spectoflow 0.18.0 → 0.20.0

package/README.md

@@ -20,6 +20,12 @@ An **agent-agnostic** spec-driven development framework with a **real-time local
 You speak in plain language; the framework classifies your intent and runs the right workflow. No
 ceremonial command to start.
 
+**Works with whichever coding agent you have** — Claude Code, Codex, Cursor, Gemini CLI, OpenCode,
+Kiro CLI, and Antigravity are auto-detected at `init`. The dashboard's topbar always shows the
+**active agent**, front and center, with a switcher: pick another and it's verified as genuinely
+installed before activating (a red **"No agent found"** if none is) — it never silently activates
+something that isn't there.
+
 ## Install
 
 ```bash
@@ -39,13 +45,17 @@ Every command works both ways — `spectoflow <cmd>` when installed globally, or
 
 ```
 spectoflow init [dir] [--agent=claude,codex]   scaffold a project (auto-detects agents; wires Playwright MCP)
-spectoflow update [--dry-run]                  refresh framework files to this kit version
+spectoflow update [--dry-run] [--force|-f]     refresh framework files to this kit version
 spectoflow status                              progress + whether the dashboard is running
 
 spectoflow dashboard [--port=NNNN]             start the control plane in the background (hands the prompt back)
 spectoflow dashboard status                    is it running? (url + pid)
 spectoflow dashboard stop   (or: stop)         stop the running dashboard
 spectoflow dashboard restart                   stop then start
+spectoflow dashboard create "..." | --auto     generate a custom dashboard
+
+spectoflow skill create "..." | --auto         generate a project skill
+spectoflow agent create "..." | --auto         generate a project agent
 
 spectoflow list                                agents, skills and the workflow at a glance
 spectoflow agents                              list the team personas
@@ -79,7 +89,10 @@ update` refreshes **framework-owned** files (engine, dashboard, `AGENTS.md`, `ca
 `policy.md`, default agents & skills) to the CLI's version, while **preserving your work** —
 `config.json`, `workflow.md`, `specs/`, `plans/`, and any agent/skill you created or edited are never
 touched. A file you edited is preserved and its new version is written next to it as `<file>.new`
-for you to merge by hand. Add `--dry-run` to preview.
+for you to merge by hand. Add `--dry-run` to preview, or `--force`/`-f` to overwrite a diverged file
+in place instead of dropping a `.new` — use it once you're sure you have no local edits worth keeping
+in that file (e.g. it's been stuck diverged since an earlier update); it still never touches
+`config.json`, `workflow.md`, `specs/` or `plans/`.
 
 First refresh the kit, then run update from your project — the flow is the same whichever way you
 installed:
@@ -154,9 +167,11 @@ A floating **💬 chat widget** (bottom-right, redesigned) and the **Chat** tab
 itself by printing `::spectoflow role=… kind=… msg=…` sentinels, which become labelled messages
 (analyst / developer / qa …); other output streams raw. The board refreshes live as it edits plans.
 Either surface can also **Orchestrate** the enabled workflow: each step runs its agent, gated by mode
-+ policy. The Agents & Skills drawer is served by the one read-only endpoint, `GET
-/api/agentfile?path=` (scoped to `.spectoflow/agents/**` + `.spectoflow/skills/**`,
-path-traversal-safe) — the framework's only other server surface is unchanged.
++ policy. The Chat tab can also **Summarize** the recent log into one digest (via the active agent)
+or **Clear** it outright, when it's grown long. The Agents & Skills drawer is served by the one
+read-only endpoint, `GET /api/agentfile?path=` (scoped to `.spectoflow/agents/**` +
+`.spectoflow/skills/**`, path-traversal-safe) — the framework's only other server surface is
+unchanged.
 
 **Customize.** Settings → **Customize** lets you extend the project's own spectoflow install: add a
 dashboard, a skill, or an agent by describing what you want (or hit **Auto** to have the agent survey
@@ -167,7 +182,17 @@ uses, so a generated dashboard automatically matches whatever design is active,
 keeps matching if you switch designs later. Blocks can bind live to project stats (`bind:
 "phases.0.pct"`) or hold a static value. Generated skills and agents follow the same gold-standard
 shape as the shipped ones, cite real domain standards (OWASP, WCAG, C4/ADR, …) instead of generic
-advice, and are marked `origin: user-generated` so they're easy to tell apart in the UI.
+advice, and are marked `origin: user-generated` so they're easy to tell apart in the UI. The same
+generators are available from the terminal:
+
+```bash
+spectoflow skill create "reviews PRs for accessibility"      # or: --auto to propose candidates
+spectoflow agent create "owns accessibility review"          # or: --auto
+spectoflow dashboard create "a KPI overview for support"     # or: --auto
+```
+
+Each streams the agent's run live and exits with its status — the same pipeline the dashboard's
+Generate/Auto buttons use, just from a shell.
 
 ## Agents vs skills
 

package/bin/spectoflow.js

@@ -10,6 +10,8 @@ const detect = require('../lib/detect');
 const ownership = require('../lib/ownership');
 const manifest = require('../lib/manifest');
 const mcp = require('../lib/mcp');
+const { startRun } = require('../templates/dashboard/runner');
+const { buildCustomizePrompt } = require('../templates/lib/customize-prompts');
 
 const KIT = path.resolve(__dirname, '..');
 const TPL = path.join(KIT, 'templates');
@@ -224,10 +226,11 @@ function update() {
     return console.log('No spectoflow project here. Run: spectoflow init');
   }
   const dryRun = argv.includes('--dry-run');
-  const r = require('../lib/update').runUpdate({ projectRoot: root, templatesDir: TPL, version: VERSION, dryRun });
+  const force = argv.includes('--force') || argv.includes('-f');
+  const r = require('../lib/update').runUpdate({ projectRoot: root, templatesDir: TPL, version: VERSION, dryRun, force });
 
   const from = r.fromVersion || 'unknown';
-  const changed = r.refreshed.length + r.created.length + r.adopted.length + r.newSidecar.length;
+  const changed = r.refreshed.length + r.created.length + r.adopted.length + r.newSidecar.length + r.forced.length;
   const row = (sym, label, list, painter, note) => {
     if (!list.length) return;
     const n = c.dim(String(list.length).padStart(2));
@@ -235,18 +238,19 @@ function update() {
     console.log(`  ${sym}  ${painter(label.padEnd(9))} ${n}   ${detail}`);
   };
   console.log(logo());
-  console.log(`  ${c.bold('spectoflow update')}   ${c.dim(from)} ${c.amber('→')} ${c.bold(r.toVersion)}${dryRun ? c.dim('   (dry-run)') : ''}`);
+  console.log(`  ${c.bold('spectoflow update')}   ${c.dim(from)} ${c.amber('→')} ${c.bold(r.toVersion)}${dryRun ? c.dim('   (dry-run)') : ''}${force ? c.y('   (force)') : ''}`);
   console.log('');
   row(c.g('✓'), 'refreshed', r.refreshed, c.g);
   row(c.cy('+'), 'created', r.created, c.cy);
   row(c.b('~'), 'adopted', r.adopted, c.b);
-  row(c.y('!'), '.new', r.newSidecar, c.y, 'you edited these — new version saved as *.new, merge by hand');
+  row(c.y('!'), 'forced', r.forced, c.y, 'overwrote a diverged file — its previous content is gone');
+  row(c.y('!'), '.new', r.newSidecar, c.y, 'you edited these — new version saved as *.new, merge by hand (or re-run with --force)');
   if (r.unchanged.length) console.log(`  ${c.dim('·')}  ${c.dim('unchanged'.padEnd(9))} ${c.dim(String(r.unchanged.length).padStart(2))}`);
   console.log(`  ${c.dim('=')}  ${c.dim('preserved'.padEnd(9))}      ${c.dim('config.json · workflow.md · specs/ · plans/ · your custom agents & skills')}`);
   console.log('');
   if (dryRun) console.log(`  ${c.dim('(dry-run — nothing was written)')}`);
   else console.log(`  ${changed ? c.g('✓ Done') : c.dim('Already up to date')}${changed ? c.dim(` · ${changed} file(s) changed`) : ''}`);
-  if (r.newSidecar.length && !dryRun) console.log(`  ${c.y('→')} ${c.dim(`${r.newSidecar.length} *.new file(s) to review and merge`)}`);
+  if (r.newSidecar.length && !dryRun) console.log(`  ${c.y('→')} ${c.dim(`${r.newSidecar.length} *.new file(s) to review and merge — or re-run: spectoflow update --force`)}`);
   console.log('');
 }
 
@@ -257,9 +261,62 @@ async function dashboard() {
   if (sub === 'stop') return stopDashboard();
   if (sub === 'status') return dashboardStatus();
   if (sub === 'restart') return restartDashboard();
+  if (sub === 'create') return runCustomize('dashboard');
   return startDashboard();
 }
 
+// ---- Customize: `spectoflow skill/agent/dashboard create` — the CLI mirror of the dashboard's
+// Settings → Customize UI. Both surfaces build the same natural-language prompt (customize-prompts.js)
+// and post it through the same pipeline (runner.js's startRun — the function /api/run itself calls),
+// so a generation triggered from the terminal behaves identically to one triggered from a click.
+function requireProjectRoot() {
+  const root = process.cwd();
+  if (!fs.existsSync(path.join(root, '.spectoflow'))) {
+    console.log('No spectoflow project here. Run: spectoflow init');
+    return null;
+  }
+  return root;
+}
+// "create <description words…> [--auto] [--agent=name]" → { description, auto, agentOverride }.
+// Words are re-joined with spaces so an unquoted multi-word description works the same as a quoted one.
+function parseCreateArgs(args) {
+  return {
+    auto: args.includes('--auto'),
+    agentOverride: (args.find((a) => a.startsWith('--agent=')) || '').split('=')[1] || undefined,
+    description: args.filter((a) => !a.startsWith('--')).join(' ').trim(),
+  };
+}
+function printCreateUsage(kind) {
+  console.log(`Usage: spectoflow ${kind} create "<description>" ${c.dim('[--agent=name]')}`);
+  console.log(`   or: spectoflow ${kind} create --auto ${c.dim('[--agent=name]')}`);
+}
+// Streams the same events the dashboard's SSE feed would show: raw output lines as-is, and
+// structured ::spectoflow sentinel messages as "[role] text" (skip the echoed user prompt — printed
+// separately, up front, so it isn't shown twice).
+function cliEmit(evt) {
+  if (evt.type === 'run-line') process.stdout.write(evt.chunk);
+  else if (evt.type === 'message' && evt.message && evt.message.role !== 'user') {
+    console.log(`${c.cy('[' + evt.message.role + ']')} ${evt.message.text}`);
+  }
+}
+async function runCustomize(kind) {
+  const root = requireProjectRoot();
+  if (!root) return;
+  if (argv[1] !== 'create') return printCreateUsage(kind);
+  const { auto, agentOverride, description } = parseCreateArgs(argv.slice(2));
+  let prompt;
+  try { prompt = buildCustomizePrompt(kind, { auto, description }); }
+  catch (e) { console.log(c.y(e.message)); console.log(''); return printCreateUsage(kind); }
+  console.log(c.dim(`→ ${prompt}`));
+  const code = await new Promise((resolve) => {
+    const r = startRun(root, { prompt, agent: agentOverride }, cliEmit);
+    if (r.error) { console.log(c.y(r.error)); return resolve(1); }
+    if (!r.child) return resolve(1); // spawn failed — cliEmit already printed the error
+    r.child.on('close', (exitCode) => resolve(exitCode == null ? 1 : exitCode));
+  });
+  process.exitCode = code;
+}
+
 // Start in the background and return control. Probes first so a second start just reports the running
 // one instead of spawning a duplicate (and never crashes on EADDRINUSE).
 async function startDashboard() {
@@ -387,7 +444,7 @@ ${c.dim('Usage:')} spectoflow ${c.g('<command>')} ${c.dim('[options]')}   ${c.di
 
 ${c.bold('Project')}
   ${c.g('init')} ${c.dim('[dir] [--agent=a,b]')}    scaffold a project (auto-detects agents; wires Playwright MCP)
-  ${c.g('update')} ${c.dim('[--dry-run]')}          refresh framework files to this kit version
+  ${c.g('update')} ${c.dim('[--dry-run|--force]')}  refresh framework files to this kit version
   ${c.g('status')}                      progress + whether the dashboard is running
 
 ${c.bold('Dashboard')}
@@ -396,6 +453,11 @@ ${c.bold('Dashboard')}
   ${c.g('dashboard stop')}               stop it ${c.dim('(alias: stop)')}
   ${c.g('dashboard restart')}            stop then start
 
+${c.bold('Customize')} ${c.dim('— same as Settings → Customize, from the terminal')}
+  ${c.g('skill create')} ${c.dim('"<description>" | --auto')}      generate a project skill
+  ${c.g('agent create')} ${c.dim('"<description>" | --auto')}      generate a project agent
+  ${c.g('dashboard create')} ${c.dim('"<description>" | --auto')}  generate a custom dashboard
+
 ${c.bold('Explore')}
   ${c.g('list')}                        agents, skills and the workflow at a glance
   ${c.g('agents')}                      list the team personas
@@ -416,17 +478,30 @@ const HELP = {
   shims; override with ${c.g('--agent=claude,codex')}. Also wires ${c.bold('Playwright MCP')} into the
   project's ${c.dim('.mcp.json')} (idempotent — never touches an existing entry).
   ${c.dim('An existing CLAUDE.md is preserved as CLAUDE.md.tomerge for you to merge on first run.')}`,
-  update: `${c.bold('spectoflow update')} ${c.dim('[--dry-run]')}\n
+  update: `${c.bold('spectoflow update')} ${c.dim('[--dry-run] [--force|-f]')}\n
   Refresh framework-owned files (engine, dashboard, default agents & skills, AGENTS.md, policy…)
   to this CLI's version, ${c.bold('preserving your work')}: config.json, workflow.md, specs/, plans/
   and any agent/skill you edited are never overwritten (an edited file's new version lands as
-  ${c.dim('*.new')} for you to merge). ${c.g('--dry-run')} previews without writing.`,
-  dashboard: `${c.bold('spectoflow dashboard')} ${c.dim('[--port=NNNN] [status|stop|restart]')}\n
+  ${c.dim('*.new')} for you to merge). ${c.g('--dry-run')} previews without writing.
+  ${c.g('--force')} (${c.g('-f')}) overwrites a diverged file in place instead of dropping a ${c.dim('*.new')}
+  — use it when you know you have no local edits worth keeping (e.g. a file stuck diverged from an
+  earlier update). It never touches config.json, workflow.md, specs/ or plans/.`,
+  dashboard: `${c.bold('spectoflow dashboard')} ${c.dim('[--port=NNNN] [status|stop|restart|create]')}\n
   Start the local control plane in the ${c.bold('background')} (default ${c.dim('4319')} or
   ${c.dim('$SPECTOFLOW_PORT')}) and hand the prompt back. Subcommands:
     ${c.g('status')}    is it running? (url + pid)
     ${c.g('stop')}      stop it            ${c.dim('(alias: spectoflow stop)')}
-    ${c.g('restart')}   stop then start`,
+    ${c.g('restart')}   stop then start
+    ${c.g('create')}    generate a custom dashboard, e.g. ${c.dim('spectoflow dashboard create "..." --auto')}`,
+  skill: `${c.bold('spectoflow skill create')} ${c.dim('"<description>" [--agent=name]')}\n${c.bold('spectoflow skill create')} ${c.dim('--auto [--agent=name]')}\n
+  Generate a project-specific skill — the CLI mirror of Settings → Customize → ${c.bold('Skills')} →
+  ${c.bold('Add skill')} in the dashboard. Describe what it should do, or pass ${c.g('--auto')} to have
+  the agent survey the project and propose candidates instead. Runs the configured agent headless
+  (${c.dim('config.json → agent')}, or override with ${c.g('--agent=')}), streaming its output live;
+  it clarifies first if the ask is ambiguous, and marks what it writes ${c.dim('origin: user-generated')}.`,
+  agent: `${c.bold('spectoflow agent create')} ${c.dim('"<description>" [--agent=name]')}\n${c.bold('spectoflow agent create')} ${c.dim('--auto [--agent=name]')}\n
+  Generate a project-specific agent — the CLI mirror of Settings → Customize → ${c.bold('Agents')} →
+  ${c.bold('Add agent')}. Same behaviour as ${c.g('spectoflow skill create')}, for an agent persona instead.`,
   status: `${c.bold('spectoflow status')}\n
   Print project progress from ${c.dim('plans/*.md')} (tasks done, specs, agents, skills, in-progress
   items) and whether the dashboard is currently running.`,
@@ -446,6 +521,8 @@ const fns = {
   agents: () => { console.log(wordmark()); printAgents(false); },
   skills: () => { console.log(wordmark()); printSkills(false); },
   workflow: () => { console.log(wordmark()); printWorkflow(false); },
+  skill: () => runCustomize('skill'),
+  agent: () => runCustomize('agent'),
 };
 const wantsHelp = argv.slice(1).some((a) => a === '-h' || a === '--help');
 

package/lib/adapters.js

@@ -67,6 +67,12 @@ Argument: \`$ARGUMENTS\`
 `;
 
 // Priority order = which agent becomes the default when several are detected.
+//
+// Considered and left out (researched, not re-added on a whim): Kimi Code CLI (MoonshotAI/kimi-cli)
+// and DeepSeek Harness (deepseek-ai/deepseek-harness) — as of Sept 2026 neither ships a genuine
+// non-interactive one-shot mode (prompt-as-trailing-arg, stdout, exit) that runner.js's spawn model
+// needs; Kimi is interactive/ACP-only, DeepSeek Harness is a local web-app framework. Revisit if
+// either ships a headless flag — a fabricated runner command would just fail silently for real users.
 const REGISTRY = [
   {
     id: 'claude',
@@ -95,6 +101,24 @@ const REGISTRY = [
     runner: 'gemini -p',
     detect: { bin: 'gemini', dirs: ['.gemini'] },
   },
+  {
+    id: 'opencode',
+    entries: [{ path: 'AGENTS.md', content: ROOT_AGENTS_MD }],
+    runner: 'opencode run --quiet',
+    detect: { bin: 'opencode', dirs: ['.opencode'] },
+  },
+  {
+    id: 'kiro',
+    entries: [{ path: 'AGENTS.md', content: ROOT_AGENTS_MD }],
+    runner: 'kiro-cli chat --no-interactive --trust-all-tools',
+    detect: { bin: 'kiro-cli', dirs: ['.kiro'] },
+  },
+  {
+    id: 'antigravity',
+    entries: [{ path: 'AGENTS.md', content: ROOT_AGENTS_MD }],
+    runner: 'agy -p',
+    detect: { bin: 'agy' },
+  },
 ];
 
 const byId = (id) => REGISTRY.find((a) => a.id === id);

package/lib/update.js

@@ -20,18 +20,20 @@ function toDisk(sf, rel) {
   return path.join(sf, rel.split('/').join(path.sep));
 }
 
-function runUpdate({ projectRoot, templatesDir, version, dryRun = false }) {
+function runUpdate({ projectRoot, templatesDir, version, dryRun = false, force = false }) {
   const sf = path.join(projectRoot, '.spectoflow');
   const prev = manifest.readManifest(sf);
   const report = {
     fromVersion: prev ? prev.version : null,
     toVersion: version,
     dryRun,
+    force,
     created: [],
     refreshed: [],
     newSidecar: [],
     adopted: [],
     unchanged: [],
+    forced: [],
   };
   const baseline = (prev && prev.files) || {};
   const nextFiles = {}; // manifest to write after this run
@@ -64,6 +66,10 @@ function runUpdate({ projectRoot, templatesDir, version, dryRun = false }) {
       write(diskPath, newBuf); // untouched framework file → safe to refresh
       report.refreshed.push(rel);
       nextFiles[rel] = newHash;
+    } else if (force) {
+      write(diskPath, newBuf); // --force: overwrite the diverged file too, no .new sidecar
+      report.forced.push(rel);
+      nextFiles[rel] = newHash;
     } else {
       write(diskPath + '.new', newBuf); // user-edited or legacy-divergent → offer, never overwrite
       report.newSidecar.push(rel);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "spectoflow",
-  "version": "0.18.0",
+  "version": "0.20.0",
   "description": "Agent-agnostic spec-driven development framework + real-time local control plane. Markdown artifacts, intent router, workflow-by-scope.",
   "keywords": [
     "spec-driven-development",

package/templates/README.md

@@ -28,14 +28,18 @@ sit at the project root and just point back here.
   ```
 - **See what you got:** `spectoflow list` (agents, skills & workflow at a glance), or `spectoflow
   agents` / `spectoflow skills` / `spectoflow workflow`. Append `-h` to any command for its help.
-- **Change how it runs** in the dashboard's **Settings** tab (autonomy mode, output language, and the
-  dashboard **design**), or by editing `config.json`.
-- **Extend spectoflow itself** from Settings → **Customize**: describe a project-specific dashboard,
+- **Change how it runs** in the dashboard's **Personalize** tab (the **active agent**, always shown
+  in the topbar too — switching is verified against what's actually installed before it activates —
+  plus autonomy mode, output language, and the dashboard **design**), or by editing `config.json`.
+- **Extend spectoflow itself** from Personalize, further down the same tab: describe a project-specific dashboard,
   skill, or agent (or hit **Auto** to have it propose candidates from your project), and it's generated
   for you — a dashboard appears in the nav immediately, a skill/agent follows the same gold-standard
-  shape as the shipped ones and is marked `origin: user-generated`.
+  shape as the shipped ones and is marked `origin: user-generated`. Same thing from the terminal:
+  `spectoflow skill create "<description>"` / `agent create` / `dashboard create` (each also takes
+  `--auto`).
 - **Update the framework** to a newer kit: `spectoflow update` (preserves your edits; a file you
-  changed is kept and its new version is written next to it as `*.new`).
+  changed is kept and its new version is written next to it as `*.new` — `--force`/`-f` overwrites it
+  in place instead, once you're sure you have nothing worth keeping there).
 
 ## Where your work lives
 

package/templates/dashboard/public/app.js

@@ -114,6 +114,17 @@ async function doOrchestrate(promptEl){
   promptEl.value='';
 }
 async function approve(decision){ await fetch('/api/orchestrate/approve',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({decision})}); }
+// ---- chat context management: condense the log via the agent, or wipe it (Chat tab only — the
+// floating widget stays "quick access", full controls live where there's room to read them) ----
+async function summarizeChat(agentEl){
+  const agent=(agentEl||$('#tabRunAgent'))?.value;
+  flash();
+  await fetch('/api/chat/summarize',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({agent})});
+}
+async function clearChat(){
+  flash();
+  await fetch('/api/chat/clear',{method:'POST'});
+}
 async function patchTask(id,patch){ flash(); await fetch('/api/task/'+encodeURIComponent(id),{method:'PATCH',headers:{'Content-Type':'application/json'},body:JSON.stringify(patch)}); }
 async function addComment(id,text,action){ flash(); await fetch('/api/task/'+encodeURIComponent(id)+'/comment',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({text,action})}); }
 async function toggleStep(name){ flash(); await fetch('/api/workflow/toggle',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({name})}); }
@@ -701,6 +712,47 @@ function fillLangSelect(sel,lang){
 }
 function setLangSelect(lang){ fillLangSelect($('#setLang'),lang); fillLangSelect($('#topLang'),lang); }
 function setModeSelects(mode){ [$('#setMode'),$('#topMode')].forEach(s=>{ if(s) s.value=mode; }); }
+
+// ---- active agent: always visible (topbar #topAgent) + editable (Settings #setAgent). Never lets
+// the user activate an agent that isn't actually installed — options for a not-installed known agent
+// are present but disabled, and P.installedAgents itself comes from the server testing each agent's
+// real CLI (bin on PATH, or the project's own config dir), not a guess.
+function fillAgentSelect(sel,known,installed,active){
+  if(!sel) return;
+  const empty=!installed.length;
+  sel.classList.toggle('is-empty',empty);
+  sel.disabled=empty;
+  if(empty){ sel.innerHTML=''; const o=document.createElement('option'); o.value=''; o.textContent=t('topbar.agent.none'); sel.append(o); return; }
+  const sig=known.map(a=>a.id+':'+(installed.includes(a.id)?1:0)).join(',');
+  if(sel.dataset.sig!==sig){
+    sel.innerHTML='';
+    known.forEach(a=>{
+      const o=document.createElement('option'); o.value=a.id; o.textContent=a.label;
+      if(!installed.includes(a.id)) o.disabled=true;
+      sel.append(o);
+    });
+    sel.dataset.sig=sig;
+  }
+  if(active) sel.value=active;
+}
+function setAgentSelects(){
+  const c=(P&&P.config)||{};
+  const known=P.knownAgents||[], installed=P.installedAgents||[];
+  fillAgentSelect($('#topAgent'),known,installed,c.agent);
+  fillAgentSelect($('#setAgent'),known,installed,c.agent);
+  const hint=$('#setAgentHint'); if(hint){ const empty=!installed.length; hint.hidden=!empty; hint.classList.toggle('is-empty',empty); if(empty) hint.textContent=t('topbar.agent.none'); }
+}
+function showAgentError(msg){
+  const hint=$('#setAgentHint'); if(!hint) return;
+  hint.hidden=false; hint.classList.add('is-empty'); hint.textContent=msg;
+  setTimeout(()=>{ if(P&&P.installedAgents&&P.installedAgents.length) hint.hidden=true; },4000);
+}
+async function saveAgent(id){
+  if(!id) return;
+  flash();
+  const r=await fetch('/api/settings',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({agent:id})});
+  if(!r.ok){ const body=await r.json().catch(()=>({})); showAgentError(body.error||t('topbar.agent.none')); setAgentSelects(); return; }
+}
 // ---- design skins (data-design) — switchable, persisted per viewer + as the project default ----
 function currentDesign(){ return document.documentElement.getAttribute('data-design')||'console'; }
 function applyDesign(id){ document.documentElement.setAttribute('data-design',id); try{ localStorage.setItem('spf-design',id); }catch{} }
@@ -710,6 +762,7 @@ function renderSettings(){
   const c=(P&&P.config)||{};
   setModeSelects(c.mode||'semi');
   setLangSelect(c.language||'en');
+  setAgentSelects();
   // design switcher — options from the DESIGNS registry (designs.js)
   const dsel=$('#setDesign');
   if(dsel){
@@ -1212,6 +1265,8 @@ $('#setLang').addEventListener('change',onLangSelectChange);
 const topModeSel=$('#topMode'); if(topModeSel) topModeSel.addEventListener('change',onModeSelectChange);
 const topLangSel=$('#topLang'); if(topLangSel) topLangSel.addEventListener('change',onLangSelectChange);
 const setDesignSel=$('#setDesign'); if(setDesignSel) setDesignSel.addEventListener('change',()=>saveDesign(setDesignSel.value));
+const topAgentSel=$('#topAgent'); if(topAgentSel) topAgentSel.addEventListener('change',(e)=>saveAgent(e.target.value));
+const setAgentSel=$('#setAgent'); if(setAgentSel) setAgentSel.addEventListener('change',(e)=>saveAgent(e.target.value));
 // theme
 (function(){ const s=localStorage.getItem('spf-theme'); if(s)document.documentElement.setAttribute('data-theme',s);
   $('#themeToggle').addEventListener('click',()=>{ const c=document.documentElement.getAttribute('data-theme'); const n=c==='dark'?'light':'dark'; document.documentElement.setAttribute('data-theme',n); localStorage.setItem('spf-theme',n); }); })();
@@ -1246,6 +1301,8 @@ $('#runPrompt').addEventListener('keydown',e=>{ if((e.metaKey||e.ctrlKey)&&e.key
 // Chat tab — same doRun/doOrchestrate/approve, its own textarea/select (#tabRunPrompt/#tabRunAgent)
 $('#tabRunBtn').addEventListener('click',()=>doRun($('#tabRunPrompt'),$('#tabRunAgent')));
 $('#tabOrchBtn').addEventListener('click',()=>doOrchestrate($('#tabRunPrompt')));
+$('#tabSummarizeBtn').addEventListener('click',()=>summarizeChat($('#tabRunAgent')));
+$('#tabClearBtn').addEventListener('click',clearChat);
 $('#tabRunPrompt').addEventListener('keydown',e=>{ if((e.metaKey||e.ctrlKey)&&e.key==='Enter')doRun($('#tabRunPrompt'),$('#tabRunAgent')); });
 $('#drawerClose').addEventListener('click',closeDrawer);
 $('#drawerScrim').addEventListener('click',closeDrawer);

package/templates/dashboard/public/i18n.js

@@ -20,18 +20,20 @@
 const I18N = {
 en: {
   'nav.board':'Board','nav.requests':'Requests','nav.attention':'Attention','nav.backlog':'Backlog',
-  'nav.workflow':'Workflow','nav.team':'Agents & Skills','nav.chat':'Chat','nav.info':'Info','nav.settings':'Settings',
+  'nav.workflow':'Workflow','nav.team':'Agents & Skills','nav.chat':'Chat','nav.info':'Info','nav.settings':'Personalize',
   'status.todo':'To do','status.in_progress':'In progress','status.to_validate':'To validate',
   'status.to_analyze':'To analyze','status.done':'Done','status.blocked':'Blocked',
   'filter.all':'All','filter.open':'Open','filter.resolved':'Resolved',
-  'action.run':'Run','action.send':'Send','action.orchestrate':'Orchestrate','action.approve':'Approve',
+  'action.run':'Run','action.send':'Send','action.orchestrate':'Orchestrate','action.approve':'Approve','action.summarize':'Summarize','action.clear':'Clear',
   'action.cancel':'Cancel','action.add':'Add','action.edit':'Edit','action.delete':'Delete',
   'action.resolve':'Resolve','action.reopen':'Reopen','action.validateToTask':'Validate → task',
   'action.addToAnalyze':'Add + to analyze',
   'topbar.navToggle':'Toggle menu','topbar.run.title':'Open the run chat',
   'topbar.theme.toggle':'Toggle light / dark theme','topbar.theme.toggle.title':'Toggle light / dark',
-  'topbar.mode.title':'Autonomy mode — change from here or in Settings',
-  'topbar.lang.title':'Output language — change from here or in Settings',
+  'topbar.mode.title':'Autonomy mode — change from here or in Personalize',
+  'topbar.lang.title':'Output language — change from here or in Personalize',
+  'topbar.agent.title':'Active agent — change from here or in Personalize',
+  'topbar.agent.none':'No agent found',
   'board.filterPlaceholder':'Filter tasks…','board.expandAll':'Expand all','board.collapseAll':'Collapse all',
   'board.expandAllTitle':'Expand or collapse all phases','board.viewList':'List','board.viewKanban':'Kanban',
   'board.viewList.title':'Grouped by phase','board.viewKanban.title':'Columns by status',
@@ -88,7 +90,7 @@ en: {
   'chat.tabSub':'Full conversation with the runner — the same run as the widget, more room to read it.',
   'chat.idle':'Type a request — the agent runs headless in this project with full memory (<code>CLAUDE.md → AGENTS.md</code>) and updates the board live.',
   'chat.inputPlaceholder':'e.g. Add a login feature with email + password',
-  'chat.orchestrateTitle':'Walk the enabled workflow',
+  'chat.orchestrateTitle':'Walk the enabled workflow','chat.summarizeTitle':'Condense the recent activity into a summary','chat.clearTitle':'Clear the chat log',
   'chat.warn':'⚠ Launches a real agent (<code>config.json → runners</code>) that can modify files &amp; run commands.',
   'info.title':'Info','info.sub':'Project overview — configuration, counts, specs and active workflow.',
   'info.project':'Project','info.projectType':'Project type','info.mode':'Mode','info.language':'Language',
@@ -107,7 +109,7 @@ en: {
   'task.failingPassing':'{f} failing, {p} passing','task.comments':'Comments','task.noComments':'No comments.',
   'task.addCommentPlaceholder':'Add a comment, a remark, feedback…',
   'task.toAnalyzeHint':'"To analyze" moves the task back so the agent picks it up next round.',
-  'customize.title':'Customize','customize.sub':'Add project-specific dashboards, skills and agents — describe what you want, or let Auto propose candidates from this project.',
+  'customize.title':'Extend spectoflow','customize.sub':'Add project-specific dashboards, skills and agents — describe what you want, or let Auto propose candidates from this project.',
   'customize.dashboards':'Dashboards','customize.skills':'Skills','customize.agents':'Agents',
   'customize.add.dashboard':'Add dashboard','customize.add.skill':'Add skill','customize.add.agent':'Add agent',
   'customize.empty.dashboard':'No custom dashboards yet.','customize.empty.skill':'No custom skills yet.','customize.empty.agent':'No custom agents yet.',
@@ -115,18 +117,20 @@ en: {
 },
 fr: {
   'nav.board':'Tableau','nav.requests':'Demandes','nav.attention':'Attention','nav.backlog':'Backlog',
-  'nav.workflow':'Workflow','nav.team':'Agents et compétences','nav.chat':'Chat','nav.info':'Infos','nav.settings':'Réglages',
+  'nav.workflow':'Workflow','nav.team':'Agents et compétences','nav.chat':'Chat','nav.info':'Infos','nav.settings':'Personnalisation',
   'status.todo':'À faire','status.in_progress':'En cours','status.to_validate':'À valider',
   'status.to_analyze':'À analyser','status.done':'Terminé','status.blocked':'Bloqué',
   'filter.all':'Tous','filter.open':'Ouverts','filter.resolved':'Résolus',
-  'action.run':'Lancer','action.send':'Envoyer','action.orchestrate':'Orchestrer','action.approve':'Approuver',
+  'action.run':'Lancer','action.send':'Envoyer','action.orchestrate':'Orchestrer','action.approve':'Approuver','action.summarize':'Résumer','action.clear':'Effacer',
   'action.cancel':'Annuler','action.add':'Ajouter','action.edit':'Modifier','action.delete':'Supprimer',
   'action.resolve':'Résoudre','action.reopen':'Rouvrir','action.validateToTask':'Valider → tâche',
   'action.addToAnalyze':'Ajouter + à analyser',
   'topbar.navToggle':'Afficher le menu','topbar.run.title':'Ouvrir le chat d’exécution',
   'topbar.theme.toggle':'Basculer thème clair / sombre','topbar.theme.toggle.title':'Basculer clair / sombre',
-  'topbar.mode.title':'Mode d’autonomie — modifiable ici ou dans Réglages',
-  'topbar.lang.title':'Langue de sortie — modifiable ici ou dans Réglages',
+  'topbar.mode.title':'Mode d’autonomie — modifiable ici ou dans Personnalisation',
+  'topbar.lang.title':'Langue de sortie — modifiable ici ou dans Personnalisation',
+  'topbar.agent.title':'Agent actif — modifiable ici ou dans Personnalisation',
+  'topbar.agent.none':'Aucun agent trouvé',
   'board.filterPlaceholder':'Filtrer les tâches…','board.expandAll':'Tout déplier','board.collapseAll':'Tout replier',
   'board.expandAllTitle':'Déplier ou replier toutes les phases','board.viewList':'Liste','board.viewKanban':'Kanban',
   'board.viewList.title':'Regroupé par phase','board.viewKanban.title':'Colonnes par statut',
@@ -183,7 +187,7 @@ fr: {
   'chat.tabSub':'Conversation complète avec l’exécuteur — la même exécution que le widget, avec plus de place pour la lire.',
   'chat.idle':'Tapez une demande — l’agent s’exécute sans supervision dans ce projet avec toute sa mémoire (<code>CLAUDE.md → AGENTS.md</code>) et met le tableau à jour en direct.',
   'chat.inputPlaceholder':'ex. Ajouter une fonctionnalité de connexion par email + mot de passe',
-  'chat.orchestrateTitle':'Parcourir le workflow activé',
+  'chat.orchestrateTitle':'Parcourir le workflow activé','chat.summarizeTitle':'Condenser l’activité récente en un résumé','chat.clearTitle':'Effacer le journal de discussion',
   'chat.warn':'⚠ Lance un agent réel (<code>config.json → runners</code>) qui peut modifier des fichiers et exécuter des commandes.',
   'info.title':'Infos','info.sub':'Vue d’ensemble du projet — configuration, comptages, specs et workflow actif.',
   'info.project':'Projet','info.projectType':'Type de projet','info.mode':'Mode','info.language':'Langue',
@@ -202,7 +206,7 @@ fr: {
   'task.failingPassing':'{f} en échec, {p} réussi(s)','task.comments':'Commentaires','task.noComments':'Aucun commentaire.',
   'task.addCommentPlaceholder':'Ajouter un commentaire, une remarque, un retour…',
   'task.toAnalyzeHint':'« À analyser » renvoie la tâche pour que l’agent la reprenne au tour suivant.',
-  'customize.title':'Personnaliser','customize.sub':'Ajoutez des dashboards, compétences et agents propres au projet — décrivez ce que vous voulez, ou laissez Auto proposer des pistes à partir de ce projet.',
+  'customize.title':'Étendre spectoflow','customize.sub':'Ajoutez des dashboards, compétences et agents propres au projet — décrivez ce que vous voulez, ou laissez Auto proposer des pistes à partir de ce projet.',
   'customize.dashboards':'Dashboards','customize.skills':'Compétences','customize.agents':'Agents',
   'customize.add.dashboard':'Ajouter un dashboard','customize.add.skill':'Ajouter une compétence','customize.add.agent':'Ajouter un agent',
   'customize.empty.dashboard':'Aucun dashboard personnalisé pour l’instant.','customize.empty.skill':'Aucune compétence personnalisée pour l’instant.','customize.empty.agent':'Aucun agent personnalisé pour l’instant.',
@@ -210,18 +214,20 @@ fr: {
 },
 es: {
   'nav.board':'Tablero','nav.requests':'Solicitudes','nav.attention':'Atención','nav.backlog':'Backlog',
-  'nav.workflow':'Workflow','nav.team':'Agentes y habilidades','nav.chat':'Chat','nav.info':'Info','nav.settings':'Ajustes',
+  'nav.workflow':'Workflow','nav.team':'Agentes y habilidades','nav.chat':'Chat','nav.info':'Info','nav.settings':'Personalizar',
   'status.todo':'Por hacer','status.in_progress':'En curso','status.to_validate':'Por validar',
   'status.to_analyze':'Por analizar','status.done':'Hecho','status.blocked':'Bloqueado',
   'filter.all':'Todos','filter.open':'Abiertos','filter.resolved':'Resueltos',
-  'action.run':'Ejecutar','action.send':'Enviar','action.orchestrate':'Orquestar','action.approve':'Aprobar',
+  'action.run':'Ejecutar','action.send':'Enviar','action.orchestrate':'Orquestar','action.approve':'Aprobar','action.summarize':'Resumir','action.clear':'Borrar',
   'action.cancel':'Cancelar','action.add':'Añadir','action.edit':'Editar','action.delete':'Eliminar',
   'action.resolve':'Resolver','action.reopen':'Reabrir','action.validateToTask':'Validar → tarea',
   'action.addToAnalyze':'Añadir + por analizar',
   'topbar.navToggle':'Mostrar menú','topbar.run.title':'Abrir el chat de ejecución',
   'topbar.theme.toggle':'Cambiar tema claro / oscuro','topbar.theme.toggle.title':'Cambiar claro / oscuro',
-  'topbar.mode.title':'Modo de autonomía — se cambia aquí o en Ajustes',
-  'topbar.lang.title':'Idioma de salida — se cambia aquí o en Ajustes',
+  'topbar.mode.title':'Modo de autonomía — se cambia aquí o en Personalizar',
+  'topbar.lang.title':'Idioma de salida — se cambia aquí o en Personalizar',
+  'topbar.agent.title':'Agente activo — se cambia aquí o en Personalizar',
+  'topbar.agent.none':'No se encontró ningún agente',
   'board.filterPlaceholder':'Filtrar tareas…','board.expandAll':'Expandir todo','board.collapseAll':'Colapsar todo',
   'board.expandAllTitle':'Expandir o colapsar todas las fases','board.viewList':'Lista','board.viewKanban':'Kanban',
   'board.viewList.title':'Agrupado por fase','board.viewKanban.title':'Columnas por estado',
@@ -278,7 +284,7 @@ es: {
   'chat.tabSub':'Conversación completa con el ejecutor — la misma ejecución que el widget, con más espacio para leerla.',
   'chat.idle':'Escribe una solicitud — el agente se ejecuta sin supervisión en este proyecto con toda su memoria (<code>CLAUDE.md → AGENTS.md</code>) y actualiza el tablero en vivo.',
   'chat.inputPlaceholder':'p. ej. Añadir un inicio de sesión con email + contraseña',
-  'chat.orchestrateTitle':'Recorrer el workflow activado',
+  'chat.orchestrateTitle':'Recorrer el workflow activado','chat.summarizeTitle':'Condensar la actividad reciente en un resumen','chat.clearTitle':'Borrar el registro del chat',
   'chat.warn':'⚠ Lanza un agente real (<code>config.json → runners</code>) que puede modificar archivos y ejecutar comandos.',
   'info.title':'Info','info.sub':'Visión general del proyecto — configuración, recuentos, specs y workflow activo.',
   'info.project':'Proyecto','info.projectType':'Tipo de proyecto','info.mode':'Modo','info.language':'Idioma',
@@ -297,7 +303,7 @@ es: {
   'task.failingPassing':'{f} fallando, {p} superadas','task.comments':'Comentarios','task.noComments':'Sin comentarios.',
   'task.addCommentPlaceholder':'Añade un comentario, una observación, feedback…',
   'task.toAnalyzeHint':'«Por analizar» devuelve la tarea para que el agente la retome en la siguiente ronda.',
-  'customize.title':'Personalizar','customize.sub':'Añade dashboards, habilidades y agentes propios del proyecto — describe lo que quieres, o deja que Auto proponga candidatos a partir de este proyecto.',
+  'customize.title':'Ampliar spectoflow','customize.sub':'Añade dashboards, habilidades y agentes propios del proyecto — describe lo que quieres, o deja que Auto proponga candidatos a partir de este proyecto.',
   'customize.dashboards':'Dashboards','customize.skills':'Habilidades','customize.agents':'Agentes',
   'customize.add.dashboard':'Añadir dashboard','customize.add.skill':'Añadir habilidad','customize.add.agent':'Añadir agente',
   'customize.empty.dashboard':'Aún no hay dashboards personalizados.','customize.empty.skill':'Aún no hay habilidades personalizadas.','customize.empty.agent':'Aún no hay agentes personalizados.',
@@ -305,18 +311,20 @@ es: {
 },
 de: {
   'nav.board':'Board','nav.requests':'Anfragen','nav.attention':'Hinweise','nav.backlog':'Backlog',
-  'nav.workflow':'Workflow','nav.team':'Agenten & Skills','nav.chat':'Chat','nav.info':'Info','nav.settings':'Einstellungen',
+  'nav.workflow':'Workflow','nav.team':'Agenten & Skills','nav.chat':'Chat','nav.info':'Info','nav.settings':'Personalisieren',
   'status.todo':'Offen','status.in_progress':'In Arbeit','status.to_validate':'Zu prüfen',
   'status.to_analyze':'Zu analysieren','status.done':'Erledigt','status.blocked':'Blockiert',
   'filter.all':'Alle','filter.open':'Offen','filter.resolved':'Erledigt',
-  'action.run':'Start','action.send':'Senden','action.orchestrate':'Orchestrieren','action.approve':'Freigeben',
+  'action.run':'Start','action.send':'Senden','action.orchestrate':'Orchestrieren','action.approve':'Freigeben','action.summarize':'Zusammenfassen','action.clear':'Löschen',
   'action.cancel':'Abbrechen','action.add':'Hinzufügen','action.edit':'Bearbeiten','action.delete':'Löschen',
   'action.resolve':'Erledigen','action.reopen':'Wieder öffnen','action.validateToTask':'Bestätigen → Aufgabe',
   'action.addToAnalyze':'Hinzufügen + zu analysieren',
   'topbar.navToggle':'Menü anzeigen','topbar.run.title':'Ausführungs-Chat öffnen',
   'topbar.theme.toggle':'Hell-/Dunkelmodus umschalten','topbar.theme.toggle.title':'Hell / Dunkel umschalten',
-  'topbar.mode.title':'Autonomiemodus — hier oder in den Einstellungen änderbar',
-  'topbar.lang.title':'Ausgabesprache — hier oder in den Einstellungen änderbar',
+  'topbar.mode.title':'Autonomiemodus — hier oder in Personalisieren änderbar',
+  'topbar.lang.title':'Ausgabesprache — hier oder in Personalisieren änderbar',
+  'topbar.agent.title':'Aktiver Agent — hier oder in Personalisieren änderbar',
+  'topbar.agent.none':'Kein Agent gefunden',
   'board.filterPlaceholder':'Aufgaben filtern…','board.expandAll':'Alle ausklappen','board.collapseAll':'Alle einklappen',
   'board.expandAllTitle':'Alle Phasen ein- oder ausklappen','board.viewList':'Liste','board.viewKanban':'Kanban',
   'board.viewList.title':'Nach Phase gruppiert','board.viewKanban.title':'Spalten nach Status',
@@ -373,7 +381,7 @@ de: {
   'chat.tabSub':'Vollständiges Gespräch mit dem Runner — derselbe Lauf wie im Widget, mit mehr Platz zum Lesen.',
   'chat.idle':'Geben Sie eine Anfrage ein — der Agent läuft eigenständig in diesem Projekt mit vollem Gedächtnis (<code>CLAUDE.md → AGENTS.md</code>) und aktualisiert das Board live.',
   'chat.inputPlaceholder':'z. B. Login mit E-Mail + Passwort hinzufügen',
-  'chat.orchestrateTitle':'Den aktivierten Workflow durchlaufen',
+  'chat.orchestrateTitle':'Den aktivierten Workflow durchlaufen','chat.summarizeTitle':'Die letzten Aktivitäten zu einer Zusammenfassung verdichten','chat.clearTitle':'Chat-Verlauf löschen',
   'chat.warn':'⚠ Startet einen echten Agenten (<code>config.json → runners</code>), der Dateien ändern und Befehle ausführen kann.',
   'info.title':'Info','info.sub':'Projektübersicht — Konfiguration, Zahlen, Specs und aktiver Workflow.',
   'info.project':'Projekt','info.projectType':'Projekttyp','info.mode':'Modus','info.language':'Sprache',
@@ -392,7 +400,7 @@ de: {
   'task.failingPassing':'{f} fehlgeschlagen, {p} bestanden','task.comments':'Kommentare','task.noComments':'Keine Kommentare.',
   'task.addCommentPlaceholder':'Kommentar, Anmerkung oder Feedback hinzufügen…',
   'task.toAnalyzeHint':'„Zu analysieren“ gibt die Aufgabe zurück, damit der Agent sie in der nächsten Runde aufgreift.',
-  'customize.title':'Anpassen','customize.sub':'Fügen Sie projektspezifische Dashboards, Skills und Agenten hinzu — beschreiben Sie, was Sie wollen, oder lassen Sie Auto Kandidaten aus diesem Projekt vorschlagen.',
+  'customize.title':'spectoflow erweitern','customize.sub':'Fügen Sie projektspezifische Dashboards, Skills und Agenten hinzu — beschreiben Sie, was Sie wollen, oder lassen Sie Auto Kandidaten aus diesem Projekt vorschlagen.',
   'customize.dashboards':'Dashboards','customize.skills':'Skills','customize.agents':'Agenten',
   'customize.add.dashboard':'Dashboard hinzufügen','customize.add.skill':'Skill hinzufügen','customize.add.agent':'Agent hinzufügen',
   'customize.empty.dashboard':'Noch keine eigenen Dashboards.','customize.empty.skill':'Noch keine eigenen Skills.','customize.empty.agent':'Noch keine eigenen Agenten.',
@@ -400,18 +408,20 @@ de: {
 },
 pt: {
   'nav.board':'Painel','nav.requests':'Pedidos','nav.attention':'Atenção','nav.backlog':'Backlog',
-  'nav.workflow':'Workflow','nav.team':'Agentes e habilidades','nav.chat':'Chat','nav.info':'Info','nav.settings':'Definições',
+  'nav.workflow':'Workflow','nav.team':'Agentes e habilidades','nav.chat':'Chat','nav.info':'Info','nav.settings':'Personalizar',
   'status.todo':'A fazer','status.in_progress':'Em curso','status.to_validate':'Por validar',
   'status.to_analyze':'Por analisar','status.done':'Concluído','status.blocked':'Bloqueado',
   'filter.all':'Todos','filter.open':'Abertos','filter.resolved':'Resolvidos',
-  'action.run':'Executar','action.send':'Enviar','action.orchestrate':'Orquestrar','action.approve':'Aprovar',
+  'action.run':'Executar','action.send':'Enviar','action.orchestrate':'Orquestrar','action.approve':'Aprovar','action.summarize':'Resumir','action.clear':'Limpar',
   'action.cancel':'Cancelar','action.add':'Adicionar','action.edit':'Editar','action.delete':'Eliminar',
   'action.resolve':'Resolver','action.reopen':'Reabrir','action.validateToTask':'Validar → tarefa',
   'action.addToAnalyze':'Adicionar + por analisar',
   'topbar.navToggle':'Mostrar menu','topbar.run.title':'Abrir o chat de execução',
   'topbar.theme.toggle':'Alternar tema claro / escuro','topbar.theme.toggle.title':'Alternar claro / escuro',
-  'topbar.mode.title':'Modo de autonomia — altere aqui ou em Definições',
-  'topbar.lang.title':'Idioma de saída — altere aqui ou em Definições',
+  'topbar.mode.title':'Modo de autonomia — altere aqui ou em Personalizar',
+  'topbar.lang.title':'Idioma de saída — altere aqui ou em Personalizar',
+  'topbar.agent.title':'Agente ativo — altere aqui ou em Personalizar',
+  'topbar.agent.none':'Nenhum agente encontrado',
   'board.filterPlaceholder':'Filtrar tarefas…','board.expandAll':'Expandir tudo','board.collapseAll':'Recolher tudo',
   'board.expandAllTitle':'Expandir ou recolher todas as fases','board.viewList':'Lista','board.viewKanban':'Kanban',
   'board.viewList.title':'Agrupado por fase','board.viewKanban.title':'Colunas por estado',
@@ -468,7 +478,7 @@ pt: {
   'chat.tabSub':'Conversa completa com o executor — a mesma execução do widget, com mais espaço para ler.',
   'chat.idle':'Escreva um pedido — o agente corre sem supervisão neste projeto com toda a sua memória (<code>CLAUDE.md → AGENTS.md</code>) e atualiza o painel em direto.',
   'chat.inputPlaceholder':'ex. Adicionar login com email + palavra-passe',
-  'chat.orchestrateTitle':'Percorrer o workflow ativado',
+  'chat.orchestrateTitle':'Percorrer o workflow ativado','chat.summarizeTitle':'Condensar a atividade recente num resumo','chat.clearTitle':'Limpar o registo do chat',
   'chat.warn':'⚠ Inicia um agente real (<code>config.json → runners</code>) que pode alterar ficheiros e executar comandos.',
   'info.title':'Info','info.sub':'Visão geral do projeto — configuração, contagens, specs e workflow ativo.',
   'info.project':'Projeto','info.projectType':'Tipo de projeto','info.mode':'Modo','info.language':'Idioma',
@@ -487,7 +497,7 @@ pt: {
   'task.failingPassing':'{f} a falhar, {p} bem-sucedido(s)','task.comments':'Comentários','task.noComments':'Sem comentários.',
   'task.addCommentPlaceholder':'Adicione um comentário, uma observação, feedback…',
   'task.toAnalyzeHint':'«Por analisar» devolve a tarefa para o agente a retomar na ronda seguinte.',
-  'customize.title':'Personalizar','customize.sub':'Adicione dashboards, habilidades e agentes específicos do projeto — descreva o que quer, ou deixe o Auto propor candidatos a partir deste projeto.',
+  'customize.title':'Estender o spectoflow','customize.sub':'Adicione dashboards, habilidades e agentes específicos do projeto — descreva o que quer, ou deixe o Auto propor candidatos a partir deste projeto.',
   'customize.dashboards':'Dashboards','customize.skills':'Habilidades','customize.agents':'Agentes',
   'customize.add.dashboard':'Adicionar dashboard','customize.add.skill':'Adicionar habilidade','customize.add.agent':'Adicionar agente',
   'customize.empty.dashboard':'Ainda sem dashboards personalizados.','customize.empty.skill':'Ainda sem habilidades personalizadas.','customize.empty.agent':'Ainda sem agentes personalizados.',
@@ -495,18 +505,20 @@ pt: {
 },
 it: {
   'nav.board':'Bacheca','nav.requests':'Richieste','nav.attention':'Attenzione','nav.backlog':'Backlog',
-  'nav.workflow':'Workflow','nav.team':'Agenti e competenze','nav.chat':'Chat','nav.info':'Info','nav.settings':'Impostazioni',
+  'nav.workflow':'Workflow','nav.team':'Agenti e competenze','nav.chat':'Chat','nav.info':'Info','nav.settings':'Personalizza',
   'status.todo':'Da fare','status.in_progress':'In corso','status.to_validate':'Da convalidare',
   'status.to_analyze':'Da analizzare','status.done':'Fatto','status.blocked':'Bloccato',
   'filter.all':'Tutti','filter.open':'Aperti','filter.resolved':'Risolti',
-  'action.run':'Avvia','action.send':'Invia','action.orchestrate':'Orchestra','action.approve':'Approva',
+  'action.run':'Avvia','action.send':'Invia','action.orchestrate':'Orchestra','action.approve':'Approva','action.summarize':'Riassumi','action.clear':'Cancella',
   'action.cancel':'Annulla','action.add':'Aggiungi','action.edit':'Modifica','action.delete':'Elimina',
   'action.resolve':'Risolvi','action.reopen':'Riapri','action.validateToTask':'Convalida → attività',
   'action.addToAnalyze':'Aggiungi + da analizzare',
   'topbar.navToggle':'Mostra menu','topbar.run.title':'Apri la chat di esecuzione',
   'topbar.theme.toggle':'Passa a tema chiaro / scuro','topbar.theme.toggle.title':'Chiaro / scuro',
-  'topbar.mode.title':'Modalità di autonomia — modificabile qui o nelle Impostazioni',
-  'topbar.lang.title':'Lingua di output — modificabile qui o nelle Impostazioni',
+  'topbar.mode.title':'Modalità di autonomia — modificabile qui o in Personalizza',
+  'topbar.lang.title':'Lingua di output — modificabile qui o in Personalizza',
+  'topbar.agent.title':'Agente attivo — modificabile qui o in Personalizza',
+  'topbar.agent.none':'Nessun agente trovato',
   'board.filterPlaceholder':'Filtra attività…','board.expandAll':'Espandi tutto','board.collapseAll':'Comprimi tutto',
   'board.expandAllTitle':'Espandi o comprimi tutte le fasi','board.viewList':'Elenco','board.viewKanban':'Kanban',
   'board.viewList.title':'Raggruppato per fase','board.viewKanban.title':'Colonne per stato',
@@ -563,7 +575,7 @@ it: {
   'chat.tabSub':'Conversazione completa con l’esecutore — la stessa esecuzione del widget, con più spazio per leggerla.',
   'chat.idle':'Digita una richiesta — l’agente viene eseguito senza supervisione in questo progetto con tutta la sua memoria (<code>CLAUDE.md → AGENTS.md</code>) e aggiorna la bacheca in diretta.',
   'chat.inputPlaceholder':'es. Aggiungi un login con email + password',
-  'chat.orchestrateTitle':'Percorri il workflow attivato',
+  'chat.orchestrateTitle':'Percorri il workflow attivato','chat.summarizeTitle':'Condensa l’attività recente in un riassunto','chat.clearTitle':'Cancella il registro della chat',
   'chat.warn':'⚠ Avvia un agente reale (<code>config.json → runners</code>) che può modificare file ed eseguire comandi.',
   'info.title':'Info','info.sub':'Panoramica del progetto — configurazione, conteggi, specs e workflow attivo.',
   'info.project':'Progetto','info.projectType':'Tipo di progetto','info.mode':'Modalità','info.language':'Lingua',
@@ -582,7 +594,7 @@ it: {
   'task.failingPassing':'{f} falliti, {p} superati','task.comments':'Commenti','task.noComments':'Nessun commento.',
   'task.addCommentPlaceholder':'Aggiungi un commento, un’osservazione, un feedback…',
   'task.toAnalyzeHint':'«Da analizzare» rimanda l’attività così l’agente la riprende al giro successivo.',
-  'customize.title':'Personalizza','customize.sub':'Aggiungi dashboard, skill e agenti specifici del progetto — descrivi cosa vuoi, oppure lascia che Auto proponga candidati a partire da questo progetto.',
+  'customize.title':'Estendi spectoflow','customize.sub':'Aggiungi dashboard, skill e agenti specifici del progetto — descrivi cosa vuoi, oppure lascia che Auto proponga candidati a partire da questo progetto.',
   'customize.dashboards':'Dashboard','customize.skills':'Skill','customize.agents':'Agenti',
   'customize.add.dashboard':'Aggiungi dashboard','customize.add.skill':'Aggiungi skill','customize.add.agent':'Aggiungi agente',
   'customize.empty.dashboard':'Ancora nessuna dashboard personalizzata.','customize.empty.skill':'Ancora nessuna skill personalizzata.','customize.empty.agent':'Ancora nessun agente personalizzato.',

package/templates/dashboard/public/index.html

@@ -26,13 +26,15 @@
           <span class="brand-project" id="projectName">—</span>
         </div>
         <div class="brand-sub" id="brandSub">
-          <select id="topMode" class="brand-mini-select" data-i18n-title="topbar.mode.title" data-i18n-aria="field.mode" aria-label="Autonomy mode" title="Autonomy mode — change from here or in Settings">
+          <select id="topAgent" class="brand-mini-select brand-agent-select" data-i18n-title="topbar.agent.title" data-i18n-aria="info.activeAgent" aria-label="Active agent" title="Active agent — change from here or in Personalize"></select>
+          <span class="brand-sub-sep">·</span>
+          <select id="topMode" class="brand-mini-select" data-i18n-title="topbar.mode.title" data-i18n-aria="field.mode" aria-label="Autonomy mode" title="Autonomy mode — change from here or in Personalize">
             <option value="autopilot">autopilot</option>
             <option value="semi">semi</option>
             <option value="manual">manual</option>
           </select>
           <span class="brand-sub-sep">·</span>
-          <select id="topLang" class="brand-mini-select" data-i18n-title="topbar.lang.title" data-i18n-aria="field.lang" aria-label="Output language" title="Output language — change from here or in Settings">
+          <select id="topLang" class="brand-mini-select" data-i18n-title="topbar.lang.title" data-i18n-aria="field.lang" aria-label="Output language" title="Output language — change from here or in Personalize">
             <option value="en">en</option>
             <option value="fr">fr</option>
             <option value="es">es</option>
@@ -47,15 +49,15 @@
       </div>
     </div>
     <nav class="tabs" id="tabs">
-      <button class="tab is-active" data-tab="board"><span class="tab-ico" data-icon="board"></span><span class="tab-label" data-i18n="nav.board">Board</span></button>
-      <button class="tab" data-tab="chat"><span class="tab-ico" data-icon="chat"></span><span class="tab-label" data-i18n="nav.chat">Chat</span></button>
-      <button class="tab" data-tab="requests"><span class="tab-ico" data-icon="requests"></span><span class="tab-label" data-i18n="nav.requests">Requests</span></button>
-      <button class="tab" data-tab="attention"><span class="tab-ico" data-icon="attention"></span><span class="tab-label" data-i18n="nav.attention">Attention</span><span class="tab-badge" id="attnBadge" hidden>0</span></button>
-      <button class="tab" data-tab="backlog"><span class="tab-ico" data-icon="backlog"></span><span class="tab-label" data-i18n="nav.backlog">Backlog</span></button>
-      <button class="tab" data-tab="workflow"><span class="tab-ico" data-icon="workflow"></span><span class="tab-label" data-i18n="nav.workflow">Workflow</span></button>
-      <button class="tab" data-tab="team"><span class="tab-ico" data-icon="agents"></span><span class="tab-label" data-i18n="nav.team">Agents &amp; Skills</span></button>
-      <button class="tab" data-tab="info"><span class="tab-ico" data-icon="info"></span><span class="tab-label" data-i18n="nav.info">Info</span></button>
-      <button class="tab" data-tab="settings"><span class="tab-ico" data-icon="settings"></span><span class="tab-label" data-i18n="nav.settings">Settings</span></button>
+      <button class="tab is-active" data-tab="board" data-i18n-title="nav.board" title="Board"><span class="tab-ico" data-icon="board"></span><span class="tab-label" data-i18n="nav.board">Board</span></button>
+      <button class="tab" data-tab="chat" data-i18n-title="nav.chat" title="Chat"><span class="tab-ico" data-icon="chat"></span><span class="tab-label" data-i18n="nav.chat">Chat</span></button>
+      <button class="tab" data-tab="requests" data-i18n-title="nav.requests" title="Requests"><span class="tab-ico" data-icon="requests"></span><span class="tab-label" data-i18n="nav.requests">Requests</span></button>
+      <button class="tab" data-tab="attention" data-i18n-title="nav.attention" title="Attention"><span class="tab-ico" data-icon="attention"></span><span class="tab-label" data-i18n="nav.attention">Attention</span><span class="tab-badge" id="attnBadge" hidden>0</span></button>
+      <button class="tab" data-tab="backlog" data-i18n-title="nav.backlog" title="Backlog"><span class="tab-ico" data-icon="backlog"></span><span class="tab-label" data-i18n="nav.backlog">Backlog</span></button>
+      <button class="tab" data-tab="workflow" data-i18n-title="nav.workflow" title="Workflow"><span class="tab-ico" data-icon="workflow"></span><span class="tab-label" data-i18n="nav.workflow">Workflow</span></button>
+      <button class="tab" data-tab="team" data-i18n-title="nav.team" title="Agents &amp; Skills"><span class="tab-ico" data-icon="agents"></span><span class="tab-label" data-i18n="nav.team">Agents &amp; Skills</span></button>
+      <button class="tab" data-tab="info" data-i18n-title="nav.info" title="Info"><span class="tab-ico" data-icon="info"></span><span class="tab-label" data-i18n="nav.info">Info</span></button>
+      <button class="tab" data-tab="settings" data-i18n-title="nav.settings" title="Personalize"><span class="tab-ico" data-icon="settings"></span><span class="tab-label" data-i18n="nav.settings">Personalize</span></button>
     </nav>
     <div class="top-right">
       <span class="sync" id="sync"><span class="sync-dot"></span><span id="syncLabel">live</span></span>
@@ -199,7 +201,11 @@
             <h2 class="panel-title" data-i18n="nav.chat">Chat</h2>
             <p class="panel-sub" data-i18n="chat.tabSub">Full conversation with the runner — the same run as the widget, more room to read it.</p>
           </div>
-          <select id="tabRunAgent" class="chat-agent" title="Agent"></select>
+          <div class="chat-tab-head-actions">
+            <button id="tabSummarizeBtn" class="mini-btn" data-i18n-title="chat.summarizeTitle" data-i18n="action.summarize" title="Condense the recent activity into a summary">Summarize</button>
+            <button id="tabClearBtn" class="mini-btn" data-i18n-title="chat.clearTitle" data-i18n="action.clear" title="Clear the chat log">Clear</button>
+            <select id="tabRunAgent" class="chat-agent" title="Agent"></select>
+          </div>
         </div>
         <div class="chat-tab-log" id="chatTabLog">
           <div class="chat-idle" data-i18n-html="chat.idle">Type a request — the agent runs headless in this project with full memory
@@ -228,9 +234,14 @@
     <!-- SETTINGS (change autonomy mode + output language → config.json) -->
     <section class="panel" data-panel="settings">
       <div class="settings-wrap">
-        <h2 class="panel-title" data-i18n="nav.settings">Settings</h2>
+        <h2 class="panel-title" data-i18n="nav.settings">Personalize</h2>
         <p class="panel-sub" data-i18n-html="settings.sub">Change how spectoflow runs. Writes <code>.spectoflow/config.json</code> — the agent picks it up on its next run.</p>
         <div class="settings-card">
+          <label class="settings-field">
+            <span class="settings-field-label" data-i18n="info.activeAgent">Active agent</span>
+            <select id="setAgent" class="settings-select"></select>
+            <span class="settings-field-hint is-empty" id="setAgentHint" data-i18n="topbar.agent.none" hidden>No agent found</span>
+          </label>
           <label class="settings-field">
             <span class="settings-field-label" data-i18n="field.design">Dashboard design</span>
             <select id="setDesign" class="settings-select"></select>
@@ -261,7 +272,7 @@
         <!-- Customize: project-specific dashboards, skills and agents — described or auto-proposed,
              generated by the configured agent through the same Run/Chat pipeline as any other ask. -->
         <div class="cz-wrap">
-          <h2 class="panel-title" data-i18n="customize.title">Customize</h2>
+          <h2 class="panel-title" data-i18n="customize.title">Extend spectoflow</h2>
           <p class="panel-sub" data-i18n="customize.sub">Add project-specific dashboards, skills and agents — describe what you want, or let Auto propose candidates from this project.</p>
           <div id="czRoot"></div>
         </div>
@@ -283,7 +294,7 @@
     <div class="footer-right">
       <a href="https://github.com/georgesmomo/spectoflow" target="_blank" rel="noopener">GitHub</a>
       <a href="https://www.npmjs.com/package/spectoflow" target="_blank" rel="noopener">npm</a>
-      <button class="footer-link" id="footerSettings" data-i18n="nav.settings">Settings</button>
+      <button class="footer-link" id="footerSettings" data-i18n="nav.settings">Personalize</button>
     </div>
   </footer>
 

package/templates/dashboard/public/styles.css

@@ -40,6 +40,9 @@ body { background:var(--bg); color:var(--ink); font-family:var(--sans); font-siz
 }
 .brand-mini-select:hover,.brand-mini-select:focus-visible { color:var(--ink); border-color:var(--line); background-color:var(--surface-2); outline:none; }
 .brand-mini-select option { color:var(--ink); background:var(--surface); text-transform:none; }
+.brand-agent-select { font-weight:700; }
+.brand-agent-select.is-empty { color:var(--s-blocked); pointer-events:none; }
+.brand-agent-select.is-empty:hover { color:var(--s-blocked); border-color:transparent; background-color:transparent; }
 .progress-meter { width:76px; height:5px; border-radius:999px; background:var(--surface-2); overflow:hidden; flex-shrink:0; margin-left:4px; }
 .progress-meter-fill { height:100%; width:0%; background:linear-gradient(90deg,var(--cool),var(--signal)); border-radius:999px; transition:width .5s ease; }
 .tabs { justify-self:center; display:flex; gap:2px; min-width:0; max-width:100%; overflow-x:auto; overflow-y:hidden; scrollbar-width:none; }
@@ -411,7 +414,8 @@ body { background:var(--bg); color:var(--ink); font-family:var(--sans); font-siz
 .chat-tab-head { display:flex; align-items:flex-start; gap:14px; margin-bottom:14px; }
 .chat-tab-head .panel-title { margin:0; }
 .chat-tab-head .panel-sub { margin:2px 0 0; }
-.chat-tab-head .chat-agent { margin-left:auto; flex-shrink:0; }
+.chat-tab-head-actions { display:flex; align-items:center; gap:8px; margin-left:auto; flex-shrink:0; flex-wrap:wrap; }
+.chat-tab-head .chat-agent { flex-shrink:0; }
 .chat-tab-log { flex:1; min-height:220px; overflow-y:auto; display:flex; flex-direction:column; gap:12px; background:var(--surface); border:1px solid var(--line); border-radius:var(--radius); padding:20px; box-shadow:var(--shadow); }
 .chat-tab-input { display:flex; gap:10px; margin-top:14px; }
 .chat-tab-input .chat-ta { flex:1; min-height:56px; max-height:180px; }
@@ -549,6 +553,8 @@ body.booting .ring-svg circle:last-of-type { transform-origin:center; animation:
 .settings-field-label { font-size:11px; text-transform:uppercase; letter-spacing:.05em; color:var(--muted); }
 .settings-select { font-family:inherit; font-size:14px; padding:9px 11px; border:1px solid var(--line); border-radius:9px; background:var(--surface-2); color:var(--ink); cursor:pointer; }
 .settings-select:focus { outline:2px solid var(--signal); outline-offset:1px; }
+.settings-field-hint { font-size:12px; }
+.settings-field-hint.is-empty { color:var(--s-blocked); }
 .settings-saved { align-self:flex-start; font-size:12px; color:var(--s-done); font-family:var(--mono); }
 .settings-saved[hidden] { display:none; }
 .settings-readonly { background:var(--surface); border:1px solid var(--line); border-radius:var(--radius); overflow:hidden; }

package/templates/dashboard/server.js

@@ -10,7 +10,9 @@ const fs = require('fs');
 const path = require('path');
 const store = require('../lib/store');
 const { startRun } = require('./runner');
+const { runSummarize } = require('./summarize');
 const orchestrator = require('./orchestrator');
+const agentsRegistry = require('../lib/agents-registry');
 
 const PORT = process.env.SPECTOFLOW_PORT ? Number(process.env.SPECTOFLOW_PORT) : 4319;
 const PUBLIC = path.join(__dirname, 'public');
@@ -30,6 +32,11 @@ function project(){
   const p = store.readProject(ROOT);
   const v = frameworkVersion(); if (v) p.version = v;
   p.projectName = path.basename(ROOT); // the actual project folder, always shown in the topbar
+  // Known vs. actually-installed agents — the topbar switcher needs both: the full list to offer,
+  // and which ones are real (bin on PATH, or the project already has that agent's config dir) so it
+  // can refuse to activate one that isn't there.
+  p.knownAgents = agentsRegistry.KNOWN_AGENTS.map((a) => ({ id: a.id, label: a.label }));
+  p.installedAgents = agentsRegistry.installedAgents(ROOT);
   return p;
 }
 function sendJSON(res,code,obj){ res.writeHead(code,{'Content-Type':'application/json; charset=utf-8'}); res.end(JSON.stringify(obj)); }
@@ -44,6 +51,20 @@ function writeConfig(patch){
   if (patch.mode && ['autopilot','semi','manual'].includes(patch.mode)) cfg.mode = patch.mode;
   if (typeof patch.language === 'string' && patch.language.trim()) cfg.language = patch.language.trim();
   if (typeof patch.design === 'string' && /^[a-z0-9-]{1,40}$/.test(patch.design)) cfg.design = patch.design;
+  if (typeof patch.agent === 'string' && patch.agent.trim()) {
+    const id = patch.agent.trim();
+    // Never activate an agent whose CLI isn't actually there — a picked-but-absent agent would just
+    // fail silently the next time something tries to run it.
+    if (!agentsRegistry.isAgentInstalled(id, ROOT)) {
+      const known = agentsRegistry.KNOWN_AGENTS.find((a) => a.id === id);
+      const label = known ? known.label : id;
+      throw new Error(`${label} isn't installed here (its command wasn't found on PATH). Install it, then try again.`);
+    }
+    cfg.agent = id;
+    // Seed a default runner if this agent was never configured (e.g. installed after init/update).
+    const known = agentsRegistry.KNOWN_AGENTS.find((a) => a.id === id);
+    if (known) { cfg.runners = cfg.runners || {}; if (!cfg.runners[id]) cfg.runners[id] = known.runner; }
+  }
   fs.writeFileSync(cp, JSON.stringify(cfg, null, 2) + '\n');
   return cfg;
 }
@@ -143,6 +164,19 @@ const server = http.createServer(async (req,res)=>{
       return sendJSON(res,200,{runId:r.runId});
     }
 
+    // ---- chat context management: condense the log via the agent, or wipe it ----
+    if (p === '/api/chat/summarize' && req.method === 'POST') {
+      const { agent } = await body(req);
+      const r = runSummarize(ROOT, { agent }, emit);
+      if (r.error) return sendJSON(res, 400, { error: r.error });
+      return sendJSON(res, 200, { ok: true });
+    }
+    if (p === '/api/chat/clear' && req.method === 'POST') {
+      const rt = store.readRuntime(ROOT); rt.messages = []; store.writeRuntime(ROOT, rt);
+      emit({ type: 'change' });
+      return sendJSON(res, 200, { ok: true });
+    }
+
     // ---- orchestrator ----
     if (p === '/api/orchestrate' && req.method === 'POST') {
       const { request } = await body(req);

package/templates/dashboard/summarize.js

@@ -0,0 +1,51 @@
+'use strict';
+/*
+ * Chat "Summarize" — condenses the recent group-chat log into one digest message, via the same
+ * configured agent runner (kept separate from runner.js since it captures the child's raw stdout as
+ * one summary, not sentinel-parsed lines; unit-testable without an HTTP server, same as runner.js).
+ */
+const { spawn } = require('child_process');
+const store = require('../lib/store');
+
+const DEFAULT_LIMIT = 40;
+
+// "role: text" lines, oldest first, capped to the most recent `limit` entries.
+function formatLog(messages, limit = DEFAULT_LIMIT) {
+  return (messages || []).slice(-limit).map((m) => `${m.role}: ${m.text}`).join('\n');
+}
+
+// Summarizes runtime.messages (excluding prior summaries, so re-summarizing doesn't compound) into
+// one new message of kind 'summary'. Returns { child } on success or { error } — mirrors startRun's
+// shape, without the sentinel-line streaming runner.js does (this is a one-shot digest, not a task run).
+function runSummarize(root, { agent } = {}, emit) {
+  const cfg = store.readConfig(root);
+  const which = agent || cfg.agent || 'claude';
+  const cmdStr = cfg.runners && cfg.runners[which];
+  if (!cmdStr) return { error: `No runner configured for "${which}".` };
+
+  const rt = store.readRuntime(root);
+  const messages = (rt.messages || []).filter((m) => m.kind !== 'summary');
+  if (!messages.length) return { error: 'Nothing to summarize yet.' };
+
+  const prompt = 'Summarize this project\'s recent activity log in 3-6 concise bullet points — '
+    + 'what was built, decided, or is blocked. Reply with the summary only, no preamble or sentinel lines.'
+    + '\n\n' + formatLog(messages);
+
+  const parts = cmdStr.split(/\s+/).filter(Boolean);
+  let child;
+  try { child = spawn(parts[0], [...parts.slice(1), prompt], { cwd: root, env: process.env }); }
+  catch (e) { return { error: e.message }; }
+  try { child.stdin && child.stdin.end(); } catch {}
+
+  let out = '';
+  child.stdout && child.stdout.on('data', (d) => { out += d.toString(); });
+  child.stderr && child.stderr.on('data', (d) => { out += d.toString(); });
+  child.on('close', (code) => {
+    const text = out.trim() || (code === 0 ? '(no output)' : `summarize failed (exit ${code})`);
+    const m = store.appendMessage(root, { role: which, kind: 'summary', text, agent: which });
+    if (emit) { emit({ type: 'message', message: m }); emit({ type: 'change' }); }
+  });
+  return { child };
+}
+
+module.exports = { runSummarize, formatLog };

package/templates/lib/agents-registry.js

@@ -0,0 +1,52 @@
+'use strict';
+/*
+ * The dashboard's own view of "which coding agents exist and is one actually installed" — a small,
+ * self-contained subset of this package's lib/adapters.js (the richer install-time registry with
+ * memory-file content). Duplicated rather than shared: .spectoflow/ must be self-contained (ships
+ * into every project), while lib/adapters.js does not ship there. test/agents-registry.test.js
+ * guards the two id/bin/runner sets from drifting apart.
+ */
+const fs = require('fs');
+const path = require('path');
+
+const KNOWN_AGENTS = [
+  { id: 'claude', label: 'Claude Code', bin: 'claude', dirs: ['.claude'], runner: 'claude -p --permission-mode acceptEdits' },
+  { id: 'codex', label: 'Codex', bin: 'codex', dirs: ['.codex'], runner: 'codex exec' },
+  { id: 'cursor', label: 'Cursor', bin: 'cursor-agent', dirs: ['.cursor'], runner: 'cursor-agent -p' },
+  { id: 'gemini', label: 'Gemini CLI', bin: 'gemini', dirs: ['.gemini'], runner: 'gemini -p' },
+  { id: 'opencode', label: 'OpenCode', bin: 'opencode', dirs: ['.opencode'], runner: 'opencode run --quiet' },
+  { id: 'kiro', label: 'Kiro CLI', bin: 'kiro-cli', dirs: ['.kiro'], runner: 'kiro-cli chat --no-interactive --trust-all-tools' },
+  { id: 'antigravity', label: 'Antigravity', bin: 'agy', dirs: [], runner: 'agy -p' },
+];
+
+// Is `bin` an executable resolvable on PATH? On win32, an extension from PATHEXT is required, so we
+// try each; we also try the bare name (covers test fixtures and extensionless shims).
+function binOnPath(bin, { env = process.env, platform = process.platform } = {}) {
+  const raw = env.PATH || env.Path || '';
+  const dirs = raw.split(path.delimiter).filter(Boolean);
+  const exts =
+    platform === 'win32' ? ['', ...(env.PATHEXT || '.COM;.EXE;.BAT;.CMD').split(';').filter(Boolean)] : [''];
+  for (const d of dirs) {
+    for (const e of exts) {
+      if (fs.existsSync(path.join(d, bin + e))) return true;
+    }
+  }
+  return false;
+}
+
+// True if `id` looks genuinely installed: its bin resolves on PATH, or the project already has its
+// config dir (a project can be set up for an agent whose bin isn't on THIS machine's PATH, e.g. a
+// remote/CI runner). Unknown ids are never "installed".
+function isAgentInstalled(id, projectRoot, opts) {
+  const a = KNOWN_AGENTS.find((x) => x.id === id);
+  if (!a) return false;
+  if (a.bin && binOnPath(a.bin, opts)) return true;
+  return (a.dirs || []).some((d) => fs.existsSync(path.join(projectRoot, d)));
+}
+
+// ids of every known agent actually installed for this project, in KNOWN_AGENTS (priority) order.
+function installedAgents(projectRoot, opts) {
+  return KNOWN_AGENTS.filter((a) => isAgentInstalled(a.id, projectRoot, opts)).map((a) => a.id);
+}
+
+module.exports = { KNOWN_AGENTS, binOnPath, isAgentInstalled, installedAgents };

package/templates/lib/customize-prompts.js

@@ -0,0 +1,34 @@
+'use strict';
+// Builds the exact natural-language prompts the dashboard's Settings → Customize UI posts to
+// /api/run (see templates/dashboard/public/app.js's CZ_KINDS) — the single source of truth so the
+// CLI (`spectoflow skill/agent/dashboard create`) and the dashboard button never drift apart. The
+// browser side can't require this Node module (no build step), so its literal strings are mirrored
+// there by hand; test/customize-prompts.test.js guards against the two falling out of sync.
+const PROMPTS = {
+  dashboard: {
+    add: (d) => `Add a custom dashboard: ${d}`,
+    auto: 'Propose dashboard candidates for this project (Auto customize)',
+  },
+  skill: {
+    add: (d) => `Create a new skill: ${d}`,
+    auto: 'Propose skill candidates for this project (Auto customize)',
+  },
+  agent: {
+    add: (d) => `Create a new agent: ${d}`,
+    auto: 'Propose agent candidates for this project (Auto customize)',
+  },
+};
+
+// buildCustomizePrompt('skill', { description: 'reviews PRs for accessibility' })
+// buildCustomizePrompt('skill', { auto: true })
+function buildCustomizePrompt(kind, opts) {
+  const p = PROMPTS[kind];
+  if (!p) throw new Error(`Unknown customize kind "${kind}" (expected dashboard, skill or agent).`);
+  const o = opts || {};
+  if (o.auto) return p.auto;
+  const d = o.description && String(o.description).trim();
+  if (!d) throw new Error('A description is required unless --auto is passed.');
+  return p.add(d);
+}
+
+module.exports = { PROMPTS, buildCustomizePrompt };