toga-ai 1.0.155 → 1.0.156

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -9,6 +9,7 @@
9
9
  | [Developer Generators (password, UUID)](features/dev-generators.md) | Two tiny **1.0 `App_` framework** convenience scripts for everyday developer needs. | test/team/generate_password.php, test/team/uuid.php |
10
10
  | [Forecast vs NetSuite Discrepancy Analysis](features/forecast-netsuite-discrepancy-analysis.md) | `team/forecast-netsuite/discrepancy_analysis.php` detects discrepancies between our **Forecast database** and **NetSuite** (the source of truth for all sales da | test/team/forecast-netsuite/discrepancy_analysis.php |
11
11
  | [TableView Builder (2.0 TableViews SQL generator)](features/tableview-builder.md) | `team/tableViewBuilder/` generates SQL `INSERT` statements for the **2.0 `TableViews`**, `TableViewFields`, and `TableViewJoins` tables from a plain SQL `SELECT | test/team/tableViewBuilder/TableViewGenerator.php, test/team/tableViewBuilder/index.php, test/team/tableViewBuilder/Instructions.md |
12
+ | [Talos Knowledge Base Pipeline (Uploader + Processor)](features/talos-kb-pipeline.md) | `team/talos/` holds the two-script web tooling that feeds the **TOGa Talos** (TOGa IQ) AI knowledge bases. | test/team/talos/kb_uploader.php, test/team/talos/kb_processor.php, test/team/talos/kb_processor.ini |
12
13
  | [TOGa 2.0 Client Onboarding SQL Generator](features/toga2-client-onboarding-sql.md) | `team/generate_toga2_onboarding_sql.php` generates the SQL needed to **onboard a new client** onto the 2.0 platform. | test/team/generate_toga2_onboarding_sql.php |
13
14
  | [TOGa 2.0 User Cross-Client Access SQL Generator](features/toga2-user-cross-client-access-sql.md) | `team/generate_toga2_user_access_sql.php` generates SQL to grant an existing 2.0 user from a **home client** access to a **cross client**. | test/team/generate_toga2_user_access_sql.php |
14
15
  | [URL & Domain Markdown Document Builder](features/url-domain-markdown-document.md) | `team/build_url_domain_markdown_document.php` generates a **markdown document of our URLs and domains** by pulling environments and domains from the 2.0 platfor | test/team/build_url_domain_markdown_document.php |
@@ -6,7 +6,7 @@ project: Test
6
6
  client: shared
7
7
  type: architecture
8
8
  status: active
9
- updated: 2026-06-16
9
+ updated: 2026-06-22
10
10
  owners: [jcardinal]
11
11
  files:
12
12
  - test/team/
@@ -20,6 +20,7 @@ related:
20
20
  - ./features/active-directory-auth-test.md
21
21
  - ./features/url-domain-markdown-document.md
22
22
  - ./features/dev-generators.md
23
+ - ./features/talos-kb-pipeline.md
23
24
  ---
24
25
 
25
26
  ## Summary
@@ -81,9 +82,12 @@ manually against the target database; they do not execute changes themselves.
81
82
  | `generate_toga2_onboarding_sql.php` | 1.0 | 2.0 client DBs | [toga2-client-onboarding-sql](./features/toga2-client-onboarding-sql.md) |
82
83
  | `generate_toga2_user_access_sql.php` | 1.0 | 2.0 client DBs | [toga2-user-cross-client-access-sql](./features/toga2-user-cross-client-access-sql.md) |
83
84
  | `generate_password.php`, `uuid.php` | 1.0 | dev utility | [dev-generators](./features/dev-generators.md) |
85
+ | `talos/kb_uploader.php`, `talos/kb_processor.php` | standalone | S3 + AWS Bedrock KBs | [talos-kb-pipeline](./features/talos-kb-pipeline.md) |
84
86
 
85
- **Retired / not documented:** `team/DOA_gitbook/` and `team/DOA_talos/` are being retired.
87
+ **Retired / not documented:** `team/DOA_gitbook/` is being retired.
86
88
 
87
89
  ## Change history
88
90
 
89
91
  - 2026-06-16 — Initial capture of the `team/` folder. Registered repo in `registry.json`.
92
+ - 2026-06-22 — Un-retired `team/DOA_talos/` → `team/talos/`; documented its uploader +
93
+ processor in [talos-kb-pipeline](./features/talos-kb-pipeline.md).
@@ -0,0 +1,145 @@
1
+ ---
2
+ title: Talos Knowledge Base Pipeline (Uploader + Processor)
3
+ framework: "1.0"
4
+ repo: test
5
+ project: Test
6
+ client: shared
7
+ type: feature
8
+ status: active
9
+ updated: 2026-06-22
10
+ owners: [jcardinal]
11
+ files:
12
+ - test/team/talos/kb_uploader.php
13
+ - test/team/talos/kb_processor.php
14
+ - test/team/talos/kb_processor.ini
15
+ related:
16
+ - ../architecture.md
17
+ ---
18
+
19
+ ## Summary
20
+
21
+ `team/talos/` holds the two-script web tooling that feeds the **TOGa Talos** (TOGa IQ) AI
22
+ knowledge bases. Content flows through S3 folders and ends up in **AWS Bedrock** knowledge
23
+ bases, which back the internal AI agent that answers developer questions.
24
+
25
+ - **`kb_uploader.php`** — the *intake* page. A user pastes a meeting transcript or uploads a
26
+ document; it lands in the S3 `upload/` folder of a chosen knowledge base.
27
+ - **`kb_processor.php`** — the *review/approve* console. A reviewer cleans transcripts (AI +
28
+ automated text rules), approves files into the `approved/` folder, writes a Bedrock metadata
29
+ sidecar, archives the raw source, and triggers a Bedrock ingestion (sync) job.
30
+
31
+ Both are **standalone PHP pages** — they do **not** bootstrap the 1.0 `App_` framework. They
32
+ load the AWS SDK from `../../vendor/autoload.php` (Composer `aws/aws-sdk-php`) and read all
33
+ settings from a shared `kb_processor.ini`. Target PHP 7.2+.
34
+
35
+ > History: this folder was previously named `DOA_talos` (the `DOA_` prefix marked it as
36
+ > deprecated-but-live). It was un-retired and renamed to `talos` on 2026-06-22.
37
+
38
+ ## S3 layout
39
+
40
+ All paths are relative to `S3_BUCKET` (`togaiq`) under `S3_ROOT_PATH` (`development-team/`).
41
+ Each **knowledge base** is a folder one level under the root (e.g. `development-team/<kb>/`),
42
+ discovered by listing common prefixes. Within each KB:
43
+
44
+ | Folder (config key) | Default | Role |
45
+ |---|---|---|
46
+ | `S3_UPLOAD_FOLDER` | `upload` | Raw intake — where the uploader drops files |
47
+ | `S3_APPROVED_FOLDER` | `approved` | Reviewed/approved content (+ `.metadata.json` sidecars) — the Bedrock data source |
48
+ | `S3_ARCHIVE_FOLDER` | `archive` | Raw source files after they've been approved |
49
+
50
+ A KB named **`general`** is treated specially: it is never auto-synced on approve/delete;
51
+ instead it is synced in bulk via the "Sync All General" flow.
52
+
53
+ ## kb_uploader.php — intake
54
+
55
+ Single page with two POST forms (toggled by a client-side mode switcher):
56
+
57
+ 1. **Meeting Transcription** (`submit_meeting`) — fields: date, meeting name, KB, raw
58
+ transcript. The meeting name is slash-sanitized; the transcript's repeated blank lines are
59
+ collapsed; the file is named **`YYYY-MM-DD - {MEETING_NAME}.txt`** and `putObject`'d to
60
+ `<root>/<kb>/upload/`. This filename pattern is what the processor later recognizes as a
61
+ meeting (`isMeetingTranscript()` regex).
62
+ 2. **File Upload** (`submit_file`) — fields: KB, document name, file. Extension is validated
63
+ against `ALLOWED_EXTENSIONS`; the file is stored as `{documentName}.{ext}` in the same
64
+ `upload/` folder via `putObject` with `SourceFile`.
65
+
66
+ Knowledge-base options come from `getKnowledgeBases()`. All output is escaped with
67
+ `htmlspecialchars()`.
68
+
69
+ ## kb_processor.php — review & approve
70
+
71
+ The page lists everything in every KB's `upload/` folder, split into **meetings** (filename
72
+ matches `^\d{4}-\d{2}-\d{2} - .+\.txt$`) and **documents**. It exposes AJAX actions (POST
73
+ `action=...`, JSON responses) plus an **Approved Files Browser** modal. Key actions:
74
+
75
+ | `action` | Purpose |
76
+ |---|---|
77
+ | `get_file_content` | Return an S3 object's text (transcript editor / file viewer) |
78
+ | `prepare_transcript` | AI clean → automated clean → prepend `date - title` header |
79
+ | `save_to_kb` | Approve a file (see flow below) |
80
+ | `download_file` | Stream any S3 key to the browser as an attachment (`Content-Disposition`) |
81
+ | `delete_file` / `delete_approved_file` | Delete from `upload/` / `approved/` (latter re-syncs the KB) |
82
+ | `list_approved_files` / `count_approved_files` | Populate the Approved Files Browser |
83
+ | `start_sync_all_general` / `sync_next_kb` / `get_sync_progress` | Session-backed batch sync of every KB's `general` data source, polled from the front-end |
84
+
85
+ ### Transcript cleaning (`prepare_transcript`)
86
+
87
+ 1. **AI cleaning** — `callAICleaningAPI()` POSTs to `AI_API_ENDPOINT` (the TOGa IQ AI service,
88
+ model `AI_MODEL_ID`, Bedrock Claude Haiku 4.5) with an `output_schema` requesting a single
89
+ `knowledge_doc` string. The lengthy `AI_SYSTEM_PROMPT` enforces a strict lossless plain-text
90
+ format (OVERVIEW + topical sections, verbatim code blocks, PascalCase tables / camelCase
91
+ columns). Retries twice with backoff on network errors and HTTP 429/5xx.
92
+ 2. **Automated cleaning** — `applyAutomatedCleaning()` strips markdown artifacts (backticks,
93
+ `> ` quoting, `# ` headings), normalizes blank lines around headings/lists, removes
94
+ `:contentReference[...]` markers, and applies the `REPLACEMENTS[...]` map from the ini
95
+ (mishear corrections, e.g. `Talus → Talos`). Replacement keys support `:c` (case-sensitive)
96
+ and `:w` (whole-word) flags.
97
+ 3. For meeting files, a `YYYY-MM-DD - Title` header is prepended.
98
+
99
+ ### Approve flow (`save_to_kb`)
100
+
101
+ 1. Write the approved content to `<root>/<newKb>/approved/<filename>` — uploading the cleaned
102
+ text for meetings, or copying the original S3 object directly for documents.
103
+ 2. Write a **Bedrock metadata sidecar** `<filename>.metadata.json`
104
+ (`buildApprovedMetadataJson()`): `metadataAttributes` with `slug`, `client` (the KB name),
105
+ `state=approved`, `approved_at`, `filename`, each `includeForEmbedding: false`.
106
+ 3. **Archive** the raw source: move `upload/<file>` → `archive/<file>` (copy + delete).
107
+ 4. **Sync** the Bedrock knowledge base unless the KB is `general`:
108
+ `syncKnowledgeBaseDataSource()` resolves the KB by name (`KB_PREFIX + kb`, e.g.
109
+ `development-team-<kb>`), finds the matching data source, and `startIngestionJob()`s it with
110
+ a UUID client token. Retries on `ConflictException` / throttling up to `KB_CONFLICT_MAX_RETRIES`.
111
+
112
+ ### Approved Files Browser
113
+
114
+ Modal listing the chosen KB's `approved/` files with size/modified, plus per-row **view**
115
+ (text types open in a viewer via `get_file_content`), **download** (⬇️ — form-POSTs
116
+ `download_file` so the browser saves the file locally), and **delete** (re-syncs the KB).
117
+ Added 2026-06-22: the per-row download button (`downloadApprovedFile()`), reusing the
118
+ existing `download_file` handler.
119
+
120
+ ## Configuration (`kb_processor.ini`)
121
+
122
+ Shared by both scripts via `parse_ini_file(..., true)`. Sections: AWS creds/region, S3 bucket
123
+ + folder names, Bedrock (`KB_PREFIX`, sync delays, conflict-retry tuning), `ALLOWED_EXTENSIONS`,
124
+ AI settings (`AI_API_ENDPOINT`, `AI_MODEL_ID`, `AI_API_KEY`, temperature/tokens/timeout, the
125
+ system prompt), and the `REPLACEMENTS[...]` mishear-correction map.
126
+
127
+ > **⚠️ Security:** `kb_processor.ini` currently stores **live AWS access/secret keys and the AI
128
+ > API key in plaintext, committed to the repo.** These should be rotated and moved out of source
129
+ > (per-developer config or environment) — do not copy these values into other files or docs.
130
+
131
+ ## Gotchas
132
+
133
+ - **Not framework code** — no `_.php` / `App_Framework_Sandbox`. Don't assume `App_*` helpers
134
+ are available here; everything goes through the AWS SDK and the ini.
135
+ - **`general` KB is special-cased** everywhere — approve/delete skip its per-file sync; use the
136
+ batch "Sync All General" flow instead.
137
+ - **Filename convention is load-bearing**: the `YYYY-MM-DD - Name.txt` pattern is the only thing
138
+ that distinguishes a meeting (gets AI cleaning) from a document (copied as-is).
139
+ - Batch sync state lives in `$_SESSION['sync_progress']` and is driven by front-end polling
140
+ (`sync_next_kb`), one KB per request with a `KB_SYNC_DELAY_SECONDS` pause between.
141
+
142
+ ## Change history
143
+
144
+ - 2026-06-22 — Un-retired the folder (`DOA_talos` → `talos`); added a per-file **Download**
145
+ button to the Approved Files Browser; initial knowledge capture of the uploader + processor.
@@ -10,7 +10,7 @@ _Auto-generated by `knowledge.js index`. Do not hand-edit._
10
10
  - **togaview** (TOGa View) — 6 doc(s) → [1.0/apps/togaview/INDEX.md](1.0/apps/togaview/INDEX.md)
11
11
  - **webhook** (Webhook) — 1 doc(s) → [1.0/apps/webhook/INDEX.md](1.0/apps/webhook/INDEX.md)
12
12
  - **walmarttechservices** (Walmart Tech Services) — 1 doc(s) → [1.0/apps/walmarttechservices/INDEX.md](1.0/apps/walmarttechservices/INDEX.md)
13
- - **test** (Test) — 10 doc(s) → [1.0/apps/test/INDEX.md](1.0/apps/test/INDEX.md)
13
+ - **test** (Test) — 11 doc(s) → [1.0/apps/test/INDEX.md](1.0/apps/test/INDEX.md)
14
14
 
15
15
  ## 2.0 framework
16
16
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "toga-ai",
3
- "version": "1.0.155",
3
+ "version": "1.0.156",
4
4
  "description": "TOGA Technology Team Claude Knowledge System — shared AI coding harness with skills, knowledge base CLI, and project installer for Claude Code.",
5
5
  "keywords": [
6
6
  "claude",