@lazyingart/agintiflow 0.20.61 → 0.20.63

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lazyingart/agintiflow",
-  "version": "0.20.61",
+  "version": "0.20.63",
   "type": "module",
   "description": "Low-cost, project-aware Web and CLI agents with DeepSeek/Venice/OpenAI routing, visible tool calls, durable sessions, scouts, AAPS, SCS, and guarded local execution.",
   "license": "Apache-2.0",

package/scripts/smoke-cli-chat.js

@@ -17,6 +17,7 @@ import {
 } from "../src/interactive-cli.js";
 import { parseArgs } from "../src/cli.js";
 import { dockerPolicyTimeoutMs, dockerUserCommand } from "../src/docker-sandbox.js";
+import { formatBehaviorContractForPrompt } from "../src/behavior-contract.js";
 
 const repoRoot = path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..");
 const tempRoot = await fs.mkdtemp(path.join(os.tmpdir(), "agintiflow-cli-chat-"));
@@ -145,6 +146,13 @@ try {
   ) {
     throw new Error("elapsed duration formatter returned an unexpected value");
   }
+  const behaviorContract = formatBehaviorContractForPrompt();
+  if (
+    !behaviorContract.includes("name the actual environment used") ||
+    !behaviorContract.includes("Do not claim compatibility across untested runtimes")
+  ) {
+    throw new Error("behavior contract must guard against host/Docker and runtime-version overclaims");
+  }
   const wrappedDockerCommand = dockerUserCommand("node --test 2>&1 | tail -30", {
     category: "general-shell",
     needsNetwork: false,

package/scripts/smoke-skills.js

@@ -72,6 +72,9 @@ assert(selectedIds("write README API docs and a tutorial").includes("docs-knowle
 assert(selectedIds("fix failing tests and add regression coverage").includes("qa-testing"), "QA prompt did not select qa-testing");
 const qaSkill = skills.find((skill) => skill.id === "qa-testing");
 assert(qaSkill?.body.includes("Do not invent staged bugs"), "QA skill does not guard against fake staged failures");
+const pythonSkill = skills.find((skill) => skill.id === "python");
+assert(pythonSkill?.body.includes("PEP 701 relaxed f-strings"), "Python skill must mention 3.12 f-string compatibility traps");
+assert(pythonSkill?.body.includes("only proves the active interpreter"), "Python skill must guard syntax-check overclaims");
 assert(selectedIds("write SQL migrations for sqlite schema").includes("database"), "database prompt did not select database");
 assert(selectedIds("debug Docker deployment logs and port config").includes("devops-deployment"), "devops prompt did not select devops-deployment");
 assert(selectedIds("review auth security and secrets handling").includes("security-review"), "security prompt did not select security-review");

package/skills/python/SKILL.md

@@ -21,4 +21,8 @@ tools:
 
 Inspect `pyproject.toml`, requirements, package layout, and tests. Prefer stdlib tests when dependency installs are unnecessary; otherwise use project-local venv/conda/uv or Docker.
 
+When reporting syntax, lint, test, or runtime results, always include the actual interpreter path/version and whether commands ran on the host or in Docker/venv. `py_compile` only proves the active interpreter accepted the files.
+
+For compatibility claims, avoid grep-only feature scans. Treat these as version-sensitive unless the target interpreter is tested: f-strings require Python 3.6+, `:=` requires 3.8+, builtin generic hints such as `list[int]` require 3.9+ unless postponed annotations are used, `match/case` and `X | Y` type unions require 3.10+, `except*`/`ExceptionGroup` require 3.11+, and PEP 701 relaxed f-strings (backslashes, comments, or previously invalid quoting/nesting inside f-string expressions) require 3.12+. If host compatibility matters and only Docker was tested, state that limitation instead of saying the code is safe on the host.
+
 For plots and artifacts, save files with clear names and send useful outputs to canvas.

package/src/behavior-contract.js

@@ -70,6 +70,7 @@ export function formatBehaviorContractForPrompt({ mode = "runtime" } = {}) {
     "Define or infer concrete success criteria for non-trivial work, then run focused checks or state why checks are unavailable.",
     "Respect the permission contract: if a tool is blocked, stop and present the safe rerun/approval path instead of retrying variants.",
     "Keep artifacts durable and discoverable with descriptive non-conflicting names; never overwrite unless the user clearly asked.",
+    "When reporting shell, language, runtime, build, or test results, name the actual environment used (host vs Docker, relevant interpreter/tool path/version when it matters). Do not claim compatibility across untested runtimes, hosts, containers, or language versions; state the caveat or run an explicit check.",
   ].join(" ");
 }
 
@@ -117,6 +118,7 @@ function baseSections() {
     "- Every changed line should trace to the task or to cleanup caused by the task.",
     "- Match existing project style even if another style is personally preferable.",
     "- If you notice unrelated issues, report them separately rather than editing them.",
+    "- When reporting shell, language, runtime, build, or test results, name the actual environment used (host vs Docker, relevant interpreter/tool path/version when it matters). Do not claim compatibility across untested runtimes, hosts, containers, or language versions; state the caveat or run an explicit check.",
     "",
     "## Verification Contract",
     "",
@@ -287,4 +289,3 @@ export function buildAgintiInstructions(template = "disciplined") {
   const body = normalized === "minimal" ? minimalSections() : `${baseSections()}${templateAppendix(normalized)}`;
   return `${body}\n`;
 }
-

package/src/task-profiles.js

@@ -143,7 +143,7 @@ export const TASK_PROFILES = {
     id: "python",
     label: "Python",
     prompt:
-      "Bias toward Python best practices without ignoring non-Python project context. Inspect pyproject/requirements, create scripts/packages/tests as files, prefer project-local venv/conda/uv or Docker for setup, and run focused smoke checks when shell is enabled.",
+      "Bias toward Python best practices without ignoring non-Python project context. Inspect pyproject/requirements, create scripts/packages/tests as files, prefer project-local venv/conda/uv or Docker for setup, and run focused smoke checks when shell is enabled. For syntax, lint, test, or runtime claims, capture the actual interpreter path/version and sandbox/host environment; never claim support for untested Python versions or host interpreters just because a Docker/venv check passed. Remember Python syntax compatibility traps: f-strings require 3.6+, walrus 3.8+, builtin generics 3.9+, match/case and union type syntax 3.10+, except*/ExceptionGroup 3.11+, and relaxed PEP 701 f-string expressions with backslashes/comment/complex quoting require 3.12+.",
     tools: ["files", "shell", "sandbox"],
   },
   shell: {