ur-agent 1.33.0 → 1.35.0

package/examples/provider_selection.md

@@ -0,0 +1,45 @@
+# Provider and model selection
+
+Check what is connected, pick a provider, then pick a model scoped to it.
+
+```sh
+ur provider list          # all providers with access type and status
+ur connect status         # connection state for every provider
+ur provider doctor        # detailed checks for the selected provider
+```
+
+Local (default, no account needed — requires a running Ollama app):
+
+```sh
+ur config set provider ollama
+ur --model qwen2.5-coder:latest
+```
+
+API key (stored securely in the OS keychain):
+
+```sh
+ur connect openai-api --key <KEY>   # or: echo "$OPENAI_API_KEY" | ur connect openai-api
+ur config set provider openai-api
+ur config set model gpt-5.5
+```
+
+Local OpenAI-compatible server (LM Studio, llama.cpp, vLLM):
+
+```sh
+ur config set provider lmstudio
+ur config set base_url http://localhost:1234/v1
+```
+
+Subscription CLI (uses the vendor's official CLI and your subscription):
+
+```sh
+ur auth chatgpt           # official Codex CLI login
+ur config set provider codex-cli
+```
+
+Inside a session, `/model` gives the same flow interactively: provider first,
+then only that provider's models. Verify the active pair any time:
+
+```sh
+ur provider status
+```

package/extensions/vscode-ur-inline-diffs/package.json

@@ -2,7 +2,7 @@
   "name": "ur-inline-diffs",
   "displayName": "UR Inline Diffs",
   "description": "Review, apply, and reject UR inline diff bundles from .ur/ide/diffs inside VS Code.",
-  "version": "1.30.6",
+  "version": "1.35.0",
   "publisher": "ur-agent",
   "engines": {
     "vscode": "^1.92.0"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ur-agent",
-  "version": "1.33.0",
+  "version": "1.35.0",
   "description": "UR-AGENT — autonomous engineering workflow engine (plan, execute, test, verify, document, benchmark, reproduce)",
   "type": "module",
   "packageManager": "bun@1.3.14",

package/docs/AGENT_UPGRADE_1.15.0.md

@@ -1,129 +0,0 @@
-# UR-AGENT 1.15.0 Upgrade Notes
-
-UR 1.15.0 keeps the agent local-first and Ollama-centered. The new agent
-platform surfaces do not add Anthropic, OpenAI, or other direct provider API
-dependencies. Network-facing behavior remains opt-in.
-
-## New Local Agent Surfaces
-
-### Background agents
-
-```sh
-ur bg run "fix the flaky parser test" --worktree
-ur bg fanout "try three parser fixes" --agents 3 --worktree
-ur bg list
-ur bg status <id>
-ur bg logs <id> --tail 100
-ur bg attach <id>
-ur bg kill <id>
-```
-
-State is stored under `.ur/background/`. Worktrees are stored under
-`.ur/worktrees/`. Pull requests are created only when `--pr` is passed and then
-use the existing local `gh` CLI path.
-
-### Context and memory retention
-
-```sh
-ur config set compaction.autoThreshold 80
-ur memory retention show
-ur memory retention set --ttl-days 90 --max-entries 5000 --decay-days 30
-ur memory retention prune
-```
-
-`compaction.autoThreshold` is a global percentage threshold from 50 to 95. The
-retention command prunes project-local `.ur/memory/*.jsonl` files.
-
-### Code-index watcher
-
-```sh
-ur config set codeIndex.autoReindex true
-ur code-index watch --graph
-ur code-index watch --dry-run --json
-```
-
-The watcher uses local filesystem events and the same local Ollama embedding
-path as `ur code-index build`.
-
-### Artifact steering
-
-```sh
-ur artifacts add --kind plan --title "Plan" --task <bg_id>
-ur artifacts comment <artifact_id> --feedback "Prefer the simpler parser path"
-```
-
-Comments are written to the artifact and to the linked background task inbox so
-long-running work has durable steering feedback. Background workers now run in
-stream-json mode and inject new inbox comments into the child agent as
-`priority: "now"` user turns while the process is active.
-
-### A2A task server
-
-```sh
-ur a2a card
-ur a2a serve --host 127.0.0.1 --port 8765 --token "$UR_A2A_TOKEN"
-ur a2a token mint --secret "$UR_A2A_DELEGATION_SECRET" --scope coding-agent
-```
-
-The server is opt-in. It refuses off-loopback binds unless a static bearer token
-or delegation secret is configured. Task execution is backed by UR background
-tasks and local `ur -p`; no external model provider API is introduced.
-
-Useful routes:
-
-- `GET /healthz`
-- `GET /.well-known/agent-card.json`
-- `POST /a2a/tasks`
-- `GET /a2a/tasks`
-- `GET /a2a/tasks/:id`
-- `GET /a2a/tasks/:id/output`
-- `POST /a2a/tasks/:id/cancel`
-- `DELETE /a2a/tasks/:id`
-
-### IDE inline diff bundles
-
-```sh
-ur ide diff capture --title "Parser fix"
-ur ide diff list
-ur ide diff show diff-1
-ur ide diff comment diff-1 --feedback "Inline note" --file src/parser.ts --line 42
-ur ide diff schema
-```
-
-Bundles are stored under `.ur/ide/diffs/` as a manifest, per-diff metadata, and
-unified patch files. The repo also ships a native VS Code extension at
-`extensions/vscode-ur-inline-diffs/` that lists bundles, opens patch previews,
-and writes comments back into the UR metadata.
-
-### Benchmark adapters
-
-```sh
-ur eval bench list
-ur eval bench swe-bench --file swe.jsonl --name local-swe
-ur eval bench terminal-bench --file terminal.jsonl --name local-terminal
-ur eval bench aider-polyglot --file aider.jsonl --name local-polyglot
-ur eval run local-swe --dry-run
-```
-
-Adapters import local JSON or JSONL exports into UR eval suites. They do not
-download datasets or call external services.
-
-## Release Verification
-
-Run these before publishing:
-
-```sh
-bun test test/agentFeatureCommands.test.ts test/agentDelegation.test.ts test/codeIndex.test.ts
-bun run typecheck
-npm pack --dry-run
-```
-
-Optional local smoke checks:
-
-```sh
-bun src/entrypoints/cli.tsx bg list
-bun src/entrypoints/cli.tsx memory retention show
-bun src/entrypoints/cli.tsx code-index watch --dry-run
-bun src/entrypoints/cli.tsx eval bench list
-bun src/entrypoints/cli.tsx ide diff schema
-```

package/docs/AGENT_UPGRADE_1.16.0.md

@@ -1,102 +0,0 @@
-# UR-AGENT 1.16.0 Upgrade Notes
-
-UR 1.16.0 adds local-network Ollama discovery. The agent can now find Ollama
-servers on your LAN, let you pick one at startup, and remember the choice for
-future sessions. The endpoint is no longer hardcoded to `localhost:11434`; it
-can be set via settings, environment, or a CLI flag.
-
-## Network Ollama Discovery
-
-### One-time discovery at startup
-
-```sh
-ur --discover-ollama
-```
-
-UR scans the active local subnets (wired Ethernet and Wi-Fi/WLAN) for hosts
-listening on port `11434`, verifies each one by fetching `/api/tags`, then shows
-a picker with:
-
-- `This computer` — your local `ollama serve` at `http://localhost:11434`
-- every discovered LAN host and the number of models it advertises
-
-Select a host and UR uses it for that session only. The choice is **not**
-persisted, so plain `ur` continues to use `localhost:11434`.
-
-### Enable discovery on every startup
-
-Add to `~/.ur/settings.json`:
-
-```json
-{
-  "ollama": {
-    "lanDiscovery": true
-  }
-}
-```
-
-The picker appears on every startup. The choice is still session-only and is
-not written to settings.
-
-### Point to a specific host without scanning
-
-```sh
-ur --ollama-host http://192.168.1.50:11434
-```
-
-This is session-only and does not write settings. It takes precedence over
-settings and `OLLAMA_HOST`.
-
-### Persistent host via settings
-
-```json
-{
-  "ollama": {
-    "host": "http://192.168.1.50:11434"
-  }
-}
-```
-
-When `ollama.host` is set, UR uses it automatically. Precedence is:
-
-1. `--ollama-host <url>` CLI flag
-2. `OLLAMA_HOST` environment variable
-3. `ollama.host` in user settings
-4. fallback `http://localhost:11434`
-
-## What Uses the Chosen Host
-
-The resolved host is used everywhere UR talks to Ollama:
-
-- interactive chat requests (`/api/chat`)
-- installed-model listing (`/api/tags`)
-- model metadata refresh (`/api/show`)
-- local embeddings (`/api/embed`)
-- `ur model-doctor`
-- `ur doctor` / `ur sysinfo`
-- startup preflight connectivity check
-
-## Security Notes
-
-LAN scanning is opt-in only. It never runs automatically unless you pass
-`--discover-ollama` or set `ollama.lanDiscovery: true`. The scan is limited to
-active local IPv4 interfaces and ignores loopback and link-local addresses. It
-uses bounded concurrency and short timeouts so it finishes in a few seconds on a
-/24 subnet.
-
-## Release Verification
-
-Run these before publishing:
-
-```sh
-bun test test/ollamaDiscovery.test.ts test/ollamaModels.test.ts test/ollamaTimeout.test.ts
-bun run typecheck
-npm pack --dry-run
-```
-
-Optional local smoke checks:
-
-```sh
-bun src/entrypoints/cli.tsx --discover-ollama --help
-bun src/entrypoints/cli.tsx --ollama-host http://localhost:11434 -p "hello"
-```

package/docs/AGENT_UPGRADE_1.17.0.md

@@ -1,32 +0,0 @@
-# UR-AGENT 1.17.0 Upgrade Notes
-
-UR 1.17.0 adds the P0 reliable repo editing surface.
-
-## Reliable Repo Editing
-
-```sh
-ur repo-edit index
-ur repo-edit search checkoutTotal
-ur repo-edit plan rename oldName --to newName
-ur repo-edit preview rename oldName --to newName
-ur repo-edit apply rename oldName --to newName --check "bun test"
-```
-
-`repo-edit` is designed for safer repository-wide code changes:
-
-- Builds `.ur/repo-edit/index.json` with file metadata, tokens, and
-  JavaScript/TypeScript symbols.
-- Searches paths, indexed tokens, content lines, and symbols.
-- Plans JavaScript/TypeScript identifier renames from AST nodes instead of
-  replacing arbitrary text.
-- Prints a unified patch preview before writing.
-- Applies all files transactionally and rolls back touched files if syntax
-  validation or an optional check command fails.
-
-## Validation
-
-```sh
-bun test test/repoEdit.test.ts
-bun run typecheck
-node ./bin/ur.js repo-edit --help
-```

package/docs/AGENT_UPGRADE_1.18.0.md

@@ -1,45 +0,0 @@
-# UR-AGENT 1.18.0 Upgrade Notes
-
-UR 1.18.0 adds the P0 test-first execution loop.
-
-## New Command
-
-```sh
-ur test-first detect
-ur test-first --dry-run
-ur test-first --max-attempts 3
-ur test-first install
-```
-
-`test-first` is designed for command-evidence-driven repo work:
-
-- Detects project languages, package managers, and quality commands from the
-  existing project files.
-- Orders detected commands as compile, test, then lint.
-- Runs the command set and reports `passed` only when the commands exit 0.
-- On failure, stores a trace under `.ur/test-first/traces/` with the command,
-  phase, exit code, stdout, and stderr.
-- Invokes a bounded fix agent between failed attempts.
-- Installs detected commands into `.ur/verify.json` with `ur test-first install`
-  so the existing verifier can run them after future edits.
-
-Aliases:
-
-```sh
-ur quality-loop --dry-run
-ur tf-loop detect
-```
-
-## Verification
-
-```sh
-bun run typecheck
-bun run lint
-bun test test/testFirstLoop.test.ts
-node ./bin/ur.js test-first --help
-node ./bin/ur.js test-first detect
-```
-
-If an external project has no lint script, `test-first` reports lint as a
-missing phase instead of inventing a command. UR itself now exposes
-`bun run lint`, so local quality detection includes compile, test, and lint.

package/docs/AGENT_UPGRADE_1.19.0.md

@@ -1,41 +0,0 @@
-# UR-AGENT 1.19.0 Upgrade Notes
-
-UR 1.19.0 adds two P0 agent reliability surfaces: project safety policy and
-project context packing.
-
-## What Changed
-
-- `ur safety status|init|check` evaluates shell command risk before execution.
-- Bash permission checks consult the project safety policy before broad allow
-  rules and sandbox auto-allow.
-- The safety policy separates read, write, execute, and network permission
-  classes.
-- Destructive commands require approval.
-- Write, execute, and network operations receive sandbox guidance.
-- Common secret-file and secret-like environment exfiltration paths are denied.
-- `ur context-pack scan|remember|compress` writes a manifest-backed repository
-  architecture summary and durable task memory.
-
-## New Project Files
-
-- `.ur/safety-policy.json` from `ur safety init`
-- `.ur/project-manifest.json` from `ur context-pack scan`
-- `.ur/context/architecture.md`
-- `.ur/context/task-memory.jsonl`
-- `.ur/context/compressed.md`
-
-Commit only shared policy and architecture files that are safe for teammates.
-Keep local task memory private when it contains local decisions, file paths, or
-operational notes that should not be shared.
-
-## Validate
-
-```sh
-ur safety status
-ur safety check --command "rm -rf build"
-ur context-pack scan
-ur context-pack remember --decision "Use manifest commands first"
-ur context-pack compress
-bun run typecheck
-bun test
-```

package/docs/AGENT_UPGRADE_1.20.0.md

@@ -1,42 +0,0 @@
-# UR-AGENT 1.20.0 Upgrade Notes
-
-UR 1.20.0 adds IDE integration and a richer tool surface for agent workflows.
-
-## What Changed
-
-- **ACP server** — `ur acp serve|stop|status` exposes an HTTP+JSON-RPC Agent
-  Communication Protocol server for VS Code/Cursor/Zed-like editors. It lists
-  tools, calls tools, sends tasks, and reports status. Existing A2A and MCP
-  support remains unchanged.
-- **`ur exec`** — run one or more prompts in non-interactive mode with optional
-  concurrency, worktrees, and output capture.
-- **New built-in tools** exposed through the agent loop, MCP server, and ACP
-  server:
-  - `GitHub` — PR/issue/repo operations via the `gh` CLI.
-  - `Api` — REST HTTP calls with JSON/text output and path extraction.
-  - `Browser` — headless browser automation (fetch/goto/click/type/evaluate/
-    screenshot); interactive actions require `UR_BROWSER_TOOL=1`.
-  - `Docker` — container and compose operations via the `docker` CLI.
-  - `TestRunner` — auto-detect and run project tests.
-  - `Database` — SQL queries against SQLite, Postgres, MySQL, and DuckDB.
-- File-system and terminal tools (`FileRead`, `FileEdit`, `FileWrite`, `Glob`,
-  `Grep`, `Bash`, `PowerShell`) remain built in and are now also reachable via
-  ACP/MCP.
-
-## New Commands
-
-```sh
-ur acp serve --host 127.0.0.1 --port 8123
-ur acp status --json
-ur exec "refactor the parser" --concurrency 2 --json
-ur exec --file prompts.jsonl --output-dir ./outputs
-```
-
-## Validate
-
-```sh
-ur acp status
-ur exec --dry-run "add tests"
-bun run typecheck
-bun test
-```

package/docs/AGENT_UPGRADE_1.21.0.md

@@ -1,58 +0,0 @@
-# Upgrading to UR-AGENT v1.21.0
-
-## What's new
-
-v1.21.0 adds a set of worktree-first agent skills and a command to manage the worktrees they create.
-
-- **Agent skill runner** — `src/services/agents/agentSkillRunner.ts` wraps
-  `startBackgroundTask({ worktree: true, pr: true })`, polls the background run to
-  completion, and returns a PR-style summary with branch, commits, PR URL, and
-  diff summary.
-- **New bundled slash skills** — each skill expands into a prompt that instructs
-  the model to work in an isolated git worktree and produce a clean branch,
-  commits, and PR:
-  - `/debug-v2` — reproduce, root-cause, and fix a bug with a regression test
-  - `/refactor` — safe, test-backed refactoring
-  - `/paper-implementation` — implement an algorithm or system from a paper/URL
-  - `/benchmark` — add or run benchmarks and commit results
-  - `/security-review` — audit code for security issues, fix low-risk items
-  - `/dockerize` — add Dockerfile, compose file, health checks, `.dockerignore`
-  - `/latex-paper` — generate or compile a LaTeX paper/report
-- **Matching agent templates** — `debug-v2`, `refactor`, `paper-implementation`,
-  `benchmark`, `security-review`, `dockerize`, and `latex-paper` are now part of
-  `AGENT_TEMPLATES`. Install them with `ur agent-templates install`.
-- **`ur worktree`** — `ur worktree list|status|clean` inspects and cleans up UR
-  agent worktrees.
-
-## Quick examples
-
-```sh
-# Run a bug-fix agent in an isolated worktree
-ur -p /debug-v2 "parser crashes on empty input"
-
-# Same through the CLI
-ur bg run "reproduce and fix parser crash on empty input" --worktree --pr
-
-# List active worktrees and clean up finished ones
-ur worktree list
-ur worktree clean --dry-run
-ur worktree clean
-
-# Install the new agent templates
-ur agent-templates install debug-v2 refactor benchmark security-review dockerize latex-paper paper-implementation
-```
-
-## For users of `/debug`
-
-The existing `/debug` skill still reads the session debug log. `/debug-v2` is a
-separate skill for bug-fix work in a worktree. You can keep using `/debug` for
-diagnosing the current session and use `/debug-v2` (or `ur bg run ... --worktree
---pr`) when you want a branch + PR.
-
-## Upgrade steps
-
-1. Pull the release and run `bun install`.
-2. Run `bun run typecheck` and `bun test` to confirm the local state.
-3. Install agent templates if you want the new `.ur/agents/*.md` files:
-   `ur agent-templates install`.
-4. No settings or project file migration is required.

package/docs/AGENT_UPGRADE_1.22.0.md

@@ -1,48 +0,0 @@
-# Upgrading to UR-AGENT v1.22.0
-
-## What's new
-
-v1.22.0 turns `ur eval` into a real agent benchmark tool.
-
-- **Execution metrics** — headless `ur -p` child runs now write a metrics JSON file
-  via `UR_EVAL_METRICS_FILE`. The eval runner reads it and also runs
-  `git diff --stat` and an optional per-case `expect.testCommand`. Captured
-  metrics: cost USD, input/output tokens, model, API duration, files changed,
-  insertions/deletions, command failures, human-edit heuristics, and test
-  pass/fail.
-- **Parallel-safe by design** — each child writes its own metrics file, so
-  future parallel eval runs will not corrupt shared cost state.
-- **Richer dashboard** — `ur eval dashboard` and
-  `ur eval report <suite> --dashboard` generate local HTML dashboards with
-  summary cards (pass rate, test pass rate, cost, tokens, files changed, command
-  failures, human edits, duration) and a per-case timeline showing model, time,
-  cost, tokens, diffs, test result, and output preview.
-- **Per-case metrics persistence** — `ur eval run <suite> --metrics` writes
-  each case's metrics to `.ur/evals/.runs/<suite>/<case>.json`.
-- **Aggregate reporting** — `ur eval report <suite>` now prints totals for cost,
-  tokens, files changed, command failures, human edits, duration, and test pass
-  rate.
-
-## Quick examples
-
-```sh
-# Run the starter suite and print JSON with metrics
-ur eval run starter --metrics --json
-
-# Write a single-suite HTML timeline dashboard
-ur eval report starter --dashboard
-
-# Generate the combined dashboard across all saved reports
-ur eval dashboard
-
-# Import a SWE-bench-style export and run it
-ur eval bench swe-bench --file issues.jsonl --name my-bench
-ur eval run my-bench --metrics
-```
-
-## Upgrade steps
-
-1. Pull the release and run `bun install`.
-2. Run `bun run typecheck` and `bun test` to confirm the local state.
-3. Rebuild the bundled CLI: `bun run bundle`.
-4. No settings or project file migration is required.