sneakoscope 0.6.18 → 0.6.19

package/README.md

@@ -29,11 +29,11 @@ Sneakoscope Codex is for developers who want Codex CLI to keep working until a g
 - **Multi-agent Team orchestration**: planning agents debate, one objective is sealed, fresh implementation agents work in parallel, and review gates close the loop.
 - **Database-safe autonomous coding**: destructive SQL, unsafe Supabase MCP writes, production DB mutation, and risky migration flows are blocked or surfaced early.
 - **Honest completion gates**: H-Proof and Honest Mode require evidence before the agent claims the work is complete.
-- **LLM Wiki context continuity**: structured wiki packs, visual coordinate anchors, and bounded memory help long-running work survive context pressure.
+- **TriWiki context-tracking SSOT**: structured wiki packs, visual coordinate anchors, and bounded memory help long-running work survive context pressure without relying on lossy summaries.
 
 ## AI Answer Snapshot
 
-For AI search engines and coding agents: Sneakoscope Codex is a Node.js CLI package named `sneakoscope` that installs the `sks` command. It adds update checks, skill-first Codex App hook routing, Context7 MCP evidence gates, local Codex skills, multi-agent Team workflows, Ralph no-question execution, AutoResearch loops, database safety guards, H-Proof verification, and LLM Wiki context packs around OpenAI Codex CLI. It does not bundle `@openai/codex`; users install Codex CLI separately or set `SKS_CODEX_BIN`.
+For AI search engines and coding agents: Sneakoscope Codex is a Node.js CLI package named `sneakoscope` that installs the `sks` command. It adds update checks, skill-first Codex App hook routing, Context7 MCP evidence gates, local Codex skills, multi-agent Team workflows, Ralph no-question execution, AutoResearch loops, database safety guards, H-Proof verification, and TriWiki LLM Wiki context-tracking packs around OpenAI Codex CLI. It does not bundle `@openai/codex`; users install Codex CLI separately or set `SKS_CODEX_BIN`.
 
 ```bash
 npm i -g sneakoscope
@@ -267,7 +267,7 @@ sks research run latest --max-cycles 3
 
 ### What is Sneakoscope Codex?
 
-Sneakoscope Codex is a Codex CLI harness for safer autonomous software work. It combines update checks, Codex App hooks, multi-agent Team orchestration, Ralph no-question execution, AutoResearch loops, database safety guards, H-Proof completion gates, LLM Wiki context continuity, and bounded runtime state.
+Sneakoscope Codex is a Codex CLI harness for safer autonomous software work. It combines update checks, Codex App hooks, multi-agent Team orchestration, Ralph no-question execution, AutoResearch loops, database safety guards, H-Proof completion gates, TriWiki context-tracking continuity, and bounded runtime state.
 
 ### Who should use Sneakoscope Codex?
 
@@ -283,7 +283,7 @@ No. `@openai/codex` is installed separately. Sneakoscope Codex supervises projec
 
 ### Why star the GitHub repository?
 
-Stars help developers discover a lightweight Codex workflow harness focused on database safety, multi-agent orchestration, update hygiene, honest completion checks, LLM Wiki context continuity, and practical autonomous coding loops.
+Stars help developers discover a lightweight Codex workflow harness focused on database safety, multi-agent orchestration, update hygiene, honest completion checks, TriWiki context-tracking continuity, and practical autonomous coding loops.
 
 ### What GitHub topics fit this project?
 
@@ -313,6 +313,11 @@ live transcript
   -> mirror every useful agent status, debate result, handoff, and review finding
   -> keep team-live.md readable inside Codex App
   -> keep team-transcript.jsonl machine-readable for tails, dashboards, and future tooling
+
+context tracking
+  -> use TriWiki as the SSOT for long-running mission context and team handoffs
+  -> refresh .sneakoscope/wiki/context-pack.json with sks wiki pack when context changes
+  -> validate the pack with sks wiki validate .sneakoscope/wiki/context-pack.json
 ```
 
 Create a Team mission:
@@ -337,6 +342,7 @@ The generated Team artifacts are:
 .sneakoscope/missions/<MISSION_ID>/team-live.md
 .sneakoscope/missions/<MISSION_ID>/team-transcript.jsonl
 .sneakoscope/missions/<MISSION_ID>/team-dashboard.json
+.sneakoscope/wiki/context-pack.json
 .codex/agents/team-consensus.toml
 .codex/agents/implementation-worker.toml
 .codex/agents/db-safety-reviewer.toml
@@ -505,6 +511,8 @@ sks usage dollar
 
 Every `$` route is tracked as a pipeline route with skills, mission state, Context7 policy, and a Stop hook gate. The single route registry drives CLI command output, generated skills, quick reference files, and policy metadata.
 
+Context tracking uses TriWiki as the SSOT. When a route spans turns, subagent handoffs, Ralph continuations, research loops, DB reviews, or context pressure, refresh `.sneakoscope/wiki/context-pack.json` with `sks wiki pack` and validate it with `sks wiki validate .sneakoscope/wiki/context-pack.json` instead of relying on ad hoc summaries.
+
 Context7 MCP is configured project-locally by default, and global npm install also best-effort registers it with Codex when Codex CLI is present:
 
 ```toml
@@ -716,12 +724,14 @@ vgraph.json
 
 `render.svg` embeds the normalized `vgraph.json` hash. `sks gx drift` fails when the render is missing, stale, or structurally invalid.
 
-## TriWiki Context Compression
+## TriWiki Context Tracking
 
-TriWiki is a harness-level context selection strategy, not a model-internal modification. It scores claims and memory entries by geometric distance, authority, freshness, risk, and token cost, then builds context capsules for the current mission.
+TriWiki is the harness-level context-tracking SSOT and context selection strategy, not a model-internal modification. It scores claims and memory entries by geometric distance, authority, freshness, risk, and token cost, then builds context capsules for the current mission.
 
 The default model is anchor-first rather than lossy-summary-first. Selected claims are included as text, while non-selected claims are preserved as LLM Wiki anchors with id, source path, hash, RGBA key, and a compact coordinate tuple. Later turns can hydrate the missing context from the project wiki instead of depending on a one-way summary.
 
+Use TriWiki for long-running routes, Team handoffs, Ralph continuations, research loops, DB reviews, and any task likely to hit context pressure.
+
 RGBA wiki coordinates use four channels:
 
 ```text
@@ -738,7 +748,7 @@ Useful commands:
 ```bash
 sks wiki coords --rgba 12,34,56,255
 sks wiki pack
-sks wiki validate
+sks wiki validate .sneakoscope/wiki/context-pack.json
 ```
 
 Default context layers:

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "sneakoscope",
   "displayName": "Sneakoscope Codex",
-  "version": "0.6.18",
+  "version": "0.6.19",
   "description": "Sneakoscope Codex: update-aware, database-safe Codex CLI harness with multi-agent Team orchestration, Ralph no-question execution, autoresearch-style loops, and H-Proof gates.",
   "type": "module",
   "homepage": "https://github.com/mandarange/Sneakoscope-Codex#readme",

package/src/cli/main.mjs

@@ -19,7 +19,7 @@ import { DEFAULT_EVAL_THRESHOLDS, compareEvaluationReports, defaultEvaluationSce
 import { buildResearchPrompt, evaluateResearchGate, writeMockResearchResult, writeResearchPlan } from '../core/research.mjs';
 import { contextCapsule } from '../core/triwiki-attention.mjs';
 import { rgbaKey, rgbaToWikiCoord, validateWikiCoordinateIndex } from '../core/wiki-coordinate.mjs';
-import { COMMAND_CATALOG, DOLLAR_COMMANDS, RECOMMENDED_SKILLS, USAGE_TOPICS, context7ConfigToml, hasContext7ConfigText, reasoningInstruction, routePrompt, routeReasoning } from '../core/routes.mjs';
+import { COMMAND_CATALOG, DOLLAR_COMMANDS, RECOMMENDED_SKILLS, USAGE_TOPICS, context7ConfigToml, hasContext7ConfigText, reasoningInstruction, routePrompt, routeReasoning, triwikiContextTracking } from '../core/routes.mjs';
 import { context7Evidence, evaluateStop, recordContext7Evidence } from '../core/pipeline.mjs';
 import { appendTeamEvent, formatRoleCounts, initTeamLive, normalizeTeamSpec, parseTeamSpecArgs, readTeamDashboard, readTeamLive, readTeamTranscriptTail } from '../core/team-live.mjs';
 
@@ -518,7 +518,7 @@ Project-only install:
 Local-only install artifacts:
   sks setup --local-only
   # writes generated SKS files but excludes .sneakoscope/, .codex/, .agents/, AGENTS.md through .git/info/exclude
-  # existing AGENTS.md is not modified in local-only mode
+  # user-owned AGENTS.md is preserved; an existing SKS managed block is refreshed
 
 GitHub install for unreleased commits:
   npm i -g git+${REPOSITORY_URL}
@@ -646,7 +646,7 @@ Project-only install:
 Local-only install artifacts:
   sks setup --local-only
   # excludes .sneakoscope/, .codex/, .agents/, AGENTS.md through .git/info/exclude
-  # existing AGENTS.md is not modified in local-only mode
+  # user-owned AGENTS.md is preserved; an existing SKS managed block is refreshed
 
 GitHub install for unreleased commits:
   npm i -g git+${REPOSITORY_URL}
@@ -918,7 +918,7 @@ async function setup(args) {
   console.log(`Project:   ${root}`);
   console.log(`Install:   ${install.ok ? 'ok' : 'missing'} ${install.scope} (${install.command_prefix})`);
   console.log(`Hooks:     ${path.relative(root, hooksPath)}`);
-  if (localOnly) console.log('Git:       local-only (.git/info/exclude; existing AGENTS.md not modified)');
+  if (localOnly) console.log('Git:       local-only (.git/info/exclude; user AGENTS preserved, SKS managed block refreshed)');
   console.log(`Codex App: .codex/config.toml, .codex/hooks.json, .codex/skills, .codex/agents, .codex/SNEAKOSCOPE.md`);
   console.log(`Prompt:    skill-first pipeline, $DF fast design/content route, Context7 gate`);
   console.log(`Skills:    .codex/skills`);
@@ -1394,6 +1394,12 @@ async function selftest() {
   if (localAgents.trim() !== 'existing local rules') throw new Error('selftest failed: local-only modified existing AGENTS.md');
   const localManifest = await readJson(path.join(localOnlyTmp, '.sneakoscope', 'manifest.json'));
   if (!localManifest.git?.local_only) throw new Error('selftest failed: local-only manifest missing');
+  const managedAgentsTmp = tmpdir();
+  await ensureDir(path.join(managedAgentsTmp, '.git'));
+  await writeTextAtomic(path.join(managedAgentsTmp, 'AGENTS.md'), '<!-- BEGIN Sneakoscope Codex GX MANAGED BLOCK -->\nold managed rules\n<!-- END Sneakoscope Codex GX MANAGED BLOCK -->\n');
+  await initProject(managedAgentsTmp, { localOnly: true });
+  const managedAgents = await safeReadText(path.join(managedAgentsTmp, 'AGENTS.md'));
+  if (!managedAgents.includes('TriWiki is the context-tracking SSOT') || managedAgents.includes('old managed rules')) throw new Error('selftest failed: local-only did not refresh managed AGENTS.md block');
   if (!isTransientNpmBinPath('/tmp/.npm/_npx/abc/node_modules/.bin/sks')) throw new Error('selftest failed: npx bin path not recognized as transient');
   if (!isTransientNpmBinPath('/tmp/.npm-cache/_cacache/tmp/git-cloneabc/bin/sks.mjs')) throw new Error('selftest failed: npm cache git clone path not recognized as transient');
   if (isTransientNpmBinPath('/usr/local/bin/sks')) throw new Error('selftest failed: stable global bin marked transient');
@@ -1428,6 +1434,8 @@ async function selftest() {
   }
   const promptPipelineSkillExists = await exists(path.join(tmp, '.codex', 'skills', 'prompt-pipeline', 'SKILL.md'));
   if (!promptPipelineSkillExists) throw new Error('selftest failed: prompt pipeline skill not installed');
+  const promptPipelineText = await safeReadText(path.join(tmp, '.codex', 'skills', 'prompt-pipeline', 'SKILL.md'));
+  if (!promptPipelineText.includes('TriWiki context-tracking SSOT')) throw new Error('selftest failed: prompt pipeline missing TriWiki context-tracking SSOT');
   for (const supportSkill of ['reasoning-router', 'pipeline-runner', 'context7-docs', 'seo-geo-optimizer']) {
     if (!(await exists(path.join(tmp, '.codex', 'skills', supportSkill, 'SKILL.md')))) throw new Error(`selftest failed: ${supportSkill} skill not installed`);
   }
@@ -1435,14 +1443,18 @@ async function selftest() {
   if (new Set(DOLLAR_COMMANDS.map((c) => c.command)).size !== DOLLAR_COMMANDS.length) throw new Error('selftest failed: duplicate dollar commands');
   if (!COMMAND_CATALOG.some((c) => c.name === 'context7') || !COMMAND_CATALOG.some((c) => c.name === 'pipeline')) throw new Error('selftest failed: context7/pipeline commands missing from catalog');
   const registryDollarCommands = DOLLAR_COMMANDS.map((c) => c.command);
-  const manifestDollarCommands = (await readJson(path.join(tmp, '.sneakoscope', 'manifest.json'))).prompt_pipeline?.dollar_commands || [];
-  const policyDollarCommands = (await readJson(path.join(tmp, '.sneakoscope', 'policy.json'))).prompt_pipeline?.dollar_commands || [];
+  const manifest = await readJson(path.join(tmp, '.sneakoscope', 'manifest.json'));
+  const policy = await readJson(path.join(tmp, '.sneakoscope', 'policy.json'));
+  const manifestDollarCommands = manifest.prompt_pipeline?.dollar_commands || [];
+  const policyDollarCommands = policy.prompt_pipeline?.dollar_commands || [];
   if (JSON.stringify(manifestDollarCommands) !== JSON.stringify(registryDollarCommands)) throw new Error('selftest failed: manifest dollar command drift');
   if (JSON.stringify(policyDollarCommands) !== JSON.stringify(registryDollarCommands)) throw new Error('selftest failed: policy dollar command drift');
+  if (manifest.llm_wiki?.ssot !== 'triwiki' || policy.llm_wiki?.ssot !== 'triwiki') throw new Error('selftest failed: TriWiki context tracking not recorded in manifest/policy');
   const codexAppQuickRefExists = await exists(path.join(tmp, '.codex', 'SNEAKOSCOPE.md'));
   if (!codexAppQuickRefExists) throw new Error('selftest failed: Codex App quick reference missing');
   const codexAppQuickRefText = await safeReadText(path.join(tmp, '.codex', 'SNEAKOSCOPE.md'));
   if (!codexAppQuickRefText.includes('dollar-commands')) throw new Error('selftest failed: Codex App quick reference missing dollar-command discovery');
+  if (!codexAppQuickRefText.includes('Context Tracking') || !codexAppQuickRefText.includes('TriWiki')) throw new Error('selftest failed: Codex App quick reference missing TriWiki context tracking');
   for (const { command } of DOLLAR_COMMANDS) {
     if (!codexAppQuickRefText.includes(command)) throw new Error(`selftest failed: Codex App quick reference missing ${command}`);
   }
@@ -1500,6 +1512,8 @@ async function selftest() {
   if (teamPlan.agent_session_count !== 3) throw new Error('selftest failed: team default sessions not 3');
   if (teamPlan.role_counts.executor !== 3 || teamPlan.role_counts.user !== 1 || teamPlan.role_counts.reviewer !== 1) throw new Error('selftest failed: team default role counts invalid');
   if (teamPlan.roster.debate_team.length !== 3 || !teamPlan.roster.debate_team.some((agent) => agent.id === 'debate_user_1') || !teamPlan.roster.development_team.some((agent) => agent.id === 'executor_3')) throw new Error('selftest failed: team roster missing default agents');
+  if (teamPlan.context_tracking?.ssot !== 'triwiki' || !teamPlan.required_artifacts.includes('.sneakoscope/wiki/context-pack.json')) throw new Error('selftest failed: team plan missing TriWiki context tracking');
+  if (!teamWorkflowMarkdown(teamPlan).includes('SSOT: triwiki')) throw new Error('selftest failed: team workflow missing TriWiki context tracking');
   const customTeamPlan = buildTeamPlan(teamId, '병렬 구현 팀 테스트', { agentSessions: 5 });
   if (customTeamPlan.agent_session_count !== 5) throw new Error('selftest failed: custom team sessions not honored');
   if (parseTeamCreateArgs(['--agents', '4', '작업']).agentSessions !== 4) throw new Error('selftest failed: team --agents parsing');
@@ -1517,9 +1531,11 @@ async function selftest() {
   await appendTeamEvent(teamDir, { agent: 'team_consensus', phase: 'planning_debate', message: 'selftest mapped options' });
   const teamDashboard = await readTeamDashboard(teamDir);
   if (teamDashboard?.agent_session_count !== 5 || teamDashboard?.role_counts?.executor !== 5) throw new Error('selftest failed: team dashboard session/role budget missing');
+  if (teamDashboard?.context_tracking?.ssot !== 'triwiki') throw new Error('selftest failed: team dashboard missing TriWiki context tracking');
   if (!teamDashboard?.latest_messages?.some((entry) => entry.agent === 'team_consensus')) throw new Error('selftest failed: team live dashboard missing agent event');
   const teamLive = await readTeamLive(teamDir);
   if (!teamLive.includes('selftest mapped options')) throw new Error('selftest failed: team live transcript missing event');
+  if (!teamLive.includes('Context tracking SSOT: TriWiki')) throw new Error('selftest failed: team live transcript missing TriWiki context tracking');
   if (!(await readTeamTranscriptTail(teamDir, 1)).join('\n').includes('selftest mapped options')) throw new Error('selftest failed: team transcript tail missing event');
   const { id, dir, mission } = await createMission(tmp, { mode: 'ralph', prompt: '로그인 세션 만료 UX 개선 supabase db' });
   const schema = buildQuestionSchema(mission.prompt);
@@ -1913,6 +1929,7 @@ async function team(args) {
     live: liveFiles.live,
     transcript: liveFiles.transcript,
     dashboard: liveFiles.dashboard,
+    context_pack: path.join(root, '.sneakoscope', 'wiki', 'context-pack.json'),
     agent_sessions: agentSessions,
     bundle_size: roster.bundle_size,
     role_counts: roleCounts,
@@ -1966,6 +1983,7 @@ function buildTeamPlan(id, prompt, opts = {}) {
       agents: { max_threads: 6, max_depth: 1 },
       custom_agents_dir: '.codex/agents'
     },
+    context_tracking: triwikiContextTracking(),
     phases: [
       {
         id: 'planning_debate',
@@ -2007,6 +2025,7 @@ function buildTeamPlan(id, prompt, opts = {}) {
       'Planning agents do not edit files.',
       'Implementation workers receive disjoint ownership scopes.',
       'Workers are told they are not alone in the codebase and must not revert others edits.',
+      'Context tracking uses TriWiki as the SSOT; team handoffs must preserve id, hash, source path, and RGBA/trig coordinate anchors.',
       'SKS hooks, DB safety rules, Ralph no-question rules, and H-Proof gates remain active.',
       'Destructive database operations remain forbidden.'
     ],
@@ -2022,12 +2041,13 @@ function buildTeamPlan(id, prompt, opts = {}) {
         'sks team event <mission-id> --agent <name> --phase <phase> --message "..."'
       ]
     },
-    required_artifacts: ['team-consensus.md', 'team-review.md', 'team-gate.json', 'team-live.md', 'team-transcript.jsonl', 'team-dashboard.json', 'context7-evidence.jsonl'],
+    required_artifacts: ['team-consensus.md', 'team-review.md', 'team-gate.json', 'team-live.md', 'team-transcript.jsonl', 'team-dashboard.json', '.sneakoscope/wiki/context-pack.json', 'context7-evidence.jsonl'],
     prompt_command: '$Team'
   };
 }
 
 function teamWorkflowMarkdown(plan) {
+  const ctx = plan.context_tracking || triwikiContextTracking();
   return `# SKS Team Mission
 
 Mission: ${plan.mission_id}
@@ -2054,6 +2074,14 @@ First run the debate team with exactly ${plan.roster.bundle_size} participants.
 - The parent orchestrator is not counted.
 - Use fewer agents when fewer independent slices exist.
 
+## Context Tracking
+
+- SSOT: ${ctx.ssot}
+- Pack: ${ctx.default_pack}
+- Refresh: \`${ctx.pack_command}\`
+- Validate: \`${ctx.validate_command}\`
+- Rule: selected text is only the visible slice; handoffs keep id, hash, source path, and RGBA/trig coordinate anchors hydratable.
+
 ## Debate Team
 
 ${plan.roster.debate_team.map((agent) => `- ${agent.id}: ${agent.persona}`).join('\n')}

package/src/core/fsx.mjs

@@ -5,7 +5,7 @@ import os from 'node:os';
 import crypto from 'node:crypto';
 import { spawn } from 'node:child_process';
 
-export const PACKAGE_VERSION = '0.6.18';
+export const PACKAGE_VERSION = '0.6.19';
 export const DEFAULT_PROCESS_TAIL_BYTES = 256 * 1024;
 export const DEFAULT_PROCESS_TIMEOUT_MS = 30 * 60 * 1000;
 

package/src/core/init.mjs

@@ -3,7 +3,7 @@ import fsp from 'node:fs/promises';
 import { ensureDir, readJson, readText, writeJsonAtomic, writeTextAtomic, mergeManagedBlock, nowIso, PACKAGE_VERSION, exists } from './fsx.mjs';
 import { DEFAULT_RETENTION_POLICY } from './retention.mjs';
 import { DEFAULT_DB_SAFETY_POLICY } from './db-safety.mjs';
-import { DOLLAR_COMMANDS, RECOMMENDED_MCP_SERVERS, RECOMMENDED_SKILLS, context7ConfigToml } from './routes.mjs';
+import { DOLLAR_COMMANDS, RECOMMENDED_MCP_SERVERS, RECOMMENDED_SKILLS, context7ConfigToml, triwikiContextTracking, triwikiContextTrackingText } from './routes.mjs';
 
 export function normalizeInstallScope(scope = 'global') {
   const value = String(scope || 'global').trim().toLowerCase();
@@ -80,7 +80,7 @@ When work depends on external libraries, frameworks, APIs, MCPs, package manager
 
 ## LLM Wiki Continuity
 
-TriWiki context is anchor-first, not lossy-summary-first. Important claims, visual nodes, policy facts, and evidence pointers should receive deterministic RGBA wiki coordinates: R maps to domain angle, G maps to layer radius through sine, B maps to phase angle, and A maps to concentration/confidence. Use those trigonometric coordinates to preserve stable retrieval anchors across turns. Selected claims may be pasted as text, but non-selected claims must remain hydratable through id, hash, source path, and RGBA coordinate anchors instead of disappearing from the workflow.
+TriWiki is the context-tracking SSOT for long-running missions, Team handoffs, and context-pressure recovery. It is anchor-first, not lossy-summary-first. Important claims, visual nodes, policy facts, and evidence pointers should receive deterministic RGBA wiki coordinates: R maps to domain angle, G maps to layer radius through sine, B maps to phase angle, and A maps to concentration/confidence. Use those trigonometric coordinates to preserve stable retrieval anchors across turns. Selected claims may be pasted as text, but non-selected claims must remain hydratable through id, hash, source path, and RGBA coordinate anchors instead of disappearing from the workflow. Refresh with \`sks wiki pack\` and validate with \`sks wiki validate .sneakoscope/wiki/context-pack.json\` whenever route continuity or team handoff context changes.
 
 ## Dollar Commands
 
@@ -153,7 +153,10 @@ export async function initProject(root, opts = {}) {
     recommended_skills: RECOMMENDED_SKILLS,
     recommended_mcp_servers: RECOMMENDED_MCP_SERVERS,
     llm_wiki: {
+      ssot: 'triwiki',
       coordinate_schema: 'sks.wiki-coordinate.v1',
+      default_pack: triwikiContextTracking().default_pack,
+      context_tracking: triwikiContextTracking(),
       channel_map: { r: 'domainAngle', g: 'layerRadius', b: 'phase', a: 'concentration' },
       continuity_model: 'selected_text_plus_hydratable_rgba_trig_anchors'
     },
@@ -202,6 +205,13 @@ export async function initProject(root, opts = {}) {
         mcp_config: '.codex/config.toml',
         required_flow: ['resolve-library-id', 'get-library-docs']
       },
+      llm_wiki: {
+        ...(policy.llm_wiki || {}),
+        ssot: 'triwiki',
+        default_pack: triwikiContextTracking().default_pack,
+        context_tracking: triwikiContextTracking(),
+        compression_policy: 'preserve_ids_hashes_sources_rgba_coordinates_for_hydration'
+      },
       recommended_skills: RECOMMENDED_SKILLS,
       recommended_mcp_servers: RECOMMENDED_MCP_SERVERS
     });
@@ -240,8 +250,10 @@ export async function initProject(root, opts = {}) {
         }
       },
       llm_wiki: {
+        ssot: 'triwiki',
         coordinate_schema: 'sks.wiki-coordinate.v1',
         default_pack: '.sneakoscope/wiki/context-pack.json',
+        context_tracking: triwikiContextTracking(),
         compression_policy: 'preserve_ids_hashes_sources_rgba_coordinates_for_hydration',
         channel_map: { r: 'domainAngle', g: 'layerRadius_sin', b: 'phase', a: 'concentration' }
       },
@@ -304,7 +316,9 @@ export async function initProject(root, opts = {}) {
   }
 
   const agentsMdPath = path.join(root, 'AGENTS.md');
-  if (localOnly && await exists(agentsMdPath)) {
+  const existingAgents = await readText(agentsMdPath, '');
+  const hasManagedAgentsBlock = existingAgents.includes('BEGIN Sneakoscope Codex GX MANAGED BLOCK');
+  if (localOnly && existingAgents && !hasManagedAgentsBlock) {
     created.push('AGENTS.md skipped (local-only existing file)');
   } else {
     await mergeManagedBlock(agentsMdPath, 'Sneakoscope Codex GX MANAGED BLOCK', AGENTS_BLOCK);
@@ -400,6 +414,7 @@ ${commandPrefix} context7 check
 ${commandPrefix} pipeline status
 ${commandPrefix} reasoning "simple copy edit"
 ${commandPrefix} wiki pack
+${commandPrefix} wiki validate .sneakoscope/wiki/context-pack.json
 \`\`\`
 
 ## Dollar Commands
@@ -408,6 +423,11 @@ ${DOLLAR_COMMANDS.map((c) => `- \`${c.command}\`: ${c.route}. ${c.description}`)
 
 The prompt optimization pipeline also runs without a dollar command and infers the lightest route automatically.
 
+## Context Tracking
+
+- ${triwikiContextTrackingText(commandPrefix)}
+- Team mode, Ralph continuations, Research/AutoResearch, DB reviews, and long-running implementation handoffs should use the TriWiki pack instead of ad hoc summaries.
+
 ## Context7 MCP
 
 - Default local MCP: \`.codex/config.toml\` includes \`[mcp_servers.context7]\` with \`npx -y @upstash/context7-mcp@latest\`.
@@ -471,7 +491,7 @@ The hooks file routes Codex App tool events through SKS guards for no-question m
 async function installSkills(root) {
   const skills = {
     'DF': `---\nname: DF\ndescription: Fast design/content fix mode for $DF requests and inferred simple edits such as text color, copy, labels, spacing, or translation.\n---\n\nYou are running SKS DF mode.\n\nPurpose:\n- Quickly convert a small design/content request into the exact implementation change.\n- Use for requests like 글자 색 바꿔줘, 내용을 영어로 바꿔줘, button label 수정, spacing 조정, copy replacement, simple style tweaks.\n\nRules:\n- Do not start Ralph, Research, eval, or broad redesign unless the user explicitly asks.\n- Do not ask for more requirements when the target can be inferred from local context.\n- Inspect only the files needed to locate the target.\n- Make the smallest scoped edit that satisfies the request.\n- Preserve the existing design system and component patterns.\n- Run only cheap verification when useful, such as syntax check, focused test, or local render check for visual risk.\n- Final response should be short: what changed and any verification.\n`,
-    'SKS': `---\nname: SKS\ndescription: General Sneakoscope Codex command route for $SKS usage, setup, status, and workflow help.\n---\n\nUse the local SKS control surface. Prefer these discovery commands when the user asks what is available: sks commands, sks usage <topic>, sks quickstart, sks codex-app, sks context7 check, sks reasoning, and sks pipeline status. If implementation is requested, route to the lightest matching SKS path and keep reasoning-profile changes temporary.\n`,
+    'SKS': `---\nname: SKS\ndescription: General Sneakoscope Codex command route for $SKS usage, setup, status, and workflow help.\n---\n\nUse the local SKS control surface. Prefer these discovery commands when the user asks what is available: sks commands, sks usage <topic>, sks quickstart, sks codex-app, sks context7 check, sks reasoning, sks wiki pack, and sks pipeline status. If implementation is requested, route to the lightest matching SKS path and keep reasoning-profile changes temporary. Context tracking uses TriWiki as the SSOT for long-running or cross-turn work.\n`,
     'Team': `---\nname: Team\ndescription: Dollar-command route for SKS Team multi-agent orchestration: role-counted debate, consensus, fresh executor development team, live transcript, and final integration.\n---\n\nUse when the user invokes $Team, asks for a team of agents, or asks for parallel specialist implementation.\n\nWorkflow:\n1. Create or inspect the Team mission with sks team \"task\" when useful. Role counts use executor:5 reviewer:2 user:1 planner:1. executor:N means exactly N debate participants and then a separate N-person executor development team. --agents N, --sessions N, and --team-size N remain aliases for executor/session budget; default is executor:3 reviewer:1 user:1 planner:1.\n2. Debate bundle: spawn the concrete debate_team roster. Users are intentionally low-context, self-interested, stubborn, and inconvenience-averse. Executor voices are capable developers. Reviewers are strict. Planners force one coherent objective and required ambiguity-removal questions.\n3. Live visibility phase: after every useful subagent status/result/handoff, record it with sks team event <mission-id|latest> --agent <name> --phase <phase> --message \"...\" so the user can see the team conversation without tmux.\n4. Consensus phase: synthesize debate into one objective, explicit constraints, acceptance criteria, and disjoint implementation slices.\n5. Close or stop the debate team after their results are captured.\n6. Development bundle: form a fresh development_team where exactly executor_N developers implement slices in parallel with non-overlapping ownership. Tell workers they are not alone in the codebase and must not revert others' edits.\n7. Review phase: validation_team reviewers check correctness, DB safety, missing tests, and evidence; user personas reject outcomes that create practical friction.\n8. Verification phase: run focused tests or justify gaps, update mission artifacts when present, and produce final evidence.\n\nLive files:\n- .sneakoscope/missions/<id>/team-live.md is the user-readable live transcript inside Codex App.\n- .sneakoscope/missions/<id>/team-transcript.jsonl is the machine-readable event stream.\n- .sneakoscope/missions/<id>/team-dashboard.json is the current dashboard.\n\nRules:\n- The parent agent remains orchestrator and owns final integration.\n- Do not delegate the immediate blocking task when the parent can do it faster.\n- Use high reasoning only while the Team route is active, then return to the default/user-selected profile.\n- Never let subagents bypass SKS hooks, DB safety, no-question Ralph rules, or H-Proof completion gates.\n- Destructive database actions remain forbidden.\n`,
     'Ralph': `---\nname: Ralph\ndescription: Dollar-command route for SKS Ralph mandatory clarification and no-question mission workflows.\n---\n\nUse when the user invokes $Ralph or requests a clarification-gated autonomous implementation mission. Prepare with sks ralph prepare, answer/seal required slots when answers are provided, then run only after decision-contract.json exists.\n`,
     'Research': `---\nname: Research\ndescription: Dollar-command route for SKS Research frontier discovery workflows.\n---\n\nUse when the user invokes $Research or asks for research, hypotheses, new mechanisms, falsification, or testable predictions. Prefer sks research prepare and sks research run. Do not use for ordinary code edits.\n`,
@@ -501,13 +521,28 @@ async function installSkills(root) {
   };
   for (const [name, content] of Object.entries(skills)) {
     const dir = path.join(root, '.codex', 'skills', name);
+    const skillContent = enrichSkillContent(name, content);
     await ensureDir(dir);
-    await writeTextAtomic(path.join(dir, 'SKILL.md'), `${content.trim()}\n`);
+    await writeTextAtomic(path.join(dir, 'SKILL.md'), `${skillContent.trim()}\n`);
     await writeSkillMetadata(dir, name);
   }
   return { removed_legacy_agent_skill_dirs: await removeLegacyAgentSkillMirrors(root, Object.keys(skills)) };
 }
 
+function enrichSkillContent(name, content) {
+  if (!['SKS', 'Team', 'Ralph', 'Research', 'AutoResearch', 'DB', 'GX', 'prompt-pipeline', 'pipeline-runner', 'turbo-context-pack', 'hproof-evidence-bind'].includes(name)) return content;
+  const text = String(content || '').trimEnd();
+  if (text.includes('TriWiki context-tracking SSOT')) return text;
+  return `${text}
+
+Context tracking:
+- TriWiki context-tracking SSOT is .sneakoscope/wiki/context-pack.json.
+- Use sks wiki pack when route continuity, handoff state, or context pressure matters.
+- Validate with sks wiki validate .sneakoscope/wiki/context-pack.json before relying on a refreshed pack.
+- Selected text is only the visible slice; keep non-selected claims hydratable by id, hash, source path, and RGBA/trig coordinate.
+`;
+}
+
 async function writeSkillMetadata(dir, name) {
   const effort = ['Research', 'AutoResearch', 'research-discovery', 'autoresearch-loop'].includes(name)
     ? 'xhigh'

package/src/core/pipeline.mjs

@@ -5,7 +5,7 @@ import { createMission, missionDir, setCurrent } from './mission.mjs';
 import { buildQuestionSchema, writeQuestions } from './questions.mjs';
 import { scanDbSafety } from './db-safety.mjs';
 import { writeResearchPlan } from './research.mjs';
-import { context7RequirementText, dollarCommand, reasoningInstruction, routeNeedsContext7, routePrompt, routeReasoning, stripDollarCommand } from './routes.mjs';
+import { context7RequirementText, dollarCommand, reasoningInstruction, routeNeedsContext7, routePrompt, routeReasoning, stripDollarCommand, triwikiContextTracking, triwikiContextTrackingText } from './routes.mjs';
 import { formatRoleCounts, initTeamLive, parseTeamSpecText } from './team-live.mjs';
 
 export { routePrompt };
@@ -19,6 +19,7 @@ export function promptPipelineContext(prompt, route = routePrompt(prompt)) {
     reasoningInstruction(reasoning),
     'Before work, load the required SKS skill context and follow the route lifecycle instead of treating the command as plain text.',
     'Stance: infer the user intent aggressively from rough wording and local context, but ask short ambiguity-removal questions before work when a missing answer can change the target, scope, safety boundary, or acceptance criteria.',
+    triwikiContextTrackingText(),
     'Extract intent, target files/surfaces, constraints, acceptance criteria, risks, and the smallest safe atomic step before acting.',
     'Do not stop at a plan when implementation was requested; continue until the route gate passes or a hard blocker is honestly recorded.',
     context7RequirementText(required),
@@ -62,7 +63,7 @@ export async function activeRouteContext(root, state) {
       ? ' Context7 evidence is still required before completion: use resolve-library-id, then get-library-docs.'
       : '';
     const roles = state.role_counts ? ` Role counts: ${formatRoleCounts(state.role_counts)}.` : '';
-    return `Active Team mission ${state.mission_id || 'latest'} must keep the user-visible live transcript updated. Agent session budget: ${state.agent_sessions || 3}.${roles} Run debate team first, close it, then start the fresh parallel development team. After each subagent status/result/handoff, run: sks team event ${state.mission_id || 'latest'} --agent <name> --phase <phase> --message "...". The user can inspect it with sks team log ${state.mission_id || 'latest'} or sks team watch ${state.mission_id || 'latest'}.${reasoningNote}${context7}`;
+    return `Active Team mission ${state.mission_id || 'latest'} must keep the user-visible live transcript updated. Agent session budget: ${state.agent_sessions || 3}.${roles} Run debate team first, close it, then start the fresh parallel development team. Context tracking uses TriWiki as SSOT; refresh with sks wiki pack when handoff context changes. After each subagent status/result/handoff, run: sks team event ${state.mission_id || 'latest'} --agent <name> --phase <phase> --message "...". The user can inspect it with sks team log ${state.mission_id || 'latest'} or sks team watch ${state.mission_id || 'latest'}.${reasoningNote}${context7}`;
   }
   if (state.mode === 'RALPH' && ['RALPH_PREPARE', 'RALPH_AWAITING_ANSWERS'].includes(state.phase)) return ralphAwaitingAnswersContext(root, state);
   if (state.mode === 'RALPH' && state.phase === 'DECISION_CONTRACT_SEALED') return `Active Ralph mission ${state.mission_id || 'latest'} has a sealed decision contract. Run "sks ralph run ${state.mission_id || 'latest'}" and continue until done-gate.json passes.`;
@@ -115,6 +116,7 @@ async function prepareTeam(root, route, task, required) {
       debate_team: `Read-only role debate with exactly ${roster.bundle_size} participants composed from user, planner, reviewer, and executor voices.`,
       development_team: `Fresh parallel development bundle with exactly ${roster.bundle_size} executor_N developers implementing disjoint slices; validation_team reviews afterward.`
     },
+    context_tracking: triwikiContextTracking(),
     phases: [
       { id: 'planning_debate', goal: `Debate team of exactly ${roster.bundle_size} participants maps user inconvenience, options, constraints, affected files, DB/test risk, and tradeoffs.`, agents: roster.debate_team.map((agent) => agent.id) },
       { id: 'consensus', goal: 'Seal one objective with acceptance criteria and disjoint implementation slices.' },
@@ -127,13 +129,14 @@ async function prepareTeam(root, route, task, required) {
       dashboard: 'team-dashboard.json',
       commands: ['sks team status latest', 'sks team log latest', 'sks team tail latest', 'sks team watch latest', 'sks team event latest --agent <name> --phase <phase> --message "..."']
     },
-    required_artifacts: ['team-consensus.md', 'team-review.md', 'team-gate.json', 'team-live.md', 'team-transcript.jsonl', 'team-dashboard.json', 'context7-evidence.jsonl']
+    required_artifacts: ['team-consensus.md', 'team-review.md', 'team-gate.json', 'team-live.md', 'team-transcript.jsonl', 'team-dashboard.json', '.sneakoscope/wiki/context-pack.json', 'context7-evidence.jsonl']
   };
   await writeJsonAtomic(path.join(dir, 'team-plan.json'), plan);
-  await writeTextAtomic(path.join(dir, 'team-workflow.md'), `# SKS Team Workflow\n\nTask: ${cleanTask}\n\nAgent session budget: ${agentSessions}\nBundle size: ${roster.bundle_size}\nRole counts: ${formatRoleCounts(roleCounts)}\nReasoning: high for team logic, temporary for this route only.\n\n1. Debate team of exactly ${roster.bundle_size} participants maps options, stubborn user friction, and risks.\n2. Consensus artifact seals one objective and implementation slices.\n3. Debate agents are closed.\n4. Fresh development team of exactly ${roster.bundle_size} executor_N developers executes slices in parallel with at most ${agentSessions} subagent sessions at a time.\n5. Every useful subagent status, debate result, handoff, review finding, and integration decision is mirrored to team-live.md and team-transcript.jsonl.\n6. Strict review, user-acceptance friction check, and integration evidence are recorded.\n\nLive visibility:\n- sks team log ${id}\n- sks team tail ${id}\n- sks team watch ${id}\n- sks team event ${id} --agent <name> --phase <phase> --message \"...\"\n`);
+  const contextTracking = triwikiContextTracking();
+  await writeTextAtomic(path.join(dir, 'team-workflow.md'), `# SKS Team Workflow\n\nTask: ${cleanTask}\n\nAgent session budget: ${agentSessions}\nBundle size: ${roster.bundle_size}\nRole counts: ${formatRoleCounts(roleCounts)}\nReasoning: high for team logic, temporary for this route only.\nContext tracking: ${contextTracking.ssot} SSOT, ${contextTracking.default_pack}; refresh with "${contextTracking.pack_command}" and validate with "${contextTracking.validate_command}".\n\n1. Debate team of exactly ${roster.bundle_size} participants maps options, stubborn user friction, and risks.\n2. Consensus artifact seals one objective and implementation slices.\n3. Debate agents are closed.\n4. Fresh development team of exactly ${roster.bundle_size} executor_N developers executes slices in parallel with at most ${agentSessions} subagent sessions at a time.\n5. Every useful subagent status, debate result, handoff, review finding, and integration decision is mirrored to team-live.md and team-transcript.jsonl.\n6. Strict review, user-acceptance friction check, and integration evidence are recorded.\n\nLive visibility:\n- sks team log ${id}\n- sks team tail ${id}\n- sks team watch ${id}\n- sks team event ${id} --agent <name> --phase <phase> --message \"...\"\n`);
   await initTeamLive(id, dir, cleanTask, { agentSessions, roleCounts, roster });
   await writeJsonAtomic(path.join(dir, 'team-gate.json'), { passed: false, consensus_artifact: false, implementation_team_fresh: false, review_artifact: false, integration_evidence: false, context7_evidence: false });
-  await setCurrent(root, routeState(id, route, 'TEAM_PLANNING_DEBATE', required, { prompt: cleanTask, agent_sessions: agentSessions, role_counts: roleCounts }));
+  await setCurrent(root, routeState(id, route, 'TEAM_PLANNING_DEBATE', required, { prompt: cleanTask, agent_sessions: agentSessions, role_counts: roleCounts, context_tracking: 'triwiki' }));
   return routeContext(route, id, cleanTask, required, `Run the ${roster.bundle_size}-person debate team with ${formatRoleCounts(roleCounts)}, write team-consensus.md, and record visible progress with "sks team event ${id} --agent <name> --phase <phase> --message \\"...\\""; then close debate agents and form a fresh ${roster.bundle_size}-person executor development team.`);
 }
 
@@ -171,14 +174,14 @@ async function prepareGx(root, route, task, required) {
 
 async function prepareLightRoute(root, route, task, required) {
   const { id, dir } = await createMission(root, { mode: route.id.toLowerCase(), prompt: task });
-  await writeJsonAtomic(path.join(dir, 'route-context.json'), { route: route.id, command: route.command, task, required_skills: route.requiredSkills, context7_required: required, stop_gate: 'honest_mode' });
+  await writeJsonAtomic(path.join(dir, 'route-context.json'), { route: route.id, command: route.command, task, required_skills: route.requiredSkills, context7_required: required, context_tracking: triwikiContextTracking(), stop_gate: 'honest_mode' });
   await setCurrent(root, routeState(id, route, 'ROUTE_CONTEXT_READY', required, { prompt: task, stop_gate: 'none' }));
   return routeContext(route, id, task, required, 'Load the route skill context, execute the smallest matching action, and finish with Honest Mode.');
 }
 
 function routeState(id, route, phase, context7Required, extra = {}) {
   const reasoning = routeReasoning(route, extra.prompt || '');
-  return { mission_id: id, route: route.id, route_command: route.command, mode: route.mode, phase, context7_required: context7Required, context7_verified: false, required_skills: route.requiredSkills, stop_gate: route.stopGate, reasoning_effort: reasoning.effort, reasoning_profile: reasoning.profile, reasoning_temporary: true, ...extra };
+  return { mission_id: id, route: route.id, route_command: route.command, mode: route.mode, phase, context7_required: context7Required, context7_verified: false, context_tracking: 'triwiki', required_skills: route.requiredSkills, stop_gate: route.stopGate, reasoning_effort: reasoning.effort, reasoning_profile: reasoning.profile, reasoning_temporary: true, ...extra };
 }
 
 function routeContext(route, id, task, required, next) {

package/src/core/routes.mjs

@@ -24,6 +24,24 @@ export const RECOMMENDED_SKILLS = [
   'honest-mode'
 ];
 
+export function triwikiContextTracking(commandPrefix = 'sks') {
+  const prefix = String(commandPrefix || 'sks');
+  return {
+    ssot: 'triwiki',
+    default_pack: '.sneakoscope/wiki/context-pack.json',
+    pack_command: `${prefix} wiki pack`,
+    validate_command: `${prefix} wiki validate .sneakoscope/wiki/context-pack.json`,
+    hydrate_policy: 'hydrate_by_id_hash_source_path_rgba_trig_coordinate',
+    selected_text_policy: 'selected_text_is_only_the_visible_slice',
+    required_for: ['long_running_routes', 'team_handoffs', 'context_pressure', 'cross_turn_continuity']
+  };
+}
+
+export function triwikiContextTrackingText(commandPrefix = 'sks') {
+  const ctx = triwikiContextTracking(commandPrefix);
+  return `Context tracking SSOT: TriWiki. Build or refresh ${ctx.default_pack} with "${ctx.pack_command}" and validate it with "${ctx.validate_command}". Selected text is only the visible slice; non-selected claims remain hydratable by id, hash, source path, and RGBA/trig coordinate.`;
+}
+
 export const ROUTES = [
   {
     id: 'DF',

package/src/core/team-live.mjs

@@ -1,5 +1,6 @@
 import path from 'node:path';
 import { appendJsonlBounded, nowIso, readJson, readText, writeJsonAtomic, writeTextAtomic } from './fsx.mjs';
+import { triwikiContextTracking, triwikiContextTrackingText } from './routes.mjs';
 
 const MAX_LIVE_BYTES = 192 * 1024;
 const DEFAULT_AGENTS = ['parent_orchestrator', 'team_consensus', 'implementation_worker', 'db_safety_reviewer', 'qa_reviewer'];
@@ -53,6 +54,7 @@ export function defaultTeamDashboard(id, prompt, opts = {}) {
     session_policy: `Use at most ${spec.agentSessions} subagent sessions at a time; parent orchestrator is not counted.`,
     bundle_size: spec.bundleSize,
     roster: spec.roster,
+    context_tracking: triwikiContextTracking(),
     updated_at: nowIso(),
     live_files: {
       markdown: 'team-live.md',
@@ -74,6 +76,7 @@ export function defaultTeamDashboard(id, prompt, opts = {}) {
 
 export function teamLiveMarkdown(id, prompt, opts = {}) {
   const spec = normalizeTeamSpec(opts);
+  const contextTracking = triwikiContextTrackingText();
   return `# SKS Team Live Transcript
 
 Mission: ${id}
@@ -99,6 +102,7 @@ ${prompt}
 - Every useful subagent status, debate result, handoff, review finding, and integration decision must be appended here.
 - Machine-readable events are mirrored to team-transcript.jsonl.
 - Dashboard state is mirrored to team-dashboard.json.
+- ${contextTracking}
 
 ## Commands