pantheon-opencode 2.1.3 → 2.1.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/AGENTS.global.md CHANGED
@@ -1,7 +1,7 @@
1
1
  <!-- pantheon:generated:begin — Pantheon-managed region; regenerated on install/upgrade, never edit inside -->
2
2
  <!-- Generated by scripts/build-agents-md.mjs — do not edit directly. -->
3
3
  <!-- Source of truth: src/instructions/*.instructions.md + the static sections in this script. -->
4
- <!-- pantheon:agents-md v2 artifact=global generated-at=2026-11-14T10:35:21-03:00 sha=0200775 -->
4
+ <!-- pantheon:agents-md v2 artifact=global generated-at=2026-11-14T14:44:27-03:00 sha=053433e -->
5
5
 
6
6
  # Pantheon Agent System — OpenCode
7
7
 
@@ -90,11 +90,11 @@ Required: position, reasoning, trade_offs, risks, confidence. Optional `agreemen
90
90
 
91
91
  Write one focused test per behavior; do not duplicate coverage.
92
92
 
93
- - **Test Slicing**: In dev loops, run ONLY the narrowest slice (<10s) for changed code (`node --test <slice>`, `pytest -k`); avoid unconstrained suites.
94
- - **Repo Selector & Receipt**: Prefer repo selectors (`test:quick`, `test:slice`); record command, targets, duration, and selection reason in `subtask_summary`.
95
- - **Failure/Fix Loops**: Rerun ONLY failed checks; never rerun passing tests without code changes.
96
- - **Mapping Gaps**: If a file lacks tests, run static check fallback plus nearest slice.
97
- - Slow install/E2E suites are opt-in, never per edit; reserve for release gates.
93
+ - **Risk-Based Slicing**: Run narrowest slice (<10s) first; broaden only on cross-module, contract, auth, or DB risk.
94
+ - **Suite Escalation**: Full suites are for CI/release; state if execution is agent recommendation or mandatory gate.
95
+ - **Failure/Fix Loops**: Rerun ONLY failed checks; never repeat passing tests. Never delete tests to speed up checks.
96
+ - **Mapping Gaps & Receipts**: Missing tests use static check plus nearest slice; record command, duration, and pending gates.
97
+ - **CLI vs Code**: Run CLI binaries (`opencode-quota`, `git`, `npm`) via `shell`/`bash`, not eval; on failure, never glob `~/.config`.
98
98
  - Test economy does NOT weaken stronger gates: auth, security, data-integrity, and required coverage.
99
99
 
100
100
  <!-- Source: src/instructions/yagni.instructions.md -->
package/AGENTS.md CHANGED
@@ -1,7 +1,7 @@
1
1
  <!-- pantheon:generated:begin — Pantheon-managed region; regenerated on install/upgrade, never edit inside -->
2
2
  <!-- Generated by scripts/build-agents-md.mjs — do not edit directly. -->
3
3
  <!-- Source of truth: project sections in scripts/build-agents-md.mjs. -->
4
- <!-- pantheon:agents-md v2 artifact=project generated-at=2026-11-14T10:35:21-03:00 sha=0200775 -->
4
+ <!-- pantheon:agents-md v2 artifact=project generated-at=2026-11-14T14:44:27-03:00 sha=053433e -->
5
5
 
6
6
  # Pantheon — Repository Setup
7
7
 
package/CHANGELOG.md CHANGED
@@ -13,6 +13,63 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
13
13
  `node scripts/versioning.mjs apply --notes` pre-fills the generated sections from conventional commits.
14
14
  Running apply will move this section to a versioned entry and reset the template below. -->
15
15
 
16
+ ## [v2.1.5] - 2026-11-14
17
+
18
+ ### 🆕 What's New
19
+
20
+ - **Risk-Based Validation (#444)**: Protocol distinguishes focused test slices, package suites, and mandatory repository CI gates.
21
+ - **Tool Execution Fallback**: Added `execute` alias executing commands via bash, preventing JS eval syntax errors on CLI commands.
22
+ - **CLI Execution Mandate**: Enforced CLI execution via `shell`/`bash` for binaries like `opencode-quota` with anti-panic search ban.
23
+ - **Scope Transparency (#444)**: `subtask_summary` explicitly reports validation scope and deferred pending repository CI pipeline gates.
24
+ - **Unified Skill Discovery**: Dynamic filesystem indexing of local and global skills; Zeus handoff passes explicit skill file paths.
25
+
26
+ ### Added
27
+
28
+ - **tools** — added `execute` fallback alias executing workspace commands and scripts via `/bin/bash` in `src/pantheon/tools.ts`.
29
+ - **permissions** — whitelisted `opencode-quota` in `SAFE_DEV_BINARIES` and installer bash permission rules.
30
+ - **skills** — added dynamic filesystem scanner for `.opencode/skills` and `.agents/skills` returning real path and scope.
31
+
32
+ ### Changed
33
+
34
+ - **tests (#444)** — enhanced `test-slice.mjs` with risk escalation triggers, validation scope reporting, and pending CI gate receipts.
35
+ - **instructions (#444)** — updated `test-economy` and `test-slicing` with risk-based validation protocol, anti-panic rules, and CLI mandate.
36
+ - **agents** — updated Zeus and specialist agent prompts in `roster.json` to mandate `SKILL.md` inspection before implementation.
37
+
38
+ ### ✅ Closed Issues
39
+
40
+ - #444 - define risk-based validation protocol for pantheon agents
41
+
42
+ &lt;!-- Add new changes here.
43
+ Canonical sections (docs/RELEASING.md): ### 🆕 What's New (curated highlights, ≤6 bullets ≤140 chars) →
44
+ ### Added / Changed / Deprecated / Removed / Fixed / Security → ### ⚠️ Known Issues / ### ✅ Closed Issues.
45
+ `node scripts/versioning.mjs apply --notes` pre-fills the generated sections from conventional commits.
46
+ Running apply will move this section to a versioned entry and reset the template below. --&gt;
47
+ ## [v2.1.4] - 2026-11-14
48
+
49
+ ### 🆕 What's New
50
+
51
+ - **Global Install Shadowing (#424)**: Doctor check (H2b) detects multiple global binaries, dangling symlinks, and prefix skew.
52
+ - **Strict Test Slice Ordering (#442)**: Fixed `--test-name-pattern` flag ordering in `test:slice` so regex filters work during fix loops.
53
+ - **Security Boundaries (#441)**: Hardened `auto-permissions` and `zeusReadGuard` to exclude `.git/` and `node_modules/` from reads.
54
+ - **Universal Session Todos (#439)**: Granted `todowrite` and `todoread` across all 14 specialist agent manifests.
55
+ - **Session Auto-Discovery (#440)**: Automated active session discovery from environment and SQLite `opencode.db` for vision.
56
+
57
+ ### Fixed
58
+
59
+ - **install (#424)** — added doctor check (H2b) for PATH shadowing, version skew check in `resolveStablePantheonRoot`, and sanitized npm env in `sync-tui.mjs`.
60
+ - **tests (#442)** — reordered `--test-name-pattern` before test path in `scripts/test-slice.mjs` so Node's test runner enforces filters during fix loops.
61
+ - **security (#441)** — enforced deterministic path checks in `auto-permissions.ts` and `zeusReadGuard` denying `.git/**`, `node_modules/**`, and traversal.
62
+ - **tools (#439)** — granted `todowrite` and `todoread` across all 14 specialist agent manifests in `src/agents/roster.json` and registered tool capabilities.
63
+ - **vision (#440)** — automated session ID resolution from `OPENCODE_SESSION_ID` and SQLite `opencode.db` (`session_v2`) for `vision_describe`.
64
+
65
+ ### ✅ Closed Issues
66
+
67
+ - #424 - detect and prevent global install shadowing and prefix version skew
68
+ - #439 - expose todowrite and todoread across specialist agents
69
+ - #440 - auto-discover OpenCode session ID for vision_describe
70
+ - #441 - enforce .git and node_modules exclusion in auto-permissions and guards
71
+ - #442 - reorder test-name-pattern filter before file path in test:slice
72
+
16
73
  ## [v2.1.3] - 2026-11-14
17
74
 
18
75
  ### 🆕 What's New
@@ -254,6 +311,30 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
254
311
 
255
312
  ## [Unreleased]
256
313
 
314
+ ### 🆕 What's New
315
+
316
+ - **Global Install Shadowing (#424)**: Doctor check (H2b) detects multiple global binaries, dangling symlinks, and prefix skew.
317
+ - **Strict Test Slice Ordering (#442)**: Fixed `--test-name-pattern` flag ordering in `test:slice` so regex filters work during fix loops.
318
+ - **Security Boundaries (#441)**: Hardened `auto-permissions` and `zeusReadGuard` to exclude `.git/` and `node_modules/` from reads.
319
+ - **Universal Session Todos (#439)**: Granted `todowrite` and `todoread` across all 14 specialist agent manifests.
320
+ - **Session Auto-Discovery (#440)**: Automated active session discovery from environment and SQLite `opencode.db` for vision.
321
+
322
+ ### Fixed
323
+
324
+ - **install (#424)** — added doctor check (H2b) for PATH shadowing, version skew check in `resolveStablePantheonRoot`, and sanitized npm env in `sync-tui.mjs`.
325
+ - **tests (#442)** — reordered `--test-name-pattern` before test path in `scripts/test-slice.mjs` so Node's test runner enforces filters during fix loops.
326
+ - **security (#441)** — enforced deterministic path checks in `auto-permissions.ts` and `zeusReadGuard` denying `.git/**`, `node_modules/**`, and traversal.
327
+ - **tools (#439)** — granted `todowrite` and `todoread` across all 14 specialist agent manifests in `src/agents/roster.json` and registered tool capabilities.
328
+ - **vision (#440)** — automated session ID resolution from `OPENCODE_SESSION_ID` and SQLite `opencode.db` (`session_v2`) for `vision_describe`.
329
+
330
+ ### ✅ Closed Issues
331
+
332
+ - #424 - detect and prevent global install shadowing and prefix version skew
333
+ - #439 - expose todowrite and todoread across specialist agents
334
+ - #440 - auto-discover OpenCode session ID for vision_describe
335
+ - #441 - enforce .git and node_modules exclusion in auto-permissions and guards
336
+ - #442 - reorder test-name-pattern filter before file path in test:slice
337
+
257
338
  <!-- Add new changes here.
258
339
  Canonical sections (docs/RELEASING.md): ### 🆕 What's New (curated highlights, ≤6 bullets ≤140 chars) →
259
340
  ### Added / Changed / Deprecated / Removed / Fixed / Security → ### ⚠️ Known Issues / ### ✅ Closed Issues.
package/README.md CHANGED
@@ -225,6 +225,25 @@ Working runnable reference projects demonstrating Pantheon orchestration in prac
225
225
 
226
226
  ---
227
227
 
228
+ ## What's New in v2.1.5
229
+
230
+ - **Risk-Based Validation Protocol (#444):** Explicit multi-tier protocol distinguishing focused test slices (<10s), package integration suites, and mandatory repository CI gates. Escalation to broad suites requires explicit triggers (cross-module, breaking contracts, auth, DB migrations).
231
+ - **Tool Execution Resilience & Fallback:** Added `execute` fallback tool alias executing commands directly in `/bin/bash`, preventing `SyntaxError: Assigning to rvalue` when models call `execute` with CLI commands or scripts.
232
+ - **CLI Execution Mandate:** Enforced CLI execution via `shell`/`bash` for binaries like `opencode-quota` with an anti-panic ban on speculative recursive `glob`/`grep` in `~/.config`.
233
+ - **Validation Scope Transparency (#444):** `subtask_summary` observable receipts report `validation_scope` and deferred `pending_validation` repository CI pipeline gates.
234
+
235
+ ---
236
+
237
+ ## What's New in v2.1.4
238
+
239
+ - **Global Install Shadowing & PATH Hygiene (#424):** Doctor check (H2b) detects multiple global binaries, dangling symlinks, and prefix skew. Version skew validation prevents `_npx` from binding to older global packages.
240
+ - **Strict Test Slice Ordering (#442):** Reordered `--test-name-pattern` flag ahead of the test file path in `scripts/test-slice.mjs` so Node's native test runner filters tests during failure/fix loops.
241
+ - **Security Boundaries on Internal Folders (#441):** Hardened `auto-permissions` and `zeusReadGuard` to exclude `.git/` and `node_modules/` from autonomous reads.
242
+ - **Universal Session Todo Tool Availability (#439):** Granted `todowrite` and `todoread` across all 14 specialist agent manifests.
243
+ - **Session Auto-Discovery for Zen Vision (#440):** Automated active session discovery from environment (`OPENCODE_SESSION_ID`) and SQLite `opencode.db` (`session_v2`) for `vision_describe`.
244
+
245
+ ---
246
+
228
247
  ## What's New in v2.0.1
229
248
 
230
249
  - **Bilingual TUI Settings (PT-BR / EN):** Visual localization for Decision Router, Autonomy & Budgets, and J1 Safety Gate tabs, with intelligent system locale auto-detection (`LANG` / `LC_ALL`) and manual preference cycling (`auto`, `pt-BR`, `en`).
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "pantheon-opencode",
3
- "version": "2.1.3",
3
+ "version": "2.1.5",
4
4
  "description": "Multi-agent orchestration for OpenCode \u2014 14 specialized agents, TDD, Themis review gates.",
5
5
  "type": "module",
6
6
  "workspaces": [
package/plugin.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "pantheon",
3
3
  "publisher": "ils15",
4
- "version": "2.1.3",
4
+ "version": "2.1.5",
5
5
  "displayName": "Pantheon",
6
6
  "description": "Conductor-Delegate orchestration platform with 14 specialized agents, 25 skills, and OpenCode support.",
7
7
  "agents": "./src/agents",
package/pyproject.toml CHANGED
@@ -4,7 +4,7 @@ build-backend = "setuptools.backends._legacy:_Backend"
4
4
 
5
5
  [project]
6
6
  name = "pantheon"
7
- version = "2.1.3"
7
+ version = "2.1.5"
8
8
  description = "Multi-agent orchestration framework"
9
9
  requires-python = ">=3.11"
10
10
  # Python sources under src/mcp/ import fastmcp/pydantic at runtime, but those
@@ -22,9 +22,17 @@
22
22
  * Exit codes: 0 = no blocking errors (warnings are allowed), 2 = errors.
23
23
  */
24
24
 
25
- import { spawn as spawnAsync, spawnSync } from 'node:child_process'
25
+ import { execSync, spawn as spawnAsync, spawnSync } from 'node:child_process'
26
26
  import { createHash } from 'node:crypto'
27
- import { existsSync, lstatSync, mkdirSync, readdirSync, readFileSync, statSync } from 'node:fs'
27
+ import {
28
+ existsSync,
29
+ lstatSync,
30
+ mkdirSync,
31
+ readdirSync,
32
+ readFileSync,
33
+ realpathSync,
34
+ statSync,
35
+ } from 'node:fs'
28
36
  import { homedir } from 'node:os'
29
37
  import { basename, dirname, isAbsolute, join, resolve } from 'node:path'
30
38
  import { fileURLToPath } from 'node:url'
@@ -1514,6 +1522,106 @@ export function checkInstallVersionDrift(args) {
1514
1522
  return 'sync'
1515
1523
  }
1516
1524
 
1525
+ // ---------------------------------------------------------------------------
1526
+ // Check H2b: PATH shadowing across global npm prefixes (Issue #424)
1527
+ // ---------------------------------------------------------------------------
1528
+
1529
+ /**
1530
+ * Check H2b: PATH shadowing across global npm prefixes.
1531
+ *
1532
+ * When multiple copies of pantheon-opencode are installed globally (e.g. across nvm,
1533
+ * ~/.npm-global, /usr/local), PATH may resolve an older or different version than the
1534
+ * package running doctor or OpenCode.
1535
+ */
1536
+ export function checkPathShadowing(args = {}) {
1537
+ section('H2b. PATH Shadowing & Binary Resolution')
1538
+ const env = args.env ?? process.env
1539
+ const isWindows = process.platform === 'win32'
1540
+ const whichCmd = isWindows ? 'where pantheon-opencode' : 'which -a pantheon-opencode'
1541
+
1542
+ let raw = ''
1543
+ try {
1544
+ raw = execSync(whichCmd, {
1545
+ env,
1546
+ encoding: 'utf8',
1547
+ stdio: ['ignore', 'pipe', 'ignore'],
1548
+ }).trim()
1549
+ } catch {
1550
+ info('No global pantheon-opencode binary found in PATH')
1551
+ return 'no-binary'
1552
+ }
1553
+
1554
+ const binaries = raw
1555
+ .split(/\r?\n/)
1556
+ .map((line) => line.trim())
1557
+ .filter(Boolean)
1558
+
1559
+ if (binaries.length === 0) {
1560
+ info('No pantheon-opencode binary found in PATH')
1561
+ return 'no-binary'
1562
+ }
1563
+
1564
+ let danglingFound = false
1565
+ const resolvedBinaries = []
1566
+
1567
+ for (const binPath of binaries) {
1568
+ let real = null
1569
+ try {
1570
+ real = realpathSync(binPath)
1571
+ } catch {
1572
+ danglingFound = true
1573
+ error(`Dangling binary symlink in PATH: ${binPath} points to a non-existent path`)
1574
+ continue
1575
+ }
1576
+
1577
+ let binVersion = null
1578
+ let curDir = dirname(real)
1579
+ for (let i = 0; i < 4; i++) {
1580
+ const candidatePkg = join(curDir, 'package.json')
1581
+ if (existsSync(candidatePkg)) {
1582
+ try {
1583
+ const p = readJson(candidatePkg)
1584
+ if (p?.name === 'pantheon-opencode') {
1585
+ binVersion = p.version
1586
+ break
1587
+ }
1588
+ } catch {}
1589
+ }
1590
+ curDir = dirname(curDir)
1591
+ }
1592
+
1593
+ resolvedBinaries.push({ path: binPath, real, version: binVersion })
1594
+ }
1595
+
1596
+ if (danglingFound) return 'dangling'
1597
+
1598
+ if (resolvedBinaries.length > 1) {
1599
+ const primary = resolvedBinaries[0]
1600
+ const others = resolvedBinaries.slice(1)
1601
+ const divergent = others.filter(
1602
+ (b) => b.version && primary.version && b.version !== primary.version,
1603
+ )
1604
+
1605
+ if (divergent.length > 0) {
1606
+ warn(
1607
+ `Multiple global pantheon-opencode installs detected across PATH prefixes:\n` +
1608
+ resolvedBinaries
1609
+ .map(
1610
+ (b, idx) =>
1611
+ ` ${idx === 0 ? '▶ (active)' : ' (shadowed)'} ${b.path} (v${b.version ?? 'unknown'})`,
1612
+ )
1613
+ .join('\n') +
1614
+ `\nStale global copies cause version skew. Clean up stale prefixes with: npm uninstall -g pantheon-opencode`,
1615
+ { code: 'path_shadowing_detected', classification: 'advisory' },
1616
+ )
1617
+ return 'shadowing'
1618
+ }
1619
+ }
1620
+
1621
+ pass(`Primary binary in PATH resolves cleanly (${binaries[0]})`)
1622
+ return 'ok'
1623
+ }
1624
+
1517
1625
  // ---------------------------------------------------------------------------
1518
1626
  // Check H3: Plugin version drift (issue #158)
1519
1627
  // ---------------------------------------------------------------------------
@@ -3053,6 +3161,7 @@ async function main() {
3053
3161
  checkCodeModeDir(args)
3054
3162
  checkCodeModeManifest(args)
3055
3163
  checkInstallVersionDrift(args)
3164
+ checkPathShadowing(args)
3056
3165
  checkPluginVersionDrift(args)
3057
3166
  await checkTuiPlugin(args)
3058
3167
  await checkInstallerFileInventory(args)
@@ -642,6 +642,22 @@ export function resolveStablePantheonRoot(root = ROOT) {
642
642
  'pantheon-opencode',
643
643
  )
644
644
  if (existsSync(join(globalCandidate, 'src', 'plugin', 'index.ts'))) {
645
+ try {
646
+ const rootPkg = readJsonConfig(join(root, 'package.json'))
647
+ const globalPkg = readJsonConfig(join(globalCandidate, 'package.json'))
648
+ if (rootPkg?.version && globalPkg?.version) {
649
+ if (rootPkg.version === globalPkg.version) {
650
+ return { root: globalCandidate, volatile: false }
651
+ }
652
+ return {
653
+ root,
654
+ volatile: true,
655
+ skew: { running: rootPkg.version, global: globalPkg.version },
656
+ }
657
+ }
658
+ } catch {
659
+ // Fallback if package.json read fails
660
+ }
645
661
  return { root: globalCandidate, volatile: false }
646
662
  }
647
663
  return { root, volatile: true }
@@ -150,6 +150,8 @@ const GLOBAL_NPM_ENV_KEYS = new Set([
150
150
  'npm_config_globalconfig',
151
151
  'npm_config_prefix',
152
152
  'npm_config_location',
153
+ 'npm_config_allow_scripts',
154
+ 'npm_config_approve_scripts',
153
155
  ])
154
156
 
155
157
  /**
@@ -200,6 +200,15 @@ export function mapFileToSlice(filePath, cwd = process.cwd()) {
200
200
  export function resolveTestSlice(files = [], options = {}) {
201
201
  const cwd = options.cwd || process.cwd()
202
202
  const failedOnly = options.failedOnly
203
+ const trigger =
204
+ options.trigger ||
205
+ (options.crossModule
206
+ ? 'cross-module'
207
+ : options.security
208
+ ? 'security-boundary'
209
+ : options.migration
210
+ ? 'schema-migration'
211
+ : undefined)
203
212
 
204
213
  // Filter for relevant source or test files
205
214
  const relevantFiles = files.filter((f) => {
@@ -207,8 +216,19 @@ export function resolveTestSlice(files = [], options = {}) {
207
216
  return SOURCE_EXTENSIONS.has(ext) || TEST_EXTENSIONS.has(ext) || f.includes('.test.')
208
217
  })
209
218
 
210
- // 1. If repo has a predefined focused script and multiple files changed, prefer it
219
+ // 1. If an explicit escalation trigger is present and repo has a suite selector, escalate (Issue #444)
211
220
  const repoScript = findRepoSliceScript(cwd)
221
+ if (trigger && repoScript) {
222
+ return {
223
+ command: repoScript.command,
224
+ type: 'package-suite-escalation',
225
+ selection: `package suite (escalation trigger: ${trigger})`,
226
+ files: relevantFiles,
227
+ scopeJustification: `Escalated from local slice due to ${trigger}`,
228
+ }
229
+ }
230
+
231
+ // 2. If repo has a predefined focused script and multiple files changed, prefer it
212
232
  if (repoScript && (relevantFiles.length === 0 || relevantFiles.length > 2)) {
213
233
  return {
214
234
  command: repoScript.command,
@@ -229,7 +249,7 @@ export function resolveTestSlice(files = [], options = {}) {
229
249
  if (slice.type.startsWith('python')) {
230
250
  finalCommand = `pytest ${slice.testFile} -k "${failedOnly}" -q --tb=short`
231
251
  } else if (slice.type.startsWith('node')) {
232
- finalCommand = `node --test ${slice.testFile} --test-name-pattern="${failedOnly}"`
252
+ finalCommand = `node --test --test-name-pattern="${failedOnly}" ${slice.testFile}`
233
253
  }
234
254
  }
235
255
  return {
@@ -299,6 +319,16 @@ export function executeTestSlice(slicePlan, options = {}) {
299
319
  const passLines = output.split('\n').filter((line) => /✔|PASS|passed/i.test(line))
300
320
  const targets = Math.max(1, passLines.length)
301
321
 
322
+ const validationScope =
323
+ options.validationScope ||
324
+ (slicePlan.type === 'package-suite-escalation'
325
+ ? 'package_suite'
326
+ : slicePlan.type === 'repo-selector'
327
+ ? 'package_suite'
328
+ : 'focused_slice')
329
+ const pendingValidation =
330
+ options.pendingValidation || 'none (CI gates deferred to repository pipeline)'
331
+
302
332
  const receipt = {
303
333
  status: isPass ? 'PASS' : 'FAIL',
304
334
  command: slicePlan.command,
@@ -306,11 +336,13 @@ export function executeTestSlice(slicePlan, options = {}) {
306
336
  durationSec,
307
337
  durationMs,
308
338
  selection: slicePlan.selection,
339
+ validationScope,
340
+ pendingValidation,
309
341
  exitCode: result.status,
310
342
  output: isPass ? output.trim().split('\n').slice(-5).join('\n') : output,
311
343
  }
312
344
 
313
- const summaryLine = `**tests:** [${receipt.status}] \`${receipt.command}\` (targets: ${receipt.targets}, duration: ${receipt.durationSec}s, selection: ${receipt.selection})`
345
+ const summaryLine = `**tests:** [${receipt.status}] \`${receipt.command}\` (targets: ${receipt.targets}, duration: ${receipt.durationSec}s, selection: ${receipt.selection})\n**validation_scope:** ${validationScope}\n**pending_validation:** ${pendingValidation}`
314
346
 
315
347
  return { ...receipt, summaryLine }
316
348
  }
package/skills-lock.json CHANGED
@@ -34,7 +34,7 @@
34
34
  "src/skills/tdd-with-agents/references/advanced-testing-patterns.md": "4aac82748bb670a61d67732168f8a265d5e66895a9183029e3da03de8e024788",
35
35
  "src/skills/tdd-with-agents/references/testing-by-layer.md": "545291301661e57d846b2db2acca1edb1219a102dce4aad7e0c3cc1a250b570a",
36
36
  "src/skills/tdd-with-agents/SKILL.md": "107495b0473a561dd570d2c25420329d68712776524a736ccc415171f7aa6c00",
37
- "src/skills/test-slicing/SKILL.md": "7361f197ab6e59a55003eb571c294f1a07388ee1d3f96b1dbfb69a84276b1bb5",
37
+ "src/skills/test-slicing/SKILL.md": "3d6f78719ee7d426f8082fd703c26a0433804c60fd3cd181d3a6886c4e94075b",
38
38
  "src/skills/verification-planning/SKILL.md": "6a3c6cb5bd5076862d369885bf0b3ba086b625495fe7a42f3f5116d07b470f3e",
39
39
  "src/skills/visual-review-pipeline/references/findings-schema.md": "776511757c0d8387fab232fced8ac9ce031a03f62b6ca539e3c38b0d45035c8e",
40
40
  "src/skills/visual-review-pipeline/references/workflow-steps.md": "13796a32076dd4d04d37a50927169929ba44a332302bd98f72ec61482ceedc09",
@@ -16,6 +16,9 @@ permission:
16
16
  question: allow
17
17
  task:
18
18
  "*": deny
19
+ todowrite: allow
20
+ todoread: allow
21
+ execute: allow
19
22
  mcp_tools:
20
23
  pantheon-resources: all
21
24
  pantheon-memory: [memory_search]
@@ -88,7 +91,8 @@ You are a frontend implementation specialist. You BUILD UI. You do NOT design ar
88
91
 
89
92
  ### Before Implementation
90
93
  1. Inspect relevant component/test/API files directly using `grep`, `glob`, and `read`; when broad or unfamiliar discovery materially reduces risk, send the discovery need to Zeus for routing
91
- 2. Read relevant frontend standards; load `skill: visual-review-pipeline` only when visual behavior changed
94
+ 2. Read relevant frontend standards; load `skill: visual-review-pipeline`
95
+ 3. When the handoff brief names an installed skill (or relevant project skill exists in `.opencode/skills/` or `.agents/skills/`), read its `SKILL.md` before writing code only when visual behavior changed
92
96
  3. Plan component/data flow when scope or risk warrants it; do not require a design phase for a bounded fix
93
97
  4. For a supplied image/screenshot, use the original reference and Zeus's visual notes as acceptance context; call `vision_describe` if the image is addressable but not directly visible. One pass per image: `vision_describe` for visual review, `vision_ocr` only when text extraction is the goal — never two vision calls on the same screenshot (each call is real gateway spend). Inspect the current state first and preserve unrelated existing work—do not delete or revert it just to match the reference unless that replacement is explicitly in scope.
94
98
  - Investigate beyond the assigned context only when necessary to resolve the task or when broader discovery materially reduces risk; verify changed behavior proportionally to its scope and risk, using focused checks for bounded changes and broader checks only when integration boundaries or risk warrant them.
@@ -15,6 +15,9 @@ permission:
15
15
  bash: deny
16
16
  task:
17
17
  "*": deny
18
+ todowrite: allow
19
+ todoread: allow
20
+ execute: deny
18
21
  mcp_tools:
19
22
  pantheon-resources: all
20
23
  pantheon-memory:
@@ -18,6 +18,8 @@ permission:
18
18
  task:
19
19
  "*": deny
20
20
  apollo: allow
21
+ todoread: allow
22
+ execute: deny
21
23
  mcp_tools:
22
24
  pantheon-resources: all
23
25
  pantheon-memory:
@@ -14,6 +14,9 @@ permission:
14
14
  webfetch: allow
15
15
  task:
16
16
  "*": deny
17
+ todowrite: allow
18
+ todoread: allow
19
+ execute: allow
17
20
  mcp_tools:
18
21
  pantheon-resources: all
19
22
  pantheon-memory: [memory_search]
@@ -79,6 +82,7 @@ A child session has a parent and must not dispatch another Pantheon agent. Inspe
79
82
  ### Before Migration
80
83
  1. Inspect relevant schema/migrations directly; ask Zeus to route any needed broad discovery to @apollo when the schema is broad or unfamiliar enough that parallel discovery reduces risk
81
84
  2. Read the relevant model/migration context to understand relationships
85
+ 3. When a skill is specified in the brief or exists in `.opencode/skills/` (e.g. database/ORM conventions), read its `SKILL.md` before generating migrations
82
86
  3. Plan migration: what changes, impact on existing data, rollback strategy
83
87
 
84
88
  ### Migration Development (TDD)
@@ -14,6 +14,9 @@ permission:
14
14
  bash: deny
15
15
  task:
16
16
  "*": deny
17
+ todowrite: allow
18
+ todoread: allow
19
+ execute: deny
17
20
  mcp_tools:
18
21
  pantheon-resources: all
19
22
  pantheon-memory: [memory_recall]
@@ -13,6 +13,9 @@ permission:
13
13
  webfetch: allow
14
14
  task:
15
15
  "*": deny
16
+ todowrite: allow
17
+ todoread: allow
18
+ execute: allow
16
19
  mcp_tools:
17
20
  pantheon-resources: all
18
21
  pantheon-memory: [memory_search]
@@ -15,6 +15,9 @@ permission:
15
15
  task:
16
16
  "*": deny
17
17
  apollo: allow
18
+ todowrite: allow
19
+ todoread: allow
20
+ execute: allow
18
21
  mcp_tools:
19
22
  pantheon-resources: all
20
23
  pantheon-memory: [memory_search]
@@ -161,7 +164,7 @@ See `skill: tdd-with-agents` for the full TDD cycle.
161
164
 
162
165
  ## Implementation & Verification Workflow
163
166
 
164
- 1. **Focused Context**: Inspect targeted files directly using `grep`, `glob`, and `read`.
167
+ 1. **Focused Context**: Inspect targeted files directly using `grep`, `glob`, and `read`. When the brief specifies a skill (or a relevant project skill exists in `.opencode/skills/`), **read its `SKILL.md` first** before implementing.
165
168
  2. **Implementation**: Implement clean async endpoints and services using Pydantic schemas, dependency injection, and parameterized queries.
166
169
  3. **Verification**: Run focused unit tests via `bash` using test slicing (e.g. `pytest tests/test_item.py -k test_create -q --tb=short`). Never run full unbounded repository suites during implementation loops.
167
170
  4. **Dependencies**: Use modern stdlib equivalents (`zoneinfo`, `tomllib`) where applicable; pin stable LTS versions.
@@ -18,6 +18,9 @@ permission:
18
18
  hashline_edit: deny
19
19
  task:
20
20
  "*": deny
21
+ todowrite: allow
22
+ todoread: allow
23
+ execute: deny
21
24
  mcp_tools:
22
25
  pantheon-resources: all
23
26
  pantheon-memory: [memory_recall]
@@ -18,6 +18,9 @@ permission:
18
18
  ".pantheon/deepwork/**": allow
19
19
  task:
20
20
  "*": deny
21
+ todowrite: allow
22
+ todoread: allow
23
+ execute: deny
21
24
  mcp_tools:
22
25
  pantheon-resources: all
23
26
  pantheon-memory:
package/src/agents/nyx.md CHANGED
@@ -14,6 +14,9 @@ permission:
14
14
  hashline_edit: deny
15
15
  task:
16
16
  "*": deny
17
+ todowrite: allow
18
+ todoread: allow
19
+ execute: allow
17
20
  mcp_tools:
18
21
  pantheon-resources: all
19
22
  pantheon-memory: [memory_search]
@@ -14,6 +14,9 @@ permission:
14
14
  webfetch: allow
15
15
  task:
16
16
  "*": deny
17
+ todowrite: allow
18
+ todoread: allow
19
+ execute: allow
17
20
  mcp_tools:
18
21
  pantheon-resources: all
19
22
  pantheon-memory: [memory_search]
@@ -13,6 +13,9 @@ permission:
13
13
  grep: allow
14
14
  task:
15
15
  "*": deny
16
+ todowrite: allow
17
+ todoread: allow
18
+ execute: allow
16
19
  mcp_tools:
17
20
  pantheon-resources: all
18
21
  pantheon-memory: [memory_recall]
@@ -14,6 +14,9 @@ permission:
14
14
  pantheon_deepwork: allow
15
15
  task:
16
16
  "*": deny
17
+ todowrite: allow
18
+ todoread: allow
19
+ execute: allow
17
20
  mcp_tools:
18
21
  pantheon-resources: all
19
22
  pantheon-memory: [memory_search]
@@ -19,6 +19,7 @@ permission:
19
19
  todoread: allow
20
20
  task:
21
21
  "*": allow
22
+ execute: deny
22
23
  mcp_tools:
23
24
  pantheon-resources: all
24
25
  pantheon-memory: [memory_search]
@@ -81,6 +82,7 @@ skills:
81
82
  - **DO NOT invoke @athena.** DO NOT force planning or ask for plan approval on direct coding requests.
82
83
  - **DO NOT invoke Council synthesis.** Council is reserved strictly for `/pantheon-council` or explicit multi-perspective trade-off requests.
83
84
  - When the specialist returns, report the concise deliverable and test result directly to the user.
85
+ - **Skill-Informed Handoff**: When a task involves an installed skill (built-in Pantheon skill or external in `.opencode/skills/` / `.agents/skills/`), Zeus includes `Skill: <name> (Path: .opencode/skills/<name>/SKILL.md)` in the brief and instructs the specialist to read `SKILL.md` before implementation.
84
86
  - **Asynchronous parallel subagents**: When running long test suites, heavy builds, or independent exploratory research, use `task(..., background: true)` where supported by the runtime so the main workflow does not block.
85
87
  - **Planning via @athena is strictly gated**:
86
88
  - ONLY invoke Athena when the user explicitly runs `/pantheon-deepwork` or explicitly requests a plan in the prompt ("plan this first", "create a plan", "planeje").
@@ -125,7 +127,7 @@ These rules are economic, measured against the 2026-11-05 usage audit. They appl
125
127
  * Granular choice: `"Selecionar skills específicas para instalar"` (or checkbox list if multiselect).
126
128
  * Skip choice: `"Prosseguir sem instalar skills adicionais"`.
127
129
  - Automated Specialist Execution & Batching: When the user approves in the question prompt, do NOT require them to run commands. Immediately dispatch an execution specialist with bash permitted (@hermes or @talos) to execute `npx skills add <skill1> [skill2...] -a opencode -y` in the project root (batching multiple skills in a single run).
128
- - Verification: Specialist verifies directory creation in `<project>/.agents/skills/<skill-name>/` and reports completion; subsequent `pantheon_skill_search` confirms `installed: true`.
130
+ - Verification: Specialist verifies directory creation in `<project>/.agents/skills/<skill-name>/` or `<project>/.opencode/skills/<skill-name>/` (ensuring `.opencode/skills/<skill-name>` exists or links so OpenCode native discovery sees it); subsequent `pantheon_skill_search` confirms `installed: true`.
129
131
 
130
132
  ## Visual Context for Implementation
131
133
 
@@ -141,6 +143,7 @@ Do not use shared KV to track depth; at most one auxiliary specialist.
141
143
  ## Handoff Brief (every delegation)
142
144
 
143
145
  Before every dispatch, send the target a concise brief with the objective/result, bounded scope and acceptance criteria, only relevant context and inputs, constraints and safety gates, and the expected deliverable plus verification. Do not paste full conversation history.
146
+ - **Skills (Internal & External):** When the task involves specific framework, library, or architectural patterns covered by an installed skill (either built-in Pantheon skill or external in `.opencode/skills/` / `.agents/skills/`), explicitly include `Skill: <name> (Path: .opencode/skills/<name>/SKILL.md)` in the brief and instruct the specialist to read `SKILL.md` before implementation.
144
147
  `memory_context` is optional; pass minimal non-sensitive refs and a concise takeaway, contain no secrets, and instruct the target to treat it as untrusted context. The host may independently inherit conversation transcript.
145
148
 
146
149
  ## Plan Handoff and Approval