fe310188ac
2 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
| aee63bf321 |
feat: phase 5 — AI agent integration (OpenAI, Ollama, Anthropic)
Add the AI layer that turns scan results into suggested next actions. Provider model (internal/ai): - Provider interface with Name() and Suggest(ctx, ScanResult) returning []Suggestion (repo_path, action, message, command, priority 0-2) - NewProvider factory resolves the configured provider and enforces that cloud providers have their API key in the configured env var; Ollama needs no key - OpenAIProvider: Chat Completions with response_format json_object - OllamaProvider: local /api/chat with format:json for structured output - AnthropicProvider: Messages API with system prompt and key/version headers; all three fall back to provider-appropriate defaults for base_url and model - Shared client: 60s timeout, 4 MiB response cap, JSON encode/decode, HTTP error surfaces the upstream status and body Prompt design (BuildPrompt): - Renders the full repository status table (name/status/branch/ahead/ behind/changes) plus strict output rules: exact suggestion schema, allowed actions, smallest-safe-step guidance, no invented repositories, return [] when healthy Parsing (parseSuggestions): - Tolerates ```json fences and surrounding prose, clamps priorities to 0-2, caps the result, and rejects replies without a JSON array Suggestion display (presenter.Suggestions): - "AI SUGGESTIONS" table (repository/action/priority/message/command) rendered below the scan table with priority color-coded (red/yellow/ green); empty results say all repositories are healthy Guarded execution (--ai-execute, experimental): - RunConfirmed executes a suggestion's command inside its repository only after explicit per-command y/N confirmation, and only for actions on an allowlist (commit/push/pull/stash/checkout) so LLM output can never run arbitrary shell commands; cancellation aborts remaining suggestions CLI wiring: - scan: AI block after the table when --ai is set and format is not json (JSON streams stay machine-readable); failures degrade to warnings - watch: AI is queried only on the first frame and when something changed since the previous frame, to avoid hammering the provider every interval - New --ai-execute flag and provider validation in config (openai/ollama/ anthropic), included in the config dump Testing: - httptest-based provider tests verifying request shape (model, auth headers, path), response parsing, API error bodies, HTTP failures, and cancellation - Parse tests: plain/fenced/prose replies, empty arrays, garbage, truncated JSON, priority clamping - Execution tests: unsafe actions and empty commands never run, declined confirmations are skipped, confirmed commands execute in the repo dir, failing commands surface errors - Presenter suggestion table and empty-state tests Verified: go build, go vet, go test -race, gofmt clean; end-to-end smoke test against a local fake Ollama server (request shape confirmed, table + suggestions rendered) and the missing-API-key warning path. |
|||
| 0c9c07e8de |
feat: phase 4 — periodic scanning (watch mode) with change detection
Add the scheduling layer and the watch command so gitflow can rescan
repositories on an interval and surface state changes.
Scheduler (internal/scheduler):
- Scheduler runs a callback immediately and then on a time.Ticker until
the context is cancelled; interval <= 0 means a single run
- Graceful shutdown: cancellation is never reported as an error (checked
on the first run and after every run), so Ctrl-C / SIGTERM exit cleanly
- A non-cancellation run error stops the loop and is propagated
Change detection (pkg/status):
- Changed(prev, next) compares snapshots per repository path — status,
branch, detached flag, ahead/behind, stash count, and file counts —
and returns the repositories whose observable state differs, including
repositories that newly appear
Watch command (cmd/gitflow):
- newWatchCmd validates that --interval is positive, prompts for the
parent directory on a TTY when --dir is absent, and drives the
scheduler with ScanOnce
- Each frame clears the screen on a terminal (or prints a RFC3339 header
when output is piped, so watch doubles as a lightweight logger) and
renders through the presenter
- Footer shows changed repositories since the previous frame ("▲ name:
status") and the next scan time; JSON format emits one document per
frame for scripting
- SIGINT/SIGTERM handled via signalContext for a clean stop
Testing:
- Scheduler: single run, periodic repetition, stop-on-error, pre-cancelled
context, and cancellation-during-run (no error reported)
- status.Changed: unchanged repos ignored; status, file-count, and
newly-appeared repos detected
Verified: go build, go vet, go test -race, gofmt clean; manual watch
smoke test over a scratch directory with 1s interval — frames render,
SIGINT and SIGTERM both exit 0 with no orphaned processes.
|
