loki-mode 8.3.3 → 8.4.0

package/README.md

@@ -15,7 +15,7 @@ _The free, source-available autonomous coding agent by [Autonomi](https://www.au
 
 [Website](https://www.autonomi.dev/) | [Documentation](wiki/Home.md) | [Installation](docs/INSTALLATION.md) | [Changelog](CHANGELOG.md) | [Purple Lab -- deprecated v7.44.0](#purple-lab)
 
-**Current release: v8.3.3**
+**Current release: v8.4.0**
 
 </div>
 
@@ -63,10 +63,15 @@ does not execute functions"
 ([continue#5696](https://github.com/continuedev/continue/issues/5696)). None of
 those seven publishes a machine-checkable completion artifact.
 
-We have not audited the closed-source products (Cursor, Devin, Replit Agent)
-feature by feature, so treat this as "unclaimed as far as we can verify" rather
-than a proven first. The receipt stands on its own either way: run the tour and
-check the hash.
+We measured every named competitor that ships a local CLI -- opencode 1.18.9,
+aider 0.86.2, codex-cli 0.146.0, Claude Code 2.1.220, cursor-agent -- and none
+exposes a command that verifies the agent's own output. Rerun it yourself with
+`bash tests/test-competitor-verify-surface.sh`.
+
+That is a measurement of the CLI surface, not of whole products: a web UI or an
+API could expose something `--help` does not, and Devin and Replit Agent ship no
+local CLI so they are not covered. The receipt stands on its own either way:
+run the tour and check the hash.
 
 **Evaluating this against something else?** [docs/EVALUATING.md](docs/EVALUATING.md)
 puts a runnable command next to every claim we make, and states plainly what we

package/VERSION

@@ -1 +1 @@
-8.3.3
+8.4.0

package/autonomy/run.sh

@@ -3971,6 +3971,30 @@ except Exception:
     print('')" "$_app_state_file" 2>/dev/null)"
     fi
 
+    # Time to first preview: how long until the user could SEE something running.
+    #
+    # WRITTEN BUT NEVER READ, until now. app-runner.sh:177 has recorded
+    # seconds_to_first_preview since v8, write-once and atomically, and NOTHING
+    # consumed it -- not this summary, not the receipt, not the dashboard. A
+    # measurement nobody surfaces cannot influence anything, so the number that
+    # research identifies as the delight peak of this whole category (a working
+    # preview at roughly four minutes) was invisible to the person who waited for
+    # it and to us.
+    #
+    # Read-only and best-effort by design: absent file means the app never
+    # previewed (or the run predates the writer), and we render nothing rather
+    # than a zero. A fabricated "0s to preview" would be worse than silence.
+    local first_preview_s=""
+    local _fp_file="$loki_dir/app-runner/first-preview.json"
+    if [ -f "$_fp_file" ]; then
+        first_preview_s="$(python3 -c "import json,sys
+try:
+    v=json.load(open(sys.argv[1])).get('seconds_to_first_preview')
+    print(int(v) if isinstance(v,(int,float)) and v>=0 else '')
+except Exception:
+    print('')" "$_fp_file" 2>/dev/null)"
+    fi
+
     # Branch + diff stats vs the run-start SHA (best-effort; non-git or empty
     # baseline yields empty values, which we render as "unknown"/"0").
     local start_sha="${_LOKI_RUN_START_SHA:-}"
@@ -4100,6 +4124,11 @@ except Exception:
         fi
         printf '%-14s %s\n' "Branch:" "$branch"
         printf '%-14s %s\n' "Files:" "$files_changed (+$insertions / -$deletions)"
+        # Only rendered when the app actually previewed. Silence beats a
+        # fabricated zero for a run where nothing ever came up.
+        if [ -n "$first_preview_s" ]; then
+            printf '%-14s %ss\n' "First preview:" "$first_preview_s"
+        fi
         printf '%-14s %s\n' "Finished:" "$ts"
         if [ -n "$delegate_branch" ]; then
             printf '%-14s %s\n' "Delegate:" "$delegate_branch"

package/dashboard/__init__.py

@@ -7,7 +7,7 @@ Modules:
     control: Session control API (start/stop/pause/resume)
 """
 
-__version__ = "8.3.3"
+__version__ = "8.4.0"
 
 # Expose the control app for easy import
 try:

package/docs/EVALUATING.md

@@ -114,6 +114,48 @@ check you run at the end and a check something else can call as a dependency.
 
 ---
 
+## 6. Head to head, on this machine, in ten seconds
+
+The one competitive comparison we will publish, because you can rerun it:
+
+```bash
+bash tests/test-competitor-verify-surface.sh
+```
+
+It runs `--help` on whatever agent CLIs are installed and checks whether any of
+them exposes a command that verifies the agent's OWN output.
+
+Measured 2026-07-30 against every one of the six named competitors that ships
+a local CLI:
+
+| CLI | Verifies its own output |
+| --- | --- |
+| opencode 1.18.9 | no |
+| aider 0.86.2 | no |
+| codex-cli 0.146.0 | no |
+| Claude Code 2.1.220 | no |
+| cursor-agent 2026.05.24 | no |
+| loki-mode 8.3.3 | yes (`loki proof verify`) |
+
+Devin and Replit Agent ship no local CLI, so they are not covered and no claim
+is made about them.
+
+**The detail that makes this honest.** aider matches eight times on
+verify/proof/attest terms. All eight are `--verify-ssl` (TLS certificate
+validation) and `--git-commit-verify` (git pre-commit hooks). Neither verifies
+the agent's output. A count-based comparison would have scored aider 8 against
+our 2 and concluded the opposite of the truth, so the test reads every hit
+instead of counting them.
+
+**Scope.** This inspects the CLI surface of locally installed tools. Devin and
+Replit Agent ship no local CLI and are not covered. Absence from a `--help`
+listing is also not proof of absence from a product: a web UI or an API could
+expose something the CLI does not. If a competitor does ship output
+verification, this test is how we would find out, and the claim above would be
+corrected rather than defended.
+
+---
+
 ## What we do not have
 
 Stating this plainly, because you will find it out anyway and it is cheaper for
@@ -124,11 +166,14 @@ both of us if you find it here.
 - **No independent third-party benchmark placement.** The SWE-bench Verified
   leaderboard is months stale and every entry on it is self-reported, ours would
   be too.
-- **No audit of the closed-source products.** We have verified that seven open
-  harnesses (OpenHands, Cline, Aider, SWE-agent, Roo-Code, OpenCode, Continue)
-  publish no machine-checkable completion artifact. Cursor, Devin, Replit Agent
-  and Claude Code we have **not** audited feature by feature, so treat the
-  receipt as "unclaimed as far as we can verify", not as a proven first.
+- **Only the CLI surface is audited, not whole products.** Section 6 measures
+  every named competitor that ships a local CLI (opencode, aider, codex, Claude
+  Code, cursor-agent) and none exposes output verification. That is a real
+  measurement you can rerun, but it is a measurement of `--help`, not of a
+  product: a web UI or an API could expose something the CLI does not. Devin and
+  Replit Agent ship no local CLI and are not covered at all. Treat this as
+  "unclaimed on every surface we can reach", which is stronger than the earlier
+  "unclaimed as far as we can verify" and still short of a proven first.
 - **Generation is not air-gapped.** See section 2.
 
 ---

package/loki-ts/dist/loki.js

@@ -1,5 +1,5 @@
 // @bun
-var S_=Object.create;var{getPrototypeOf:y_,defineProperty:oK,getOwnPropertyNames:b_}=Object;var __=Object.prototype.hasOwnProperty;function f_(Z){return this[Z]}var h_,v_,g_=(Z,X,Q)=>{var Y=Z!=null&&typeof Z==="object";if(Y){var J=X?h_??=new WeakMap:v_??=new WeakMap,z=J.get(Z);if(z)return z}Q=Z!=null?S_(y_(Z)):{};let K=X||!Z||!Z.__esModule?oK(Q,"default",{value:Z,enumerable:!0}):Q;for(let $ of b_(Z))if(!__.call(K,$))oK(K,$,{get:f_.bind(Z,$),enumerable:!0});if(Y)J.set(Z,K);return K};var VQ=(Z,X)=>()=>(X||Z((X={exports:{}}).exports,X),X.exports);var m_=(Z)=>Z;function u_(Z,X){this[Z]=m_.bind(null,X)}var c0=(Z,X)=>{for(var Q in X)oK(Z,Q,{get:X[Q],enumerable:!0,configurable:!0,set:u_.bind(X,Q)})};var p=(Z,X)=>()=>(Z&&(X=Z(Z=0)),X);var t0=import.meta.require;var IO={};c0(IO,{lokiDir:()=>j0,homeLokiDir:()=>I4,findRepoRootForVersion:()=>tK,REPO_ROOT:()=>Q8});import{resolve as s7,dirname as rK}from"path";import{fileURLToPath as p_}from"url";import{existsSync as qQ}from"fs";import{homedir as d_}from"os";function c_(){let Z=EO;for(let X=0;X<6;X++){if(qQ(s7(Z,"VERSION"))&&qQ(s7(Z,"autonomy/run.sh")))return Z;let Q=rK(Z);if(Q===Z)break;Z=Q}return s7(EO,"..","..","..")}function tK(Z){let X=Z;for(let Q=0;Q<6;Q++){if(qQ(s7(X,"VERSION"))&&qQ(s7(X,"autonomy/run.sh")))return X;let Y=rK(X);if(Y===X)break;X=Y}return s7(Z,"..","..","..")}function j0(){return process.env.LOKI_DIR??s7(process.cwd(),".loki")}function I4(){return s7(d_(),".loki")}var EO,Q8;var G8=p(()=>{EO=rK(p_(import.meta.url));Q8=c_()});import{readFileSync as l_}from"fs";import{resolve as i_,dirname as a_}from"path";import{fileURLToPath as s_}from"url";function _3(){if(f5!==null)return f5;let Z="8.3.3";if(typeof Z==="string"&&Z.length>0)return f5=Z,f5;try{let X=a_(s_(import.meta.url)),Q=tK(X);f5=l_(i_(Q,"VERSION"),"utf-8").trim()}catch{f5="unknown"}return f5}var f5=null;var GQ=p(()=>{G8()});var xO={};c0(xO,{runOrThrow:()=>qf,run:()=>E0,readStreamCapped:()=>HQ,commandVersion:()=>Hf,commandExists:()=>X9,ShellError:()=>eK,MAX_STDOUT_BYTES:()=>kO});async function HQ(Z,X=kO){let Q=Z.getReader(),Y=new TextDecoder,J="",z=0;try{while(z<X){let{done:K,value:$}=await Q.read();if(K)break;if(!$)continue;if(z+=$.byteLength,z>X){let W=$.byteLength-(z-X);J+=Y.decode($.subarray(0,W),{stream:!0});break}J+=Y.decode($,{stream:!0})}J+=Y.decode()}finally{try{await Q.cancel()}catch{}Q.releaseLock()}return J}async function E0(Z,X={}){let Q=Bun.spawn({cmd:[...Z],stdout:"pipe",stderr:"pipe",env:X.env?{...process.env,...X.env}:process.env,cwd:X.cwd}),Y,J;if(X.timeoutMs&&X.timeoutMs>0)Y=setTimeout(()=>{try{Q.kill("SIGTERM")}catch{}J=setTimeout(()=>{try{Q.kill("SIGKILL")}catch{}},2000)},X.timeoutMs);try{let[z,K,$]=await Promise.all([HQ(Q.stdout),new Response(Q.stderr).text(),Q.exited]);return{stdout:z,stderr:K,exitCode:$}}finally{if(Y)clearTimeout(Y);if(J)clearTimeout(J)}}async function qf(Z,X={}){let Q=await E0(Z,X);if(Q.exitCode!==0)throw new eK(`command failed (${Q.exitCode}): ${Z.join(" ")}`,Q.exitCode,Q.stdout,Q.stderr);return Q}async function X9(Z){let X=Gf(Z),Q=await E0(["sh","-c",`command -v ${X}`],{timeoutMs:5000});if(Q.exitCode===0)return Q.stdout.trim()||null;return null}function Gf(Z){if(!/^[A-Za-z0-9._/-]+$/.test(Z))throw Error(`refused to shell-escape suspect token: ${Z}`);return Z}async function Hf(Z,X="--version"){if(!await X9(Z))return null;let Y=await E0([Z,X],{timeoutMs:5000});if(Y.exitCode!==0)return null;return((Y.stdout||Y.stderr).split(/\r?\n/)[0]?.trim()??"")||null}var kO=16777216,eK;var k9=p(()=>{eK=class eK extends Error{message;exitCode;stdout;stderr;constructor(Z,X,Q,Y){super(Z);this.message=Z;this.exitCode=X;this.stdout=Q;this.stderr=Y;this.name="ShellError"}}});function n7(Z){return Uf?"":Z}var Uf,M0,k8,l0,dW0,a0,H8,Q9,g;var S6=p(()=>{Uf=(process.env.NO_COLOR??"").length>0;M0=n7("\x1B[0;31m"),k8=n7("\x1B[0;32m"),l0=n7("\x1B[1;33m"),dW0=n7("\x1B[0;34m"),a0=n7("\x1B[0;36m"),H8=n7("\x1B[1m"),Q9=n7("\x1B[2m"),g=n7("\x1B[0m")});import{existsSync as Ff}from"fs";async function F7(){if(R4!==void 0)return R4;let Z="/opt/homebrew/bin/python3.12";if(Ff(Z))return R4=Z,Z;let X=await X9("python3.12");if(X)return R4=X,X;let Q=await X9("python3");return R4=Q,Q}async function D7(Z,X={}){let Q=await F7();if(!Q)return{stdout:"",stderr:"python3 not found",exitCode:127};return E0([Q,"-c",Z],X)}var R4;var o7=p(()=>{k9()});var rO={};c0(rO,{runStatus:()=>tf});import{existsSync as Y9,readFileSync as h3,readdirSync as dO,statSync as cO}from"fs";import{resolve as h8,basename as pf}from"path";import{homedir as df}from"os";function lO(Z){let X=Math.trunc(Z);if(X>=1e6)return`${(Math.trunc(X/1e6*10)/10).toFixed(1)}M`;if(X>=1000)return`${(Math.trunc(X/1000*10)/10).toFixed(1)}K`;return String(X)}function iO(Z,X,Q){if(X===0)return null;let Y=Math.trunc(Z*100/X),J=Math.trunc(Z*BQ/X);if(J>BQ)J=BQ;let z=BQ-J,K=k8;if(Y>=80)K=M0;else if(Y>=50)K=l0;let $="=".repeat(Math.max(0,J))+" ".repeat(Math.max(0,z)),W=lO(Z),V=lO(X);return`  ${H8}${Q}${g} ${K}[${$}]${g} ${Y}% (${W} / ${V})`}async function lf(){if(await X9("jq"))return!0;return process.stdout.write(`${M0}Error: jq is required but not installed.${g}
+var S_=Object.create;var{getPrototypeOf:y_,defineProperty:oK,getOwnPropertyNames:b_}=Object;var __=Object.prototype.hasOwnProperty;function f_(Z){return this[Z]}var h_,v_,g_=(Z,X,Q)=>{var Y=Z!=null&&typeof Z==="object";if(Y){var J=X?h_??=new WeakMap:v_??=new WeakMap,z=J.get(Z);if(z)return z}Q=Z!=null?S_(y_(Z)):{};let K=X||!Z||!Z.__esModule?oK(Q,"default",{value:Z,enumerable:!0}):Q;for(let $ of b_(Z))if(!__.call(K,$))oK(K,$,{get:f_.bind(Z,$),enumerable:!0});if(Y)J.set(Z,K);return K};var VQ=(Z,X)=>()=>(X||Z((X={exports:{}}).exports,X),X.exports);var m_=(Z)=>Z;function u_(Z,X){this[Z]=m_.bind(null,X)}var c0=(Z,X)=>{for(var Q in X)oK(Z,Q,{get:X[Q],enumerable:!0,configurable:!0,set:u_.bind(X,Q)})};var p=(Z,X)=>()=>(Z&&(X=Z(Z=0)),X);var t0=import.meta.require;var IO={};c0(IO,{lokiDir:()=>j0,homeLokiDir:()=>I4,findRepoRootForVersion:()=>tK,REPO_ROOT:()=>Q8});import{resolve as s7,dirname as rK}from"path";import{fileURLToPath as p_}from"url";import{existsSync as qQ}from"fs";import{homedir as d_}from"os";function c_(){let Z=EO;for(let X=0;X<6;X++){if(qQ(s7(Z,"VERSION"))&&qQ(s7(Z,"autonomy/run.sh")))return Z;let Q=rK(Z);if(Q===Z)break;Z=Q}return s7(EO,"..","..","..")}function tK(Z){let X=Z;for(let Q=0;Q<6;Q++){if(qQ(s7(X,"VERSION"))&&qQ(s7(X,"autonomy/run.sh")))return X;let Y=rK(X);if(Y===X)break;X=Y}return s7(Z,"..","..","..")}function j0(){return process.env.LOKI_DIR??s7(process.cwd(),".loki")}function I4(){return s7(d_(),".loki")}var EO,Q8;var G8=p(()=>{EO=rK(p_(import.meta.url));Q8=c_()});import{readFileSync as l_}from"fs";import{resolve as i_,dirname as a_}from"path";import{fileURLToPath as s_}from"url";function _3(){if(f5!==null)return f5;let Z="8.4.0";if(typeof Z==="string"&&Z.length>0)return f5=Z,f5;try{let X=a_(s_(import.meta.url)),Q=tK(X);f5=l_(i_(Q,"VERSION"),"utf-8").trim()}catch{f5="unknown"}return f5}var f5=null;var GQ=p(()=>{G8()});var xO={};c0(xO,{runOrThrow:()=>qf,run:()=>E0,readStreamCapped:()=>HQ,commandVersion:()=>Hf,commandExists:()=>X9,ShellError:()=>eK,MAX_STDOUT_BYTES:()=>kO});async function HQ(Z,X=kO){let Q=Z.getReader(),Y=new TextDecoder,J="",z=0;try{while(z<X){let{done:K,value:$}=await Q.read();if(K)break;if(!$)continue;if(z+=$.byteLength,z>X){let W=$.byteLength-(z-X);J+=Y.decode($.subarray(0,W),{stream:!0});break}J+=Y.decode($,{stream:!0})}J+=Y.decode()}finally{try{await Q.cancel()}catch{}Q.releaseLock()}return J}async function E0(Z,X={}){let Q=Bun.spawn({cmd:[...Z],stdout:"pipe",stderr:"pipe",env:X.env?{...process.env,...X.env}:process.env,cwd:X.cwd}),Y,J;if(X.timeoutMs&&X.timeoutMs>0)Y=setTimeout(()=>{try{Q.kill("SIGTERM")}catch{}J=setTimeout(()=>{try{Q.kill("SIGKILL")}catch{}},2000)},X.timeoutMs);try{let[z,K,$]=await Promise.all([HQ(Q.stdout),new Response(Q.stderr).text(),Q.exited]);return{stdout:z,stderr:K,exitCode:$}}finally{if(Y)clearTimeout(Y);if(J)clearTimeout(J)}}async function qf(Z,X={}){let Q=await E0(Z,X);if(Q.exitCode!==0)throw new eK(`command failed (${Q.exitCode}): ${Z.join(" ")}`,Q.exitCode,Q.stdout,Q.stderr);return Q}async function X9(Z){let X=Gf(Z),Q=await E0(["sh","-c",`command -v ${X}`],{timeoutMs:5000});if(Q.exitCode===0)return Q.stdout.trim()||null;return null}function Gf(Z){if(!/^[A-Za-z0-9._/-]+$/.test(Z))throw Error(`refused to shell-escape suspect token: ${Z}`);return Z}async function Hf(Z,X="--version"){if(!await X9(Z))return null;let Y=await E0([Z,X],{timeoutMs:5000});if(Y.exitCode!==0)return null;return((Y.stdout||Y.stderr).split(/\r?\n/)[0]?.trim()??"")||null}var kO=16777216,eK;var k9=p(()=>{eK=class eK extends Error{message;exitCode;stdout;stderr;constructor(Z,X,Q,Y){super(Z);this.message=Z;this.exitCode=X;this.stdout=Q;this.stderr=Y;this.name="ShellError"}}});function n7(Z){return Uf?"":Z}var Uf,M0,k8,l0,dW0,a0,H8,Q9,g;var S6=p(()=>{Uf=(process.env.NO_COLOR??"").length>0;M0=n7("\x1B[0;31m"),k8=n7("\x1B[0;32m"),l0=n7("\x1B[1;33m"),dW0=n7("\x1B[0;34m"),a0=n7("\x1B[0;36m"),H8=n7("\x1B[1m"),Q9=n7("\x1B[2m"),g=n7("\x1B[0m")});import{existsSync as Ff}from"fs";async function F7(){if(R4!==void 0)return R4;let Z="/opt/homebrew/bin/python3.12";if(Ff(Z))return R4=Z,Z;let X=await X9("python3.12");if(X)return R4=X,X;let Q=await X9("python3");return R4=Q,Q}async function D7(Z,X={}){let Q=await F7();if(!Q)return{stdout:"",stderr:"python3 not found",exitCode:127};return E0([Q,"-c",Z],X)}var R4;var o7=p(()=>{k9()});var rO={};c0(rO,{runStatus:()=>tf});import{existsSync as Y9,readFileSync as h3,readdirSync as dO,statSync as cO}from"fs";import{resolve as h8,basename as pf}from"path";import{homedir as df}from"os";function lO(Z){let X=Math.trunc(Z);if(X>=1e6)return`${(Math.trunc(X/1e6*10)/10).toFixed(1)}M`;if(X>=1000)return`${(Math.trunc(X/1000*10)/10).toFixed(1)}K`;return String(X)}function iO(Z,X,Q){if(X===0)return null;let Y=Math.trunc(Z*100/X),J=Math.trunc(Z*BQ/X);if(J>BQ)J=BQ;let z=BQ-J,K=k8;if(Y>=80)K=M0;else if(Y>=50)K=l0;let $="=".repeat(Math.max(0,J))+" ".repeat(Math.max(0,z)),W=lO(Z),V=lO(X);return`  ${H8}${Q}${g} ${K}[${$}]${g} ${Y}% (${W} / ${V})`}async function lf(){if(await X9("jq"))return!0;return process.stdout.write(`${M0}Error: jq is required but not installed.${g}
 `),process.stdout.write(`Install with:
 `),process.stdout.write(`  brew install jq    (macOS)
 `),process.stdout.write(`  apt install jq     (Debian/Ubuntu)
@@ -1206,4 +1206,4 @@ Set LOKI_LEGACY_BASH=1 to force the bash CLI for every command.
 `),2}case"start":{let{runStart:Y}=await Promise.resolve().then(() => (R_(),P_));return Y(Q)}default:return process.stderr.write(`Unknown command: ${X}
 `),process.stderr.write(k_),2}}uO();process.on("SIGINT",()=>process.exit(130));process.on("SIGTERM",()=>process.exit(143));var DW0=await FW0(Bun.argv.slice(2));process.exit(DW0);
 
-//# debugId=0387E93FA065D70464756E2164756E21
+//# debugId=8285CFE544C8109D64756E2164756E21

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "loki-mode",
   "mcpName": "io.github.asklokesh/loki-mode",
-  "version": "8.3.3",
+  "version": "8.4.0",
   "description": "Loki Mode by Autonomi. Autonomous spec-to-product system: takes a PRD, GitHub issue, OpenAPI/JSON/YAML, or one-line brief to a deployed app via the RARV-C closure loop with 8 quality gates. Provider-agnostic (Claude Code, OpenAI Codex, Cline, Aider).",
   "keywords": [
     "agent",

package/plugins/loki-mode/.claude-plugin/plugin.json

@@ -2,7 +2,7 @@
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "loki-mode",
   "displayName": "Loki Mode",
-  "version": "8.3.3",
+  "version": "8.4.0",
   "description": "Autonomous spec-to-product build system with a built-in trust layer (RARV-C closure loop, 8 quality gates, completion council). Ships Loki's spec-hardening, drift-detection, and deterministic PR verification commands plus the Loki MCP server.",
   "author": {
     "name": "Autonomi",