retorna `UNSUPPORTED`, sem fallback. Devolve tabela markdown de custo + tokens por
 agente nos ultimos N dias (default 7). Uso: quando Nyx pedir visibilidade de

package/src/instructions/memory-protocol.instructions.md

@@ -54,15 +54,21 @@ Each agent file defines overrides in its `## 🧠 Memory Protocol` section:
 
 Para otimizar decisoes de delegacao e reduzir gasto de tokens:
 
-1. **memory_search(task_prompt, top_k=2)** antes de aplicar a arvore de roteamento
+1. **memory_search(query=task_prompt, top_k=2)** antes de aplicar a arvore de roteamento
 2. Se score > 0.85 → reutiliza agente + background_mode do cache
 3. Se score ≤ 0.85 → aplica regras estaticas e memory_store() com:
    - key: deleg:<task_type>
-   - value: {agent, background, pattern}
-   - metadata: {type: "decision", score: N}
+   - value: JSON.stringify({agent, background, pattern})
+   - metadata: JSON.stringify({type: "decision", score: N})
 
-4. **kv_store("deleg:<pattern>", ...)** para padroes recorrentes de delegacao
-5. **kv_get("deleg:<pattern>")** para reusar decisoes ja tomadas
+   `value` E `metadata` sao JSON-encoded strings — o runtime exige que o
+   chamador serialize os dois. Objeto cru em `metadata` e rejeitado pelo
+   `BeforeValidator` do servidor com `metadata must be a JSON object encoded
+   as a string (got object). json.dumps it first. Example: metadata='{"type":
+   "decision", "score": 0.9}'`.
+
+4. **kv_store(namespace="deleg", key="deleg:<pattern>", value=...)** para padroes recorrentes de delegacao
+5. **kv_get(namespace="deleg", key="deleg:<pattern>")** para reusar decisoes ja tomadas
 
 Isso elimina ~300 tokens de reasoning por delegacao quando o cache acerta.
 
@@ -76,7 +82,7 @@ After every `/pantheon` council synthesis completes, Zeus stores:
 memory_store({
   namespace: "council_decisions",
   key: "council:<yyyy-mm-dd>:<slug>",
-  value: {
+  value: JSON.stringify({
     question: "original question",
     specialists: ["@agent1", "@agent2"],
     recommendation: "final recommendation",
@@ -87,20 +93,25 @@ memory_store({
     themis_audit: "approved|issues",
     precedent_used: false,
     timestamp: "<ISO-8601>"
-  },
-  metadata: {
+  }),
+  metadata: JSON.stringify({
     type: "council_decision",
     specialist_count: N,
     model_tier_used: "premium|default|fast"
-  }
+  })
 })
 ```
 `value` is JSON-serialized before storing (the MCP `memory_store.value` argument is a string).
+`metadata` works the same way — it is a JSON object encoded as a **string**, so pass
+`JSON.stringify({...})`. A raw object is rejected with `metadata must be a JSON object
+encoded as a string (got object). json.dumps it first. Example: metadata='{"type":
+"decision", "score": 0.9}'` (enforced by a `BeforeValidator` in
+`memory_mcp.py`). **Both `value` and `metadata` must be serialized by the caller.**
 
 ### Read Path (Precedent Fast-Path)
 Before dispatching a new council, Zeus runs:
 ```
-memory_search(question, top_k=2, namespace="council_decisions")
+memory_search(query=question, top_k=2, namespace="council_decisions")
 ```
 
 Result interpretation:

package/src/instructions/zeus-anti-stall.instructions.md

@@ -56,19 +56,27 @@ All session state lives in **pantheon-persistence** (namespace `checkpoint:<slug
 No file I/O, no checkpoint_session.py — TTL (4h) handles cleanup automatically.
 
 ### Heartbeat Check
-- If `context_get(slug, "heartbeat")` returns a checkin older than 300s, log a stall warning and resume
+- If `context_get(slug=slug, key="heartbeat", session_id=SESSION_ID)` returns a checkin older than 300s, log a stall warning and resume
 - Write heartbeat after every anti-stall recovery action:
   ```
-  context_save(slug, "heartbeat", json({"status": "alive", "last_action": "...", "turn_count": N}))
+  context_save(slug=slug, key="heartbeat", content=JSON.stringify({"status": "alive", "last_action": "...", "turn_count": N}), session_id=SESSION_ID)
   ```
 
 ### Checkpoint Auto-Save (Pré-Compactação)
 Before ANY delegate dispatch, save a checkpoint:
 ```
-context_save(slug, "phase:N", json({
-  "phase": N, "turn_count": N, "agent": "...", "summary": "..."
+context_save(slug=slug, key="phase:N", content=JSON.stringify({
+  "phase": {"current": N, "total": M, "name": "..."},
+  "turn_count": N, "agent": "...", "summary": "..."
 }), session_id=SESSION_ID)
 ```
+**`content.phase` must be an OBJECT, not a bare number.** `{"phase": 1}` is rejected with
+`context_save: content.phase must be an object (got number)`. The server reads `current`
+and `total` (non-negative ints) and `name` (string, max 256 **bytes** — the guard
+measures `len(value.encode())`, so multi-byte characters count as more than one). That
+set is not closed: any other key inside `phase` is accepted and ignored, never an
+error. The `"phase:N"` slot KEY above is unrelated to `content.phase` and stays a plain
+string.
 All checkpoints auto-expire after 4h (TTL=14400).
 
 ### Gatilho de Pré-Compactação (Anti-perda de estado)
@@ -78,15 +86,15 @@ o Zeus DEVE salvar o estado atual:
 2. Salve heartbeat + phase atual + tarefas pendentes
 3. Só então permita que a compactação prossiga
 ```
-# Ao iniciar sessão:
-result = context_save(slug, "init", session_state)
+# Ao iniciar sessão (session_id é REQUIRED — use o id da sessão corrente):
+result = context_save(slug=slug, key="init", content=JSON.stringify(session_state), session_id=SESSION_ID)
 SESSION_ID = result.session_id   # ← guarde para toda a sessão
 
 # Antes de CADA delegação:
-context_save(slug, f"pre:{agent}", current_state, session_id=SESSION_ID)
+context_save(slug=slug, key=f"pre:{agent}", content=JSON.stringify(current_state), session_id=SESSION_ID)
 
 # Após retorno do agente:
-context_save(slug, f"post:{agent}", result_state, session_id=SESSION_ID)
+context_save(slug=slug, key=f"post:{agent}", content=JSON.stringify(result_state), session_id=SESSION_ID)
 ```
 Isso garante que o estado sobreviva à compactação — o "latest" pointer
 sempre aponta para o checkpoint mais recente, mesmo após compactação.
@@ -95,9 +103,9 @@ sempre aponta para o checkpoint mais recente, mesmo após compactação.
 ### Context Retrieval
 Next-phase agents retrieve previous context via:
 ```
-context_get(slug, "latest")        # most recent checkpoint
-context_get(slug, "phase:3")      # specific phase
-context_list(slug)                 # all checkpoints
+context_get(slug=slug, session_id=SESSION_ID)                    # most recent checkpoint ("latest" is the default key)
+context_get(slug=slug, key="phase:3", session_id=SESSION_ID)     # specific phase
+context_list(slug=slug, session_id=SESSION_ID)                   # all checkpoints
 ```
 
 ### Long-Session Progress