@javierfrancoleal/accountability-auditor 0.0.0-stage → 1.0.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 J.Franco
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -1,3 +1,42 @@
-# Temporary Holding Version
+---
+description: "Mandatory reflection and verification pass that emits an Accountability Report before every final output — stage 3 of the agentic workflow."
+kind: "package-bundle"
+---
 
-This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
+# @javierfrancoleal/accountability-auditor
+
+> ⚡ **Super Agent Ready** — part of the J.Franco governance layer for autonomous agents.
+
+Created by **J.Franco**.
+
+## Summary
+
+Stage 3 of the J.Franco agentic workflow:
+
+`Input → Detection (AI + human) → AI processes, decides, and proposes → Human reflects, verifies, and decides → Output`
+
+This plugin registers the `accountability-auditor` skill on the host skill
+registry. The skill teaches the agent to run a mandatory reflection and
+verification pass before every final output: audit the input trail
+(`input-detector`), the decision trail (`decision-router`), and the actions
+taken, then emit a compact Accountability Report the human reviews.
+
+## Use this plugin
+
+In Desktop, open **Plugins** and enable **Accountability Auditor**. The skill
+appears in the session skill catalog; agents load it automatically before
+final outputs, and you can invoke it yourself by name to audit any previous
+work.
+
+## How it works
+
+- [`lib/index.js`](lib/index.js) registers a bundled skill provider on
+  `ctx.skills`, following the `@deepseek-ai/dsh-skill-badge` provider pattern.
+- [`assets/accountability-auditor.md`](assets/accountability-auditor.md) holds
+  the skill body.
+- [`cordis.patch.yml`](cordis.patch.yml) inserts the provider row into the host
+  plane of the Cordis tree, so its skills land in the global registry layer.
+
+## License
+
+MIT — see the package license field.

package/assets/accountability-auditor.md

@@ -0,0 +1,128 @@
+# Accountability Auditor
+
+Final stage of the J.Franco agentic workflow:
+
+`Input → Detection (AI + human) → AI processes, decides, and proposes → Human reflects, verifies, and decides → Output`
+
+## Purpose
+
+Prepare the human's verification and decision before any output is delivered
+as final. You reflect on and verify everything that happened, and hand the
+human a compact Accountability Report with two clear sections: the automatic
+actions to VERIFY, and the proposals to DECIDE. The human is the decider of
+record; the output is not final until the human has decided.
+
+## When to run
+
+- Before the FINAL answer of any task that changed files, ran commands, or
+  made decisions (mandatory).
+- On demand, when the human asks for an audit or review (class REVIEW).
+
+## Milestone reports
+
+Long missions are reported by milestone, not only at the end:
+
+- Before each milestone completes (or at the human's cadence), deliver the
+  compact report for the stretch just finished.
+- Mark what changed since the last report, and what remains open.
+- The final report at mission end summarizes every milestone; each milestone
+  entry is also appended to the supervision log.
+- If the human is away (autonomous mode), leave the report queued in the log
+  and say so on their return.
+
+## Watchdog pass
+
+Beyond verifying results, watch for progress itself, like a watchdog:
+
+- Is the work actually advancing, or repeating the same step?
+- Is the direction still aligned with the goal and the mission scope?
+- Are there stalled background jobs or unanswered proposals?
+
+When something is stuck or off-track, say so in the report — mark it URGENT
+when the human must intervene — and propose the recovery, following the
+recovery protocol (retry once with a correction, then classify as
+RECOVERABLE / NEEDS HUMAN / SAFETY and act accordingly; never hide a
+failure).
+
+## Verification checklist
+
+Run through these checks before delivering:
+
+1. **Inputs** — were all inputs detected, split into actions, and classified,
+   with their source (user or third party) and expectation?
+   (`input-detector`)
+2. **Decisions** — does every action have a tier, justification, and approval
+   reference? (`decision-router`)
+3. **Actions** — did every action claimed actually happen, and in the right
+   place? Re-check that files exist, values were noted in the right
+   application, commands ran, and outputs are real. Never claim an unverified
+   step.
+4. **Proposals** — is what you are handing the human clear enough to decide
+   on? State the options and your recommendation.
+5. **Results** — do the results match what was asked? Re-read the request
+   (including any third-party expectation) and compare.
+6. **Side effects** — anything modified beyond the stated scope? List it.
+7. **Gaps** — anything skipped, assumed, or left unverified? Say so
+   explicitly.
+
+## Accountability Report
+
+Deliver this compact report right BEFORE the output, addressed to the human,
+with the two sections separated:
+
+- **Worker:** the worker tag when provided (this is how the owner knows who
+  did what)
+- **Inputs detected:** one line (source + expectation, split into actions)
+- **Automatic actions — for you to verify:** one line per GREEN or
+  SEMI-AUTOMATIC action with evidence (what was done, where, and the value or
+  result, for example "noted the phone number 555-1234 in the client app");
+  mark SEMI-AUTOMATIC actions with a ⚑ flag — they are mandatory for the
+  human to check
+- **Proposals — for you to decide:** one line per YELLOW proposal, with the
+  options, your recommendation, and what each option changes
+- **Decisions:** one line per decision (what, tier, why, approval)
+- **Verification:** what you checked and what you found
+- **Review points:** what the human should look at before trusting the result
+- **Gaps:** anything unverified, plainly stated
+
+Keep it tight; offer the full trail separately only if the human asks.
+
+## Supervision log
+
+If a supervision file exists — `supervision.md` in the workspace root,
+`.dsh/supervision.md` in the workspace, or `~/.dsh/supervision.md` — append
+each Accountability Report to it after delivery: a dated entry with the
+inputs detected, the decisions (tiers and approvals), the automatic actions,
+and the human's decisions. This file is the owner's living record of what
+the AI did in the company; never edit or remove earlier entries. The
+`autonomy-grader` skill reads this log to propose tier moves.
+
+## Worked example: bank advisor at reception
+
+After noting the client's phone number (GREEN) and preparing a loan quote
+(YELLOW), the report reads:
+
+- **Inputs detected:** 3 actions from the client — note phone number, loan
+  quote, mortgage rate.
+- **Automatic actions — for you to verify:** noted phone number 555-1234 in
+  the client app.
+- **Proposals — for you to decide:** send the client the loan quote for
+  $120,000 at 36 months (my recommendation: confirm the amount first);
+  answer the client with today's mortgage rate of 3.2% (source: internal
+  rates page).
+- **Verification:** re-opened the client app and confirmed the number is
+  saved under the right record.
+- **Review points:** check the quote amount matches what the client asked.
+- **Gaps:** the client's income documents were not requested yet.
+
+## Rules
+
+- The report is part of the output, not an optional appendix. If the final
+  output takes three lines, the report takes three more — that cost is the
+  point.
+- If verification FAILED, do not present the result as final. State what is
+  unverified and stop for the human.
+- If the decision trail has holes, report them as gaps; never invent
+  justifications after the fact.
+- The human's decision on a YELLOW proposal is the anchor of accountability;
+  quote it, don't paraphrase it.

package/cordis.patch.yml

@@ -0,0 +1,6 @@
+# @javierfrancoleal/accountability-auditor — J.Franco's agentic-workflow plugin, stage 3.
+# Inserts the skill provider into the host plane of the Cordis tree, so every
+# agent preset (standard included) sees its skill through the global layer.
+- insert:
+    - id: accountability-auditor
+      name: '@javierfrancoleal/accountability-auditor'

package/icon.svg

@@ -0,0 +1,6 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 64 64" width="64" height="64">
+  <rect width="64" height="64" rx="14" fill="#4D6BFE"/>
+  <circle cx="32" cy="36" r="16" fill="none" stroke="#ffffff" stroke-width="4"/>
+  <path d="M32 28 v8 l6 4" fill="none" stroke="#ffffff" stroke-width="4" stroke-linecap="round"/>
+  <path d="M14 8 l6 4 M50 8 l-6 4" stroke="#ffffff" stroke-width="4" stroke-linecap="round"/>
+</svg>

package/lib/index.js

@@ -0,0 +1,56 @@
+// @javierfrancoleal/accountability-auditor — agentic-workflow stage 3: reflection and
+// verification. Created by J.Franco. Registers a bundled skill provider on the
+// host skill registry, following the @deepseek-ai/dsh-skill-badge provider pattern.
+import { readFile } from "node:fs/promises";
+import { fileURLToPath } from "node:url";
+import { BUNDLED_SKILL_RANK } from "@deepseek-ai/dsh-skill";
+
+const PROVIDER_NAME = "jfranco-accountability-auditor";
+const SKILL_BODY_URL = new URL("../assets/accountability-auditor.md", import.meta.url);
+const RESOURCE_BASE = {
+  kind: "directory",
+  path: fileURLToPath(new URL("../assets/", import.meta.url)),
+};
+
+const CANDIDATE = {
+  name: "accountability-auditor",
+  description: "Mandatory reflection and verification pass before any output is final: audit the input trail, decision trail, and actions, then hand the human a compact Accountability Report separating the automatic actions to verify from the proposals to decide. Stage 3 of the J.Franco agentic workflow.",
+  invocation: {
+    modelInvocable: true,
+    userInvocable: true,
+  },
+  provider: PROVIDER_NAME,
+  source: "bundled",
+  resourceBase: RESOURCE_BASE,
+  rank: BUNDLED_SKILL_RANK,
+  locator: SKILL_BODY_URL,
+};
+
+const provider = {
+  name: PROVIDER_NAME,
+  list: () => Promise.resolve([CANDIDATE]),
+  async get(_candidate) {
+    return {
+      name: CANDIDATE.name,
+      description: CANDIDATE.description,
+      invocation: CANDIDATE.invocation,
+      provider: CANDIDATE.provider,
+      source: CANDIDATE.source,
+      resourceBase: RESOURCE_BASE,
+      content: await readFile(SKILL_BODY_URL, "utf8"),
+    };
+  },
+};
+
+/** Cordis plugin name. */
+const name = "accountability-auditor";
+
+/** Service required by the bundled provider. */
+const inject = ["skills"];
+
+/** Register the bundled skill provider on ctx.skills. */
+function apply(ctx) {
+  ctx.skills.registerProvider(() => provider);
+}
+
+export { apply, inject, name };

package/locale/ar.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "مدقّق المساءلة",
+    "description": "تأمل وتحقق قبل كل نتيجة، مع تقرير يوضح ما عليك أن تقرره. المرحلة الثالثة."
+  }
+}

package/locale/de.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "Rechenschafts-Prüfer",
+    "description": "Reflexion und Prüfung vor jedem Ergebnis, mit einem Bericht über Ihre Entscheidungen. Stufe 3."
+  }
+}

package/locale/en.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "Accountability Auditor",
+    "description": "Reflection and verification before every output, with a report stating what you need to decide. Stage 3 of the agentic workflow."
+  }
+}

package/locale/es.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "Auditor de Responsabilidad",
+    "description": "Reflexión y verificación antes de cada resultado, con un reporte que indica lo que debes decidir. Etapa 3."
+  }
+}

package/locale/fr.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "Auditeur de Responsabilité",
+    "description": "Réflexion et vérification avant chaque résultat, avec un rapport indiquant ce que vous devez décider. Étape 3."
+  }
+}

package/locale/it.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "Revisore di Responsabilità",
+    "description": "Riflessione e verifica prima di ogni risultato, con un rapporto su cosa decidere. Fase 3."
+  }
+}

package/locale/ja.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "説明責任監査人",
+    "description": "結果の前に振り返りと検証を行い、決定事項を報告。第3段階。"
+  }
+}

package/locale/ko.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "책임 감사관",
+    "description": "모든 결과 전에 성찰과 검증, 결정할 사항을 보고합니다. 3단계."
+  }
+}

package/locale/pt.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "Auditor de Responsabilidade",
+    "description": "Reflexão e verificação antes de cada resultado, com um relatório do que você deve decidir. Etapa 3."
+  }
+}

package/locale/zh.json

@@ -0,0 +1,6 @@
+{
+  "meta": {
+    "title": "问责审计员",
+    "description": "每个结果前的反思与验证，并附报告说明你需要决定什么。第3阶段。"
+  }
+}

package/package.json

@@ -1,6 +1,55 @@
 {
   "name": "@javierfrancoleal/accountability-auditor",
-  "version": "0.0.0-stage",
-  "stub": true,
-  "description": "Temporary package placeholder for staged publishing"
-}
+  "version": "1.0.0",
+  "description": "Agentic-workflow stage 3: mandatory reflection and verification pass that emits an Accountability Report before every final output.",
+  "author": "J.Franco",
+  "license": "MIT",
+  "type": "module",
+  "main": "lib/index.js",
+  "exports": {
+    ".": "./lib/index.js",
+    "./cordis.patch.yml": "./cordis.patch.yml",
+    "./package.json": "./package.json",
+    "./locale/*.json": "./locale/*.json"
+  },
+  "icon": "./icon.svg",
+  "files": [
+    "icon.svg",
+    "locale/*.json",
+    "lib/index.js",
+    "assets",
+    "cordis.patch.yml",
+    "README.md"
+  ],
+  "engines": {
+    "node": ">=24",
+    "dsh": ">=0.2.1-rc.0"
+  },
+  "dsh": {
+    "manifestVersion": 1,
+    "bundle": {
+      "patch": "./cordis.patch.yml"
+    }
+  },
+  "peerDependencies": {
+    "@deepseek-ai/cordis": "~4.0.5",
+    "@deepseek-ai/dsh-skill": ">=0.2.1-rc.5"
+  },
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/JavierFrancoLeal/jfranco-agentic-suite.git",
+    "directory": "jfranco-accountability-auditor"
+  },
+  "homepage": "https://github.com/JavierFrancoLeal/jfranco-agentic-suite",
+  "keywords": [
+    "deepseek-harness",
+    "agentic-workflow",
+    "super-agent",
+    "ai-assistant",
+    "accountability",
+    "plugins"
+  ],
+  "publishConfig": {
+    "access": "public"
+  }
+}