docs: consolidate repository guides and preserve audit history
This commit is contained in:
cesnimda
@@ -0,0 +1,13 @@
|
||||
# AI
|
||||
|
||||
AI is an optional assistive subsystem. Manual job tracking and document editing remain usable without it.
|
||||
|
||||
- [Architecture and feature boundaries](architecture.md)
|
||||
- [Providers and routing](providers.md)
|
||||
- [Prompt and output rules](prompts.md)
|
||||
- [Safety and privacy](safety.md)
|
||||
- [Document processing](document-processing.md)
|
||||
|
||||
Implemented behavior must be verified in the API and `tools/summarizer/`; future ideas are tracked only in [roadmap](../roadmap/README.md).
|
||||
|
||||
For local service commands and parser limits, see the [summarizer service README](../../tools/summarizer/README.md).
|
||||
@@ -1,55 +0,0 @@
|
||||
# AI Principles
|
||||
|
||||
## 1. User First
|
||||
|
||||
The user owns all generated content.
|
||||
|
||||
---
|
||||
|
||||
## 2. Never Invent Facts
|
||||
|
||||
AI must not create:
|
||||
|
||||
- Jobs.
|
||||
- Companies.
|
||||
- Qualifications.
|
||||
- Skills.
|
||||
- Experience.
|
||||
|
||||
---
|
||||
|
||||
## 3. Explain Changes
|
||||
|
||||
Where possible show:
|
||||
|
||||
Before
|
||||
|
||||
↓
|
||||
|
||||
After
|
||||
|
||||
---
|
||||
|
||||
## 4. Preserve Context
|
||||
|
||||
AI should understand:
|
||||
|
||||
- User career.
|
||||
- Target job.
|
||||
- Industry.
|
||||
|
||||
---
|
||||
|
||||
## 5. Allow Manual Editing
|
||||
|
||||
Every AI output must be editable.
|
||||
|
||||
---
|
||||
|
||||
## 6. Fail Safely
|
||||
|
||||
If AI fails:
|
||||
|
||||
- Show error.
|
||||
- Allow retry.
|
||||
- Do not lose user data.
|
||||
@@ -0,0 +1,7 @@
|
||||
# AI architecture
|
||||
|
||||
The API calls the internal FastAPI service at `AI_SERVICE_BASE_URL`. The service provides local job summarisation, text extraction/OCR, and CV normalization/classification/rewrite routes. The local summarizer uses `sshleifer/distilbart-cnn-12-6`; structured CV tasks use Ollama by default and may use Gemini or Groq only when external processing is enabled and policy permits it.
|
||||
|
||||
Current feature surfaces include job summaries, deterministic match scoring, candidate-fit/focus-plan narratives, CV extraction and normalization, section rewriting, tailored CVs, cover-letter/application drafts, recruiter/follow-up drafts, and interview preparation. AI output is reviewable/editable; sending mail is a separate user action.
|
||||
|
||||
The API owns user authorization, quotas, and feature policy. The sidecar owns provider routing, parser limits, and the shared service-token check. `/health` is the only intentionally open sidecar path.
|
||||
@@ -1,39 +0,0 @@
|
||||
# AI Cost Control
|
||||
|
||||
## Purpose
|
||||
|
||||
Prevent uncontrolled AI expenses.
|
||||
|
||||
---
|
||||
|
||||
# Requirements
|
||||
|
||||
Track:
|
||||
|
||||
- User requests.
|
||||
- Tokens.
|
||||
- Provider usage.
|
||||
|
||||
---
|
||||
|
||||
# Future SaaS
|
||||
|
||||
Support:
|
||||
|
||||
Free:
|
||||
|
||||
Limited usage.
|
||||
|
||||
Premium:
|
||||
|
||||
Higher limits.
|
||||
|
||||
---
|
||||
|
||||
# Optimisation
|
||||
|
||||
Prefer:
|
||||
|
||||
- Smaller models for simple tasks.
|
||||
- Caching where appropriate.
|
||||
- Reusing generated results.
|
||||
@@ -1,49 +0,0 @@
|
||||
# Cover Letter AI
|
||||
|
||||
## Purpose
|
||||
|
||||
Generate personalised cover letters.
|
||||
|
||||
---
|
||||
|
||||
# Inputs
|
||||
|
||||
Use:
|
||||
|
||||
- Career Profile.
|
||||
- Master CV.
|
||||
- Job description.
|
||||
|
||||
---
|
||||
|
||||
# Features
|
||||
|
||||
Generate:
|
||||
|
||||
- First draft.
|
||||
- Improvements.
|
||||
- Tone changes.
|
||||
|
||||
---
|
||||
|
||||
# User Control
|
||||
|
||||
Allow:
|
||||
|
||||
- Editing.
|
||||
- Regeneration.
|
||||
- Manual writing.
|
||||
|
||||
---
|
||||
|
||||
# Rules
|
||||
|
||||
Avoid generic letters.
|
||||
|
||||
Focus on:
|
||||
|
||||
Why this person.
|
||||
|
||||
Why this company.
|
||||
|
||||
Why this role.
|
||||
@@ -1,58 +0,0 @@
|
||||
# CV AI Features
|
||||
|
||||
## Purpose
|
||||
|
||||
Improve CV creation.
|
||||
|
||||
---
|
||||
|
||||
# Features
|
||||
|
||||
## Improve Writing
|
||||
|
||||
Example:
|
||||
|
||||
Original:
|
||||
|
||||
"Worked on websites"
|
||||
|
||||
AI:
|
||||
|
||||
"Developed and maintained customer-facing web applications."
|
||||
|
||||
---
|
||||
|
||||
## Suggest Content
|
||||
|
||||
AI can suggest:
|
||||
|
||||
- Missing skills.
|
||||
- Better wording.
|
||||
|
||||
---
|
||||
|
||||
## Rewrite
|
||||
|
||||
Options:
|
||||
|
||||
- More professional.
|
||||
- More concise.
|
||||
- More technical.
|
||||
|
||||
---
|
||||
|
||||
## Generate
|
||||
|
||||
Create:
|
||||
|
||||
- Summary.
|
||||
- Bullet points.
|
||||
- Sections.
|
||||
|
||||
---
|
||||
|
||||
# Rules
|
||||
|
||||
AI suggestions require approval.
|
||||
|
||||
Never overwrite original content.
|
||||
@@ -0,0 +1,5 @@
|
||||
# Document processing
|
||||
|
||||
Supported extraction includes PDF, DOCX, TXT, Markdown, and PNG/JPEG/WEBP image inputs. Scanned documents can use Tesseract OCR. `/extract-text` performs size/signature checks in the web process and delegates decoding to a bounded child process; temporary work directories are removed and stale work is reconciled.
|
||||
|
||||
Relevant limits are configured with `PARSER_TIMEOUT_SECONDS`, `PARSER_CPU_SECONDS`, `PARSER_MEMORY_MIB`, and `PARSER_STALE_WORK_SECONDS`. `PARSER_TEMP_ROOT` is an optional local override. When a document exceeds safe limits, reject it or ask the user for a smaller/cleaner file; do not disable validation to make a parser pass.
|
||||
@@ -1,49 +0,0 @@
|
||||
# Future AI Features
|
||||
|
||||
Possible future features:
|
||||
|
||||
---
|
||||
|
||||
## AI Career Assistant
|
||||
|
||||
A contextual assistant aware of:
|
||||
|
||||
- Career history.
|
||||
- Goals.
|
||||
- Applications.
|
||||
|
||||
---
|
||||
|
||||
## Interview Coach
|
||||
|
||||
Generate:
|
||||
|
||||
- Questions.
|
||||
- Answers.
|
||||
- Feedback.
|
||||
|
||||
---
|
||||
|
||||
## Career Recommendations
|
||||
|
||||
Suggest:
|
||||
|
||||
- Skills.
|
||||
- Roles.
|
||||
- Learning paths.
|
||||
|
||||
---
|
||||
|
||||
## Automatic Application Assistant
|
||||
|
||||
Potential:
|
||||
|
||||
- Find relevant jobs.
|
||||
- Prepare documents.
|
||||
- Track progress.
|
||||
|
||||
---
|
||||
|
||||
# Priority
|
||||
|
||||
Only build after core workflows are excellent.
|
||||
@@ -1,48 +0,0 @@
|
||||
# Job AI Features
|
||||
|
||||
## Purpose
|
||||
|
||||
Help users understand opportunities.
|
||||
|
||||
---
|
||||
|
||||
# Features
|
||||
|
||||
## Job Analysis
|
||||
|
||||
Extract:
|
||||
|
||||
- Required skills.
|
||||
- Responsibilities.
|
||||
- Seniority.
|
||||
- Keywords.
|
||||
|
||||
---
|
||||
|
||||
## Matching
|
||||
|
||||
Compare:
|
||||
|
||||
Job
|
||||
|
||||
against
|
||||
|
||||
Career Profile
|
||||
|
||||
---
|
||||
|
||||
Output:
|
||||
|
||||
- Strengths.
|
||||
- Gaps.
|
||||
- Suggestions.
|
||||
|
||||
---
|
||||
|
||||
## Application Assistance
|
||||
|
||||
Suggest:
|
||||
|
||||
- CV improvements.
|
||||
- Cover letter topics.
|
||||
- Interview preparation.
|
||||
@@ -1,60 +0,0 @@
|
||||
# AI Model Selection
|
||||
|
||||
## Goal
|
||||
|
||||
Use appropriate models for tasks.
|
||||
|
||||
---
|
||||
|
||||
# Task Categories
|
||||
|
||||
## Simple Tasks
|
||||
|
||||
Examples:
|
||||
|
||||
- Grammar correction.
|
||||
- Shortening text.
|
||||
|
||||
Use:
|
||||
|
||||
Fast/cheap models.
|
||||
|
||||
---
|
||||
|
||||
## Complex Tasks
|
||||
|
||||
Examples:
|
||||
|
||||
- CV generation.
|
||||
- Career analysis.
|
||||
|
||||
Use:
|
||||
|
||||
Higher quality models.
|
||||
|
||||
---
|
||||
|
||||
## Long Context Tasks
|
||||
|
||||
Examples:
|
||||
|
||||
- Analysing full CV.
|
||||
- Comparing job descriptions.
|
||||
|
||||
Use:
|
||||
|
||||
Large context models.
|
||||
|
||||
---
|
||||
|
||||
# Principles
|
||||
|
||||
Do not use expensive models unnecessarily.
|
||||
|
||||
Balance:
|
||||
|
||||
Quality.
|
||||
|
||||
Speed.
|
||||
|
||||
Cost.
|
||||
@@ -1,72 +0,0 @@
|
||||
# AI System Overview
|
||||
|
||||
## Purpose
|
||||
|
||||
AI enhances Jobjakt workflows.
|
||||
|
||||
It helps users:
|
||||
|
||||
- Create better applications.
|
||||
- Improve documents.
|
||||
- Understand opportunities.
|
||||
- Prepare for interviews.
|
||||
|
||||
AI does not replace user decisions.
|
||||
|
||||
---
|
||||
|
||||
# AI Philosophy
|
||||
|
||||
Jobjakt is not an AI chatbot.
|
||||
|
||||
AI should appear where it provides clear value.
|
||||
|
||||
Examples:
|
||||
|
||||
Good:
|
||||
|
||||
"Improve this CV bullet point"
|
||||
|
||||
Good:
|
||||
|
||||
"Tailor this CV for this job"
|
||||
|
||||
Good:
|
||||
|
||||
"Suggest interview questions"
|
||||
|
||||
Bad:
|
||||
|
||||
Empty chat window with no context.
|
||||
|
||||
---
|
||||
|
||||
# AI Sources
|
||||
|
||||
AI can use:
|
||||
|
||||
- Career Profile.
|
||||
- Master CV.
|
||||
- Job descriptions.
|
||||
- User-provided documents.
|
||||
- Application history.
|
||||
|
||||
---
|
||||
|
||||
# User Control
|
||||
|
||||
AI output must:
|
||||
|
||||
- Be visible.
|
||||
- Be editable.
|
||||
- Require approval before saving.
|
||||
|
||||
---
|
||||
|
||||
# Important Rule
|
||||
|
||||
Never silently modify:
|
||||
|
||||
- Career profile.
|
||||
- Master CV.
|
||||
- Application documents.
|
||||
@@ -1,37 +0,0 @@
|
||||
# AI Privacy
|
||||
|
||||
## Principles
|
||||
|
||||
Users control their data.
|
||||
|
||||
---
|
||||
|
||||
# External AI Providers
|
||||
|
||||
Before sending data:
|
||||
|
||||
Clearly communicate:
|
||||
|
||||
- What is sent.
|
||||
- Why it is needed.
|
||||
- Provider used.
|
||||
|
||||
---
|
||||
|
||||
# Future SaaS
|
||||
|
||||
Support:
|
||||
|
||||
- User consent.
|
||||
- Data deletion.
|
||||
- Export.
|
||||
|
||||
---
|
||||
|
||||
# Sensitive Information
|
||||
|
||||
Be careful with:
|
||||
|
||||
- Personal details.
|
||||
- Employment history.
|
||||
- Documents.
|
||||
+3
-49
@@ -1,51 +1,5 @@
|
||||
# AI Prompt Management
|
||||
# Prompt and output rules
|
||||
|
||||
## Purpose
|
||||
Prompts are assembled by feature services/controllers from the job, career profile, reviewed correspondence, and explicitly selected attachments. Inputs are bounded before leaving the API/sidecar. Structured tasks request typed JSON or constrained sections; callers validate and sanitize the result before storing or rendering it.
|
||||
|
||||
Centralise prompts.
|
||||
|
||||
Avoid random prompts scattered throughout code.
|
||||
|
||||
---
|
||||
|
||||
# Requirements
|
||||
|
||||
Prompts should:
|
||||
|
||||
- Be version controlled.
|
||||
- Be documented.
|
||||
- Have tests where possible.
|
||||
|
||||
---
|
||||
|
||||
# Prompt Structure
|
||||
|
||||
Every prompt should define:
|
||||
|
||||
Purpose.
|
||||
|
||||
Inputs.
|
||||
|
||||
Expected output.
|
||||
|
||||
Restrictions.
|
||||
|
||||
---
|
||||
|
||||
# Example
|
||||
|
||||
Task:
|
||||
|
||||
Improve CV bullet point.
|
||||
|
||||
Input:
|
||||
|
||||
Existing bullet.
|
||||
|
||||
Output:
|
||||
|
||||
Improved bullet.
|
||||
|
||||
Restrictions:
|
||||
|
||||
Do not add unsupported achievements.
|
||||
Prompts must preserve factual career data, avoid inventing credentials or experience, respect the selected locale, and keep user edits recoverable. Do not add automatic employer-facing sends to an AI feature without a separate review and delivery design.
|
||||
|
||||
+3
-64
@@ -1,66 +1,5 @@
|
||||
# AI Providers
|
||||
# AI providers and routing
|
||||
|
||||
## Purpose
|
||||
`AI_PROVIDER` selects the structured CV provider: `ollama` (default/local), `gemini`, or `groq`. `AI_ROUTING_MODE` is `local_only`, `local_first`, or `external_only`; external processing also requires the administrator gate, an allowed task, and the user's applicable opt-in/entitlement. Job summarisation remains local in the sidecar.
|
||||
|
||||
Support multiple AI providers.
|
||||
|
||||
---
|
||||
|
||||
# Current/Future Providers
|
||||
|
||||
Possible:
|
||||
|
||||
- OpenAI.
|
||||
- Anthropic.
|
||||
- Google Gemini.
|
||||
- Groq.
|
||||
- Ollama.
|
||||
- Other compatible APIs.
|
||||
|
||||
---
|
||||
|
||||
# Provider Architecture
|
||||
|
||||
Use abstraction:
|
||||
|
||||
AI Provider Interface
|
||||
|
||||
↓
|
||||
|
||||
Provider Implementation
|
||||
|
||||
↓
|
||||
|
||||
Model
|
||||
|
||||
---
|
||||
|
||||
# Requirements
|
||||
|
||||
Users should eventually be able to:
|
||||
|
||||
- Select provider.
|
||||
- Select model.
|
||||
- Configure API keys.
|
||||
|
||||
---
|
||||
|
||||
# Admin Controls
|
||||
|
||||
Admins should control:
|
||||
|
||||
- Available providers.
|
||||
- Allowed models.
|
||||
- Usage limits.
|
||||
|
||||
---
|
||||
|
||||
# SaaS Consideration
|
||||
|
||||
Different plans may have:
|
||||
|
||||
Free models.
|
||||
|
||||
Premium models.
|
||||
|
||||
BYO-key support.
|
||||
Ollama uses `OLLAMA_BASE_URL` and `OLLAMA_MODEL`. Gemini and Groq use their model, base URL, and API-key variables. Keep provider keys in the environment only. If a provider/model is unavailable, show a retriable/unavailable state rather than silently selecting an unevaluated model.
|
||||
|
||||
+3
-6
@@ -1,8 +1,5 @@
|
||||
# safety
|
||||
# AI safety and privacy
|
||||
|
||||
AI output is suggestion-only and cannot overwrite the Career Profile automatically. Prompts delimit
|
||||
untrusted source text, require preservation of factual claims, and instruct providers not to invent
|
||||
employers, titles, dates, or metrics. Attachments are included only when explicitly selected.
|
||||
Local processing is the default. External CV processing is opt-in and gated; provider selection and task type are recorded in the AI interaction/usage paths. The sidecar is private to the backend network in Compose, requires `AI_SERVICE_TOKEN` on protected endpoints, runs with reduced container privileges, and bounds parser work.
|
||||
|
||||
Provider outages degrade AI features without blocking core tracking. Calls and tokens are metered and
|
||||
limited by plan. See `docs/ai/security.md` and `docs/ai/privacy.md`.
|
||||
Known limits: local tests do not prove provider equivalence, production privacy configuration, or real outage behavior. Do not paste personal CVs, tokens, or provider responses into audit notes. See [security notes](../security/security-notes.md) for open verification gates.
|
||||
|
||||
@@ -1,44 +0,0 @@
|
||||
# AI Security
|
||||
|
||||
## Purpose
|
||||
|
||||
Protect user data.
|
||||
|
||||
---
|
||||
|
||||
# Prompt Injection Protection
|
||||
|
||||
Job descriptions and uploaded documents are untrusted input.
|
||||
|
||||
Never allow external text to override system instructions.
|
||||
|
||||
---
|
||||
|
||||
# Data Protection
|
||||
|
||||
Protect:
|
||||
|
||||
- CV data.
|
||||
- Personal information.
|
||||
- Documents.
|
||||
|
||||
---
|
||||
|
||||
# Validation
|
||||
|
||||
Validate:
|
||||
|
||||
- AI outputs.
|
||||
- Generated files.
|
||||
- User inputs.
|
||||
|
||||
---
|
||||
|
||||
# File Processing
|
||||
|
||||
Uploaded files may contain:
|
||||
|
||||
- Malicious content.
|
||||
- Hidden instructions.
|
||||
|
||||
Treat all files as untrusted.
|
||||
@@ -1,44 +0,0 @@
|
||||
# JobTracker AI workload inventory
|
||||
|
||||
Date: 2026-08-02
|
||||
|
||||
This inventory is code-derived, not a production measurement. Typical sizes and latency targets are initial evaluation bands for synthetic benchmarks; they are not observed production SLOs. Current provider means the repository default path, not a verified production setting.
|
||||
|
||||
Privacy classes used by the next policy package:
|
||||
|
||||
- `P0 public`: synthetic/public job text without user data.
|
||||
- `P1 account`: user settings or application metadata without CV/email bodies.
|
||||
- `P2 private`: CV/profile, notes, drafts, contacts or email content.
|
||||
- `P3 credential`: provider tokens/secrets; never a model input.
|
||||
|
||||
| ID | Task and reachable path | Input / typical → max | Output | Language | Target / quality | Privacy and fallback candidate | Mode | Deterministic? | Current provider/model | Pro |
|
||||
|---|---|---|---|---|---|---|---|---|---|---|
|
||||
| DOC-EXTRACT | CV/attachment text extraction; `/extract-text` | PDF/DOCX/image/text; 50 KB–2 MB → API 5 MB, sidecar 8 MB | text + OCR metadata | any OCR-supported | background ≤30 s; high | P2; external prohibited | interactive/background | parser/OCR, not generative | local parser/Tesseract | Yes when used for AI CV import/context |
|
||||
| CV-NORMALIZE | CV structure normalization; `/cv/normalize` | extracted CV text; 2k–15k → 50k chars | strict profile JSON | EN/NO/mixed | background ≤45 s; critical factuality | P2; local default, external only explicit future consent | background | deterministic parser first; AI only unresolved structure | configurable `/cv/*`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| CV-CLASSIFY | ambiguous CV block classification; `/cv/classify-block` | one block; 100–2k → 6k chars | strict classification JSON | EN/NO/mixed | background ≤10 s; medium/high | P2; same as CV-NORMALIZE | background | deterministic headings first | configurable `/cv/*`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| PROFILE-EXTRACT | Career Profile extraction in `BuildStructuredCvAsync` | CV text; 2k–15k → composed prompt 20k | strict `StructuredCvProfile` JSON | EN/NO/mixed | background ≤60 s; critical | P2; external prohibited by default | background | deterministic parsing/fallback remains required | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| PROFILE-DIFF | conservative import/merge suggestions | old/new structured profile | field diff with explicit accept/reject | any | interactive <1 s; critical | P2; no fallback needed | interactive | Yes: `CvProfileDiffService`; keep AI out | local C# | No model call |
|
||||
| JOB-CLEAN | language, stop words, keyword/phrase/skill cleanup | job advert; 200–15k → 20k chars | language/tags/phrases | EN/NO/mixed | <1 s; high precision | P0/P1; no external need | interactive/background | Yes for language/basic tags; semantic phrase extraction may be evaluated separately | local `LanguageDetector`/`SkillTagger` | No model call |
|
||||
| JOB-SUMMARY | create/detail/refresh/enrichment summary | job text; 500–10k → 20k chars | short text + sidecar signals | EN/NO | interactive ≤5 s or queued; medium | P0–P2 when notes included; local default | both | AI for abstractive summary; tags stay deterministic | local `sshleifer/distilbart-cnn-12-6` `/summarize` | Yes for model call |
|
||||
| JOB-MATCH | skills/missing skills/score | job text + reviewed profile | score, matched/missing skills, evidence | EN/NO | <1 s; high/reproducible | P2; no external need | interactive | Yes: `JobCvMatchService`/`ApplicationIntelligenceService` | local C# | No model call |
|
||||
| STRATEGY | candidate fit, focus plan, Strategy Snapshot | job + CV + optional notes/email/attachments; 4k–20k | typed DTO/markdown narrative | EN/NO/mixed | queued target ≤60 s; high | P2; external prohibited by default | interactive → durable background planned | deterministic match evidence first; AI narrative | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| CV-TAILOR | tailored CV/headline/bullets | reviewed profile + job + optional private context; 4k–20k | editable CV draft | EN/NO | queued ≤90 s; critical factuality | P2; external prohibited by default | interactive → background planned | No, but deterministic evidence selection should constrain prompt | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| APPLICATION-DRAFT | cover letter/application answer/recruiter message | CV + job + optional email/notes; 4k–20k | editable text variants | EN/NO | queued ≤90 s; high | P2; external prohibited by default | interactive → background planned | No | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| FOLLOWUP-DRAFT | follow-up email draft | job/application state + saved drafts + optional email; 1k–20k | editable subject/body | EN/NO | interactive/queued ≤30 s; high | P2; external prohibited by default; never auto-send | interactive | No; mode/recipient validation deterministic | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| EMAIL-CLASSIFY | pipeline status from received email | subject/body; 50–10k | status suggestion + signal | EN/NO | <1 s; high precision | P2; no external need | background/interactive | Yes: `EmailStatusClassifier` | local C# | No model call |
|
||||
| RECRUITMENT-DETECT | detect/link recruitment messages | headers/body/job/company signals | suggested link/category/confidence | EN/NO | background <1 s; high precision | P2; no external need for current rules | background | Yes until measured semantic gap | local C# matching/import rules | No model call |
|
||||
| INTERVIEW | generated interview brief/questions | job + profile + saved state | editable brief/questions | EN/NO | queued ≤45 s; high | P2; external prohibited by default | interactive → background planned | checklist/board CRUD deterministic; brief generative | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| WRITING | CV Builder improve/shorten/expand/ATS/bullets/summary/tailor | selected user text + optional role; 50–20k | editable suggestion | EN/NO/mixed | interactive ≤20 s; high factuality | P2; external prohibited by default | interactive | grammar-only may use local deterministic tools later; no current equivalent | `/cv/rewrite`, default Ollama `qwen2.5:7b` | Yes |
|
||||
| HEALTH-PROBE | service capability/latency probe | fixed synthetic sentence | metrics only | EN | ≤10 s; operational | P0; external provider probe requires later explicit policy | background/admin | fixed local request | `/summarize`; local DistilBART | Admin/operational |
|
||||
|
||||
## Reachability and policy observations
|
||||
|
||||
- `/summarize` always uses local DistilBART in the current sidecar. `/cv/*` routes dispatch through `AI_PROVIDER`, whose repository default is local Ollama but which can currently be set globally to Gemini or Groq.
|
||||
- The current global provider switch cannot enforce the task/privacy distinctions above. POL-002/AI-002 must replace it before external fallback is enabled.
|
||||
- `ISummarizerService` truncates composed rewrite prompts at 20,000 characters. Sidecar request models cap rewrite text at 20,000, CV normalization at 50,000, classification blocks at 6,000, and extraction bytes at 8 MB; the application CV upload boundary is 5 MB.
|
||||
- Deterministic keyword cleanup, skill matching, profile diffing and email classification must not be routed through a model.
|
||||
- P3 credentials are configuration inputs only and must never appear in prompts, fixtures, logs or provider payloads.
|
||||
|
||||
## Synthetic evaluation data
|
||||
|
||||
Machine-readable cases are in `JobTrackerApi.Tests/Fixtures/AiEvaluation/cases.json`. They cover all required English/Norwegian/mixed/noisy/sparse/technology/email/follow-up/strategy/strict-JSON/adversarial/injection/long/empty/invalid categories. Expected assertions are constraints rather than golden prose so later model comparisons do not reward one exact wording.
|
||||
Reference in New Issue
Block a user