Reapply "Merge pull request #14 from Smartling/settings-extraction"

This reverts commit 002453a.

Author: PavelLoparev

.env.example

@@ -51,3 +51,33 @@ LLM_CONCURRENCY=5
 # SMARTLING_ACCOUNT_UID=your_account_uid
 # SMARTLING_USER_IDENTIFIER=your_user_identifier
 # SMARTLING_USER_SECRET=your_user_secret
+
+# -----------------------------------------------------------------------------
+# Prompt Templates (optional — defaults shown)
+# -----------------------------------------------------------------------------
+# Directory containing prompt template files
+# PROMPTS_DIR=./prompts
+
+# -----------------------------------------------------------------------------
+# Per-Service LLM Settings (optional — defaults shown)
+# -----------------------------------------------------------------------------
+# Commit Analyzer
+# ANALYZER_TEMPERATURE=0
+# ANALYZER_MAX_TOKENS=256
+
+# Chat Agent (conversational assistant)
+# CHAT_AGENT_TEMPERATURE=0.3
+# CHAT_AGENT_MAX_TOKENS=1500
+# CHAT_AGENT_MAX_ITERATIONS=5
+
+# Developer Summary
+# SUMMARY_TEMPERATURE=0.7
+# SUMMARY_MAX_TOKENS=512
+
+# Report Highlights (report-over-report comparison)
+# HIGHLIGHTS_TEMPERATURE=0.5
+# HIGHLIGHTS_MAX_TOKENS=512
+
+# LLM Connection Test
+# LLM_TEST_TEMPERATURE=0
+# LLM_TEST_MAX_TOKENS=32

CLAUDE.md

@@ -23,6 +23,7 @@ Glooker is a Next.js 15 web app that generates developer impact reports for a Gi
 - **Dual DB support** — SQLite (default, zero config) and MySQL (opt-in via `DB_TYPE=mysql`). The SQLite wrapper translates MySQL-dialect SQL on the fly
 - **In-memory progress store** (globalThis Map) — survives Next.js HMR, acceptable for single-user local use
 - **AI detection** has two layers: trailer parsing (confirmed) and LLM heuristic (maybe_ai)
+- **Prompt template system** — LLM prompts live in `prompts/` dir (configurable via `PROMPTS_DIR`), loaded by `prompt-loader.ts` with in-memory caching. Templates use `{{PLACEHOLDER}}` syntax. All LLM settings (temperature, max_tokens, max_iterations) are configurable via env vars with hardcoded defaults.
 
 ## Environment
 
@@ -45,3 +46,6 @@ Glooker is a Next.js 15 web app that generates developer impact reports for a Gi
 - `@octokit/rest` is ESM-only — any test file that imports from `github.ts` (directly or transitively) must `jest.mock('@octokit/rest')` before the import
 - Tests use Jest + ts-jest with `@/` path alias — config in `jest.config.ts`
 - CI runs on all pull requests and pushes to main (`.github/workflows/test.yml`)
+- `PROMPTS_DIR` defaults to `./prompts` relative to CWD — in Docker, ensure the directory is mounted or `outputFileTracingIncludes` is configured in `next.config.ts`
+- Prompt loader caches template files in memory — restart the server after changing prompt template files (or call `clearPromptCache()` in dev)
+- Prompt template files have Jest snapshot tests that assert exact text — after editing any file in `prompts/`, run `npm test -- -u` to update snapshots (or `npm test -- --testPathPattern="analyzer" -u` for a specific service). Review the snapshot diff to confirm the change is intentional.

README.md

@@ -113,6 +113,34 @@ DB_NAME=glooker
 ```
 Then initialize: `mysql -u root < schema.sql`
 
+### Prompt Customization
+
+LLM prompts are stored as template files in the `prompts/` directory. You can customize prompts by editing these files or pointing to a different directory:
+
+```env
+PROMPTS_DIR=./my-custom-prompts
+```
+
+Template files use `{{PLACEHOLDER}}` syntax for dynamic values injected at runtime.
+
+Each LLM-powered service has configurable temperature and max_tokens settings:
+
+| Setting | Default | Description |
+|---------|---------|-------------|
+| `ANALYZER_TEMPERATURE` | 0 | Commit analysis (deterministic) |
+| `ANALYZER_MAX_TOKENS` | 256 | Commit analysis response limit |
+| `CHAT_AGENT_TEMPERATURE` | 0.3 | Chat assistant |
+| `CHAT_AGENT_MAX_TOKENS` | 1500 | Chat assistant response limit |
+| `CHAT_AGENT_MAX_ITERATIONS` | 5 | Max tool-call rounds per chat |
+| `SUMMARY_TEMPERATURE` | 0.7 | Developer summaries |
+| `SUMMARY_MAX_TOKENS` | 512 | Developer summary response limit |
+| `HIGHLIGHTS_TEMPERATURE` | 0.5 | Report comparison highlights |
+| `HIGHLIGHTS_MAX_TOKENS` | 512 | Highlights response limit |
+
+All settings are optional — defaults match the original hardcoded values.
+
+> **After editing prompt templates:** Run `npm test -- -u` to update Jest snapshots, then review the diff to confirm the change is intentional. Snapshots assert the exact prompt text to prevent accidental regressions.
+
 ## Docker
 
 ```bash