ur-agent 1.13.0 → 1.13.2
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.
- package/CHANGELOG.md +47 -1
- package/dist/cli.js +3081 -1672
- package/docs/AGENT_FEATURES.md +23 -0
- package/docs/AGENT_TRENDS.md +7 -1
- package/docs/VALIDATION.md +1 -1
- package/package.json +1 -1
package/docs/AGENT_FEATURES.md
CHANGED
|
@@ -54,3 +54,26 @@ capabilities easier to discover and reuse.
|
|
|
54
54
|
Network-facing behavior, such as a full A2A task server or a GitHub bot that can
|
|
55
55
|
push code, should remain explicitly opt-in because it changes the trust and
|
|
56
56
|
permission boundary.
|
|
57
|
+
|
|
58
|
+
## v1.13 Additions
|
|
59
|
+
|
|
60
|
+
Five additions from a fresh comparison with current Claude Code, Cursor, Codex,
|
|
61
|
+
Cline/Roo, and Copilot workflows.
|
|
62
|
+
|
|
63
|
+
| Addition | Surface | What it adds |
|
|
64
|
+
| --- | --- | --- |
|
|
65
|
+
| AGENTS.md runtime context | automatic | Loads `AGENTS.md` from project roots as project memory (before `UR.md`), for drop-in compatibility with the cross-tool standard |
|
|
66
|
+
| Semantic code index | `ur code-index build\|search\|status` + `CodeSearch` tool | Local embedding-based code retrieval (Ollama embeddings, incremental). The opt-in `CodeSearch` tool (`UR_CODE_INDEX=1`) finds code by meaning alongside Grep/Glob |
|
|
67
|
+
| OS-level execution sandbox | `sandbox.enabled` setting + `/sandbox` | Real enforcement on macOS (Seatbelt) and Linux/WSL (bubblewrap): writes confined to the workspace, optional network block (`UR_SANDBOX_BLOCK_NETWORK`) |
|
|
68
|
+
| Self-review PR gate | `ur agent-task pr --create` | Deterministic diff review that blocks PR creation on merge-conflict markers, hardcoded secrets, and focused tests (override `--force`, skip `--no-review`) |
|
|
69
|
+
| Named role modes | `ur role-mode list\|show\|install` | Architect / Code / Debug / Ask roles with scoped toolsets, installed as `.ur/agents/*.md` so they work with the existing Agent tool |
|
|
70
|
+
|
|
71
|
+
### Design notes
|
|
72
|
+
|
|
73
|
+
- The code index and sandbox are local-first and opt-in. The index uses the
|
|
74
|
+
same local Ollama endpoint UR already uses; the sandbox enforces only when
|
|
75
|
+
the user sets `sandbox.enabled`.
|
|
76
|
+
- Role modes reuse the agent system rather than inventing a parallel runtime
|
|
77
|
+
concept — installing a mode just writes a scoped agent definition.
|
|
78
|
+
- The self-review gate is heuristic and deterministic; it is the automatic
|
|
79
|
+
safety net on the PR path, not a replacement for the model-driven review.
|
package/docs/AGENT_TRENDS.md
CHANGED
|
@@ -19,6 +19,10 @@ ur automation run-due
|
|
|
19
19
|
ur agent-task pr --create --dry-run
|
|
20
20
|
ur a2a serve --dry-run
|
|
21
21
|
ur semantic-memory build
|
|
22
|
+
ur code-index build
|
|
23
|
+
ur code-index search "where is the rate limiter configured"
|
|
24
|
+
ur role-mode install all
|
|
25
|
+
ur agent-task pr --create --dry-run # runs the self-review gate first
|
|
22
26
|
ur claim-ledger validate
|
|
23
27
|
ur browser-qa validate
|
|
24
28
|
```
|
|
@@ -40,10 +44,12 @@ Inside an interactive session:
|
|
|
40
44
|
| Durable workflows and checkpoints | Partial | resume, rewind, background tasks, cron/workflow internals, file restore | Publish a checkpointed workflow format for repeated automations |
|
|
41
45
|
| Multi-agent orchestration | Covered | built-in planning, exploration, verification, and general-purpose agents; custom agents | Document reusable team patterns and role selection |
|
|
42
46
|
| Long-term memory | Partial | `/remember`, `/forget`, `.ur/memory`, research notes, team memory, consolidation | Add optional local semantic/vector memory with retention controls |
|
|
47
|
+
| Semantic codebase retrieval | Covered | local embedding-based code index (`ur code-index`), opt-in `CodeSearch` tool, incremental re-index, Ollama embeddings | Add a file-watcher for auto-reindex on change |
|
|
48
|
+
| AGENTS.md interoperability | Covered | `AGENTS.md` loaded as runtime project context (before `UR.md`), plus imported at `ur init` | Keep aligned as the AGENTS.md spec evolves |
|
|
43
49
|
| Browser and computer-use workflows | Covered | `/browser`, `/chrome`, Playwright-aware tasks, WebSearch, WebFetch, risky-action approval | Add more release fixtures with screenshots and replay assertions |
|
|
44
50
|
| Provenance and citations | Partial | WebFetch source URLs, `/cite`, `/graph`, `/trace`, evidence ledgers | Add claim-to-source mapping for web/MCP answers |
|
|
45
51
|
| Evals and observability | Partial | verifier gates, `.ur/verify.json`, `/verify`, `/trace`, OpenTelemetry hooks, release checks | Publish replayable eval suites and dashboards |
|
|
46
|
-
| Security and prompt-injection resistance | Covered | allow/ask/deny permissions, shell safety analysis, secret scan, untrusted web-content guidance | Continuously test web/MCP injection cases |
|
|
52
|
+
| Security and prompt-injection resistance | Covered | allow/ask/deny permissions, shell safety analysis, secret scan, untrusted web-content guidance, OS-level execution sandbox (macOS Seatbelt, Linux bubblewrap) | Continuously test web/MCP injection cases |
|
|
47
53
|
| Agent identity and delegated authorization | Partial | MCP OAuth/XAA helpers, local trust boundaries, permission rules | Add portable cross-agent identity only with an opt-in A2A task adapter |
|
|
48
54
|
| Multimodal workflows | Partial | `/image`, `/video`, `/youtube`, `/voice`, browser workflows | Add model-aware multimodal capability reporting for local Ollama setups |
|
|
49
55
|
|
package/docs/VALIDATION.md
CHANGED